Training in progress, step 1250, checkpoint
Browse files- checkpoint-1250/README.md +202 -0
- checkpoint-1250/adapter_config.json +34 -0
- checkpoint-1250/adapter_model.safetensors +3 -0
- checkpoint-1250/added_tokens.json +13 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1250/global_step1249/mp_rank_00_model_states.pt +3 -0
- checkpoint-1250/latest +1 -0
- checkpoint-1250/rng_state_0.pth +3 -0
- checkpoint-1250/rng_state_1.pth +3 -0
- checkpoint-1250/rng_state_2.pth +3 -0
- checkpoint-1250/rng_state_3.pth +3 -0
- checkpoint-1250/rng_state_4.pth +3 -0
- checkpoint-1250/rng_state_5.pth +3 -0
- checkpoint-1250/rng_state_6.pth +3 -0
- checkpoint-1250/rng_state_7.pth +3 -0
- checkpoint-1250/scheduler.pt +3 -0
- checkpoint-1250/special_tokens_map.json +30 -0
- checkpoint-1250/tokenizer.json +0 -0
- checkpoint-1250/tokenizer.model +3 -0
- checkpoint-1250/tokenizer_config.json +133 -0
- checkpoint-1250/trainer_state.json +2308 -0
- checkpoint-1250/training_args.bin +3 -0
- checkpoint-1250/zero_to_fp32.py +674 -0
checkpoint-1250/README.md
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: microsoft/Phi-3-mini-4k-instruct
|
| 3 |
+
library_name: peft
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Model Card for Model ID
|
| 7 |
+
|
| 8 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
## Model Details
|
| 13 |
+
|
| 14 |
+
### Model Description
|
| 15 |
+
|
| 16 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
- **Developed by:** [More Information Needed]
|
| 21 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 22 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 23 |
+
- **Model type:** [More Information Needed]
|
| 24 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 25 |
+
- **License:** [More Information Needed]
|
| 26 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 27 |
+
|
| 28 |
+
### Model Sources [optional]
|
| 29 |
+
|
| 30 |
+
<!-- Provide the basic links for the model. -->
|
| 31 |
+
|
| 32 |
+
- **Repository:** [More Information Needed]
|
| 33 |
+
- **Paper [optional]:** [More Information Needed]
|
| 34 |
+
- **Demo [optional]:** [More Information Needed]
|
| 35 |
+
|
| 36 |
+
## Uses
|
| 37 |
+
|
| 38 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 39 |
+
|
| 40 |
+
### Direct Use
|
| 41 |
+
|
| 42 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 43 |
+
|
| 44 |
+
[More Information Needed]
|
| 45 |
+
|
| 46 |
+
### Downstream Use [optional]
|
| 47 |
+
|
| 48 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 49 |
+
|
| 50 |
+
[More Information Needed]
|
| 51 |
+
|
| 52 |
+
### Out-of-Scope Use
|
| 53 |
+
|
| 54 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 55 |
+
|
| 56 |
+
[More Information Needed]
|
| 57 |
+
|
| 58 |
+
## Bias, Risks, and Limitations
|
| 59 |
+
|
| 60 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 61 |
+
|
| 62 |
+
[More Information Needed]
|
| 63 |
+
|
| 64 |
+
### Recommendations
|
| 65 |
+
|
| 66 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 67 |
+
|
| 68 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 69 |
+
|
| 70 |
+
## How to Get Started with the Model
|
| 71 |
+
|
| 72 |
+
Use the code below to get started with the model.
|
| 73 |
+
|
| 74 |
+
[More Information Needed]
|
| 75 |
+
|
| 76 |
+
## Training Details
|
| 77 |
+
|
| 78 |
+
### Training Data
|
| 79 |
+
|
| 80 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 81 |
+
|
| 82 |
+
[More Information Needed]
|
| 83 |
+
|
| 84 |
+
### Training Procedure
|
| 85 |
+
|
| 86 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 87 |
+
|
| 88 |
+
#### Preprocessing [optional]
|
| 89 |
+
|
| 90 |
+
[More Information Needed]
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
#### Training Hyperparameters
|
| 94 |
+
|
| 95 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 96 |
+
|
| 97 |
+
#### Speeds, Sizes, Times [optional]
|
| 98 |
+
|
| 99 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 100 |
+
|
| 101 |
+
[More Information Needed]
|
| 102 |
+
|
| 103 |
+
## Evaluation
|
| 104 |
+
|
| 105 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 106 |
+
|
| 107 |
+
### Testing Data, Factors & Metrics
|
| 108 |
+
|
| 109 |
+
#### Testing Data
|
| 110 |
+
|
| 111 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 112 |
+
|
| 113 |
+
[More Information Needed]
|
| 114 |
+
|
| 115 |
+
#### Factors
|
| 116 |
+
|
| 117 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 118 |
+
|
| 119 |
+
[More Information Needed]
|
| 120 |
+
|
| 121 |
+
#### Metrics
|
| 122 |
+
|
| 123 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 124 |
+
|
| 125 |
+
[More Information Needed]
|
| 126 |
+
|
| 127 |
+
### Results
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
#### Summary
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
|
| 135 |
+
## Model Examination [optional]
|
| 136 |
+
|
| 137 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 138 |
+
|
| 139 |
+
[More Information Needed]
|
| 140 |
+
|
| 141 |
+
## Environmental Impact
|
| 142 |
+
|
| 143 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 144 |
+
|
| 145 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 146 |
+
|
| 147 |
+
- **Hardware Type:** [More Information Needed]
|
| 148 |
+
- **Hours used:** [More Information Needed]
|
| 149 |
+
- **Cloud Provider:** [More Information Needed]
|
| 150 |
+
- **Compute Region:** [More Information Needed]
|
| 151 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 152 |
+
|
| 153 |
+
## Technical Specifications [optional]
|
| 154 |
+
|
| 155 |
+
### Model Architecture and Objective
|
| 156 |
+
|
| 157 |
+
[More Information Needed]
|
| 158 |
+
|
| 159 |
+
### Compute Infrastructure
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
#### Hardware
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Software
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
## Citation [optional]
|
| 172 |
+
|
| 173 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 174 |
+
|
| 175 |
+
**BibTeX:**
|
| 176 |
+
|
| 177 |
+
[More Information Needed]
|
| 178 |
+
|
| 179 |
+
**APA:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
## Glossary [optional]
|
| 184 |
+
|
| 185 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 186 |
+
|
| 187 |
+
[More Information Needed]
|
| 188 |
+
|
| 189 |
+
## More Information [optional]
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## Model Card Authors [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Contact
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
### Framework versions
|
| 201 |
+
|
| 202 |
+
- PEFT 0.14.0
|
checkpoint-1250/adapter_config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "microsoft/Phi-3-mini-4k-instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"eva_config": null,
|
| 7 |
+
"exclude_modules": null,
|
| 8 |
+
"fan_in_fan_out": false,
|
| 9 |
+
"inference_mode": true,
|
| 10 |
+
"init_lora_weights": true,
|
| 11 |
+
"layer_replication": null,
|
| 12 |
+
"layers_pattern": null,
|
| 13 |
+
"layers_to_transform": null,
|
| 14 |
+
"loftq_config": {},
|
| 15 |
+
"lora_alpha": 16,
|
| 16 |
+
"lora_bias": false,
|
| 17 |
+
"lora_dropout": 0.0,
|
| 18 |
+
"megatron_config": null,
|
| 19 |
+
"megatron_core": "megatron.core",
|
| 20 |
+
"modules_to_save": null,
|
| 21 |
+
"peft_type": "LORA",
|
| 22 |
+
"r": 8,
|
| 23 |
+
"rank_pattern": {},
|
| 24 |
+
"revision": null,
|
| 25 |
+
"target_modules": [
|
| 26 |
+
"qkv_proj",
|
| 27 |
+
"gate_up_proj",
|
| 28 |
+
"o_proj",
|
| 29 |
+
"down_proj"
|
| 30 |
+
],
|
| 31 |
+
"task_type": "CAUSAL_LM",
|
| 32 |
+
"use_dora": false,
|
| 33 |
+
"use_rslora": false
|
| 34 |
+
}
|
checkpoint-1250/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b7db9ab3f6be7df165471d843086ffe9ebe8fc537e0d87ccd5681782028cf9b
|
| 3 |
+
size 25200088
|
checkpoint-1250/added_tokens.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<|assistant|>": 32001,
|
| 3 |
+
"<|endoftext|>": 32000,
|
| 4 |
+
"<|end|>": 32007,
|
| 5 |
+
"<|placeholder1|>": 32002,
|
| 6 |
+
"<|placeholder2|>": 32003,
|
| 7 |
+
"<|placeholder3|>": 32004,
|
| 8 |
+
"<|placeholder4|>": 32005,
|
| 9 |
+
"<|placeholder5|>": 32008,
|
| 10 |
+
"<|placeholder6|>": 32009,
|
| 11 |
+
"<|system|>": 32006,
|
| 12 |
+
"<|user|>": 32010
|
| 13 |
+
}
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a480755cca2f2ebf3a3cb34b235a7902bfd8ba64aa692ef6d64f283d28eedfd6
|
| 3 |
+
size 18881328
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:caab6b1fa8ce39f604ea919d7b6aea4abd6b02e3dfd8afea787a7881c64351e6
|
| 3 |
+
size 18881328
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ecd51203a602dee84974f83ad1b0120e15ca3adad7d89e25201ac08964a3544e
|
| 3 |
+
size 18881328
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a735da857a7333bf248143fd9fe91e6b2629eb9e1a85a0b316c1a8473841e9a
|
| 3 |
+
size 18881392
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b74ff2a77faf1c5c02a20b54bfb3a3e04297f64bf2672162e2c76627722ab97
|
| 3 |
+
size 18881392
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:002ba7ca027f00de9dde3f80baff8297cdafa3a964a1e1fefd80209e7c6090ee
|
| 3 |
+
size 18881392
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e338914a2297c1c65c73538211da339a48ce494c399cc645f6d4515abbfc7893
|
| 3 |
+
size 18881392
|
checkpoint-1250/global_step1249/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:196dbefa616bd48dd48da37ea8572723b492d08f93cbafeb0b0d16f988dd3ceb
|
| 3 |
+
size 18881392
|
checkpoint-1250/global_step1249/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a5cf1ebaf68c4624423463a91922242cb7b4cf6d64b2ee13c2c666518c700059
|
| 3 |
+
size 25379244
|
checkpoint-1250/latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step1249
|
checkpoint-1250/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c49abc3bdedbec1fc8e1028ef422150f19ee7470d7b542e1ad8869fc044d2af
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df12ca4106ff0831785a55b5da88f6c86f6f67bd3d09b2dced4f20b539b14f72
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05fc0786faff729a3a1582f98b806b68d4f0b76aebb25cbad4431b73176b11c1
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a3117e1218a2dd3f7f8c516a840af48f6b93660d852cca124269f78c21f8577c
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cdaa30c82476bf6a65e4eb9ca2ae7b95f1b38f41a6f5b2f1cbdda9af86a4a7a0
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:09cdde6931807139efa184e8a98108b74bb05730bc511336966b254b68dc93ee
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a2268136932c55b3857d38c7cf3fc4bd3cdad532c156b9addebc6d26374374a
|
| 3 |
+
size 15984
|
checkpoint-1250/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:56cd9a502015b79e0ab94c92a04bd96c99aaf79ef8d64bf81d81eb702c10c2a8
|
| 3 |
+
size 15984
|
checkpoint-1250/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85a32e7d79d4bce24e8039d66138f2f6eb81d15d973d067af5d2c4296fc1dd0b
|
| 3 |
+
size 1064
|
checkpoint-1250/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|end|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "<|endoftext|>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
checkpoint-1250/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1250/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
checkpoint-1250/tokenizer_config.json
ADDED
|
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": true,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": false
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "<|endoftext|>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
},
|
| 38 |
+
"32001": {
|
| 39 |
+
"content": "<|assistant|>",
|
| 40 |
+
"lstrip": false,
|
| 41 |
+
"normalized": false,
|
| 42 |
+
"rstrip": true,
|
| 43 |
+
"single_word": false,
|
| 44 |
+
"special": true
|
| 45 |
+
},
|
| 46 |
+
"32002": {
|
| 47 |
+
"content": "<|placeholder1|>",
|
| 48 |
+
"lstrip": false,
|
| 49 |
+
"normalized": false,
|
| 50 |
+
"rstrip": true,
|
| 51 |
+
"single_word": false,
|
| 52 |
+
"special": true
|
| 53 |
+
},
|
| 54 |
+
"32003": {
|
| 55 |
+
"content": "<|placeholder2|>",
|
| 56 |
+
"lstrip": false,
|
| 57 |
+
"normalized": false,
|
| 58 |
+
"rstrip": true,
|
| 59 |
+
"single_word": false,
|
| 60 |
+
"special": true
|
| 61 |
+
},
|
| 62 |
+
"32004": {
|
| 63 |
+
"content": "<|placeholder3|>",
|
| 64 |
+
"lstrip": false,
|
| 65 |
+
"normalized": false,
|
| 66 |
+
"rstrip": true,
|
| 67 |
+
"single_word": false,
|
| 68 |
+
"special": true
|
| 69 |
+
},
|
| 70 |
+
"32005": {
|
| 71 |
+
"content": "<|placeholder4|>",
|
| 72 |
+
"lstrip": false,
|
| 73 |
+
"normalized": false,
|
| 74 |
+
"rstrip": true,
|
| 75 |
+
"single_word": false,
|
| 76 |
+
"special": true
|
| 77 |
+
},
|
| 78 |
+
"32006": {
|
| 79 |
+
"content": "<|system|>",
|
| 80 |
+
"lstrip": false,
|
| 81 |
+
"normalized": false,
|
| 82 |
+
"rstrip": true,
|
| 83 |
+
"single_word": false,
|
| 84 |
+
"special": true
|
| 85 |
+
},
|
| 86 |
+
"32007": {
|
| 87 |
+
"content": "<|end|>",
|
| 88 |
+
"lstrip": false,
|
| 89 |
+
"normalized": false,
|
| 90 |
+
"rstrip": false,
|
| 91 |
+
"single_word": false,
|
| 92 |
+
"special": true
|
| 93 |
+
},
|
| 94 |
+
"32008": {
|
| 95 |
+
"content": "<|placeholder5|>",
|
| 96 |
+
"lstrip": false,
|
| 97 |
+
"normalized": false,
|
| 98 |
+
"rstrip": true,
|
| 99 |
+
"single_word": false,
|
| 100 |
+
"special": true
|
| 101 |
+
},
|
| 102 |
+
"32009": {
|
| 103 |
+
"content": "<|placeholder6|>",
|
| 104 |
+
"lstrip": false,
|
| 105 |
+
"normalized": false,
|
| 106 |
+
"rstrip": true,
|
| 107 |
+
"single_word": false,
|
| 108 |
+
"special": true
|
| 109 |
+
},
|
| 110 |
+
"32010": {
|
| 111 |
+
"content": "<|user|>",
|
| 112 |
+
"lstrip": false,
|
| 113 |
+
"normalized": false,
|
| 114 |
+
"rstrip": true,
|
| 115 |
+
"single_word": false,
|
| 116 |
+
"special": true
|
| 117 |
+
}
|
| 118 |
+
},
|
| 119 |
+
"bos_token": "<s>",
|
| 120 |
+
"chat_template": "{% set system_message = 'You are a helpful AI assistant.' %}{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '<s>' + '<|system|>\n' + system_message + '<|end|>\n' }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|user|>\n' + content + '<|end|>\n<|assistant|>\n' }}{% elif message['role'] == 'assistant' %}{{ content + '<|end|>' + '\n' }}{% endif %}{% endfor %}",
|
| 121 |
+
"clean_up_tokenization_spaces": false,
|
| 122 |
+
"eos_token": "<|end|>",
|
| 123 |
+
"extra_special_tokens": {},
|
| 124 |
+
"legacy": false,
|
| 125 |
+
"model_max_length": 4096,
|
| 126 |
+
"pad_token": "<|endoftext|>",
|
| 127 |
+
"padding_side": "right",
|
| 128 |
+
"sp_model_kwargs": {},
|
| 129 |
+
"split_special_tokens": false,
|
| 130 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 131 |
+
"unk_token": "<unk>",
|
| 132 |
+
"use_default_system_prompt": false
|
| 133 |
+
}
|
checkpoint-1250/trainer_state.json
ADDED
|
@@ -0,0 +1,2308 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.1479269553233031,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 1250,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00918800964741013,
|
| 13 |
+
"grad_norm": 0.036612071096897125,
|
| 14 |
+
"learning_rate": 4.999451708687114e-06,
|
| 15 |
+
"logits/chosen": 15.01579761505127,
|
| 16 |
+
"logits/rejected": 15.359031677246094,
|
| 17 |
+
"logps/chosen": -0.2681262791156769,
|
| 18 |
+
"logps/rejected": -0.31947994232177734,
|
| 19 |
+
"loss": 0.9551,
|
| 20 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 21 |
+
"rewards/chosen": -0.40218934416770935,
|
| 22 |
+
"rewards/margins": 0.07703053951263428,
|
| 23 |
+
"rewards/rejected": -0.479219913482666,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01837601929482026,
|
| 28 |
+
"grad_norm": 0.05575725808739662,
|
| 29 |
+
"learning_rate": 4.997807075247147e-06,
|
| 30 |
+
"logits/chosen": 14.570712089538574,
|
| 31 |
+
"logits/rejected": 15.321355819702148,
|
| 32 |
+
"logps/chosen": -0.2867889404296875,
|
| 33 |
+
"logps/rejected": -0.3514837622642517,
|
| 34 |
+
"loss": 0.923,
|
| 35 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 36 |
+
"rewards/chosen": -0.43018341064453125,
|
| 37 |
+
"rewards/margins": 0.09704220294952393,
|
| 38 |
+
"rewards/rejected": -0.5272256135940552,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02756402894223039,
|
| 43 |
+
"grad_norm": 0.0492466576397419,
|
| 44 |
+
"learning_rate": 4.9950668210706795e-06,
|
| 45 |
+
"logits/chosen": 14.748420715332031,
|
| 46 |
+
"logits/rejected": 14.969354629516602,
|
| 47 |
+
"logps/chosen": -0.28405922651290894,
|
| 48 |
+
"logps/rejected": -0.32855403423309326,
|
| 49 |
+
"loss": 0.9357,
|
| 50 |
+
"rewards/accuracies": 0.5,
|
| 51 |
+
"rewards/chosen": -0.426088809967041,
|
| 52 |
+
"rewards/margins": 0.06674225628376007,
|
| 53 |
+
"rewards/rejected": -0.4928310811519623,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.03675203858964052,
|
| 58 |
+
"grad_norm": 0.05719422921538353,
|
| 59 |
+
"learning_rate": 4.9912321481237616e-06,
|
| 60 |
+
"logits/chosen": 14.28278923034668,
|
| 61 |
+
"logits/rejected": 14.76964282989502,
|
| 62 |
+
"logps/chosen": -0.27940627932548523,
|
| 63 |
+
"logps/rejected": -0.3408831059932709,
|
| 64 |
+
"loss": 0.9215,
|
| 65 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 66 |
+
"rewards/chosen": -0.41910940408706665,
|
| 67 |
+
"rewards/margins": 0.09221524000167847,
|
| 68 |
+
"rewards/rejected": -0.5113246440887451,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04594004823705065,
|
| 73 |
+
"grad_norm": 0.06247895210981369,
|
| 74 |
+
"learning_rate": 4.986304738420684e-06,
|
| 75 |
+
"logits/chosen": 14.943578720092773,
|
| 76 |
+
"logits/rejected": 14.936178207397461,
|
| 77 |
+
"logps/chosen": -0.2819541394710541,
|
| 78 |
+
"logps/rejected": -0.3245392441749573,
|
| 79 |
+
"loss": 0.9464,
|
| 80 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 81 |
+
"rewards/chosen": -0.4229312539100647,
|
| 82 |
+
"rewards/margins": 0.06387762725353241,
|
| 83 |
+
"rewards/rejected": -0.4868088662624359,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.04594004823705065,
|
| 88 |
+
"eval_logits/chosen": 14.7594575881958,
|
| 89 |
+
"eval_logits/rejected": 15.193694114685059,
|
| 90 |
+
"eval_logps/chosen": -0.2807807922363281,
|
| 91 |
+
"eval_logps/rejected": -0.36209535598754883,
|
| 92 |
+
"eval_loss": 0.9397181868553162,
|
| 93 |
+
"eval_rewards/accuracies": 0.5681818127632141,
|
| 94 |
+
"eval_rewards/chosen": -0.4211711883544922,
|
| 95 |
+
"eval_rewards/margins": 0.12197184562683105,
|
| 96 |
+
"eval_rewards/rejected": -0.5431429743766785,
|
| 97 |
+
"eval_runtime": 24.9762,
|
| 98 |
+
"eval_samples_per_second": 28.187,
|
| 99 |
+
"eval_steps_per_second": 3.523,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.05512805788446078,
|
| 104 |
+
"grad_norm": 0.11519577354192734,
|
| 105 |
+
"learning_rate": 4.980286753286196e-06,
|
| 106 |
+
"logits/chosen": 14.996228218078613,
|
| 107 |
+
"logits/rejected": 15.37781810760498,
|
| 108 |
+
"logps/chosen": -0.2809831202030182,
|
| 109 |
+
"logps/rejected": -0.35486167669296265,
|
| 110 |
+
"loss": 0.9318,
|
| 111 |
+
"rewards/accuracies": 0.5625,
|
| 112 |
+
"rewards/chosen": -0.4214746952056885,
|
| 113 |
+
"rewards/margins": 0.1108178049325943,
|
| 114 |
+
"rewards/rejected": -0.5322924852371216,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.06431606753187091,
|
| 119 |
+
"grad_norm": 0.06691388040781021,
|
| 120 |
+
"learning_rate": 4.973180832407471e-06,
|
| 121 |
+
"logits/chosen": 14.612454414367676,
|
| 122 |
+
"logits/rejected": 15.678136825561523,
|
| 123 |
+
"logps/chosen": -0.2569667100906372,
|
| 124 |
+
"logps/rejected": -0.40047627687454224,
|
| 125 |
+
"loss": 0.9158,
|
| 126 |
+
"rewards/accuracies": 0.75,
|
| 127 |
+
"rewards/chosen": -0.3854501247406006,
|
| 128 |
+
"rewards/margins": 0.21526429057121277,
|
| 129 |
+
"rewards/rejected": -0.600714385509491,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.07350407717928104,
|
| 134 |
+
"grad_norm": 0.05976058170199394,
|
| 135 |
+
"learning_rate": 4.964990092676263e-06,
|
| 136 |
+
"logits/chosen": 14.873895645141602,
|
| 137 |
+
"logits/rejected": 15.50474739074707,
|
| 138 |
+
"logps/chosen": -0.28742527961730957,
|
| 139 |
+
"logps/rejected": -0.37555089592933655,
|
| 140 |
+
"loss": 0.9372,
|
| 141 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 142 |
+
"rewards/chosen": -0.43113788962364197,
|
| 143 |
+
"rewards/margins": 0.13218846917152405,
|
| 144 |
+
"rewards/rejected": -0.5633264183998108,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.08269208682669117,
|
| 149 |
+
"grad_norm": 0.0602131113409996,
|
| 150 |
+
"learning_rate": 4.9557181268217225e-06,
|
| 151 |
+
"logits/chosen": 14.356691360473633,
|
| 152 |
+
"logits/rejected": 14.895658493041992,
|
| 153 |
+
"logps/chosen": -0.2613506317138672,
|
| 154 |
+
"logps/rejected": -0.3317110538482666,
|
| 155 |
+
"loss": 0.9324,
|
| 156 |
+
"rewards/accuracies": 0.5625,
|
| 157 |
+
"rewards/chosen": -0.3920259475708008,
|
| 158 |
+
"rewards/margins": 0.10554064810276031,
|
| 159 |
+
"rewards/rejected": -0.4975665509700775,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.0918800964741013,
|
| 164 |
+
"grad_norm": 0.07126503437757492,
|
| 165 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 166 |
+
"logits/chosen": 14.862826347351074,
|
| 167 |
+
"logits/rejected": 15.257089614868164,
|
| 168 |
+
"logps/chosen": -0.2707213759422302,
|
| 169 |
+
"logps/rejected": -0.3511395752429962,
|
| 170 |
+
"loss": 0.9353,
|
| 171 |
+
"rewards/accuracies": 0.637499988079071,
|
| 172 |
+
"rewards/chosen": -0.4060820937156677,
|
| 173 |
+
"rewards/margins": 0.1206272691488266,
|
| 174 |
+
"rewards/rejected": -0.5267094373703003,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.0918800964741013,
|
| 179 |
+
"eval_logits/chosen": 14.664334297180176,
|
| 180 |
+
"eval_logits/rejected": 15.113536834716797,
|
| 181 |
+
"eval_logps/chosen": -0.2750833034515381,
|
| 182 |
+
"eval_logps/rejected": -0.36540210247039795,
|
| 183 |
+
"eval_loss": 0.9324077367782593,
|
| 184 |
+
"eval_rewards/accuracies": 0.5795454382896423,
|
| 185 |
+
"eval_rewards/chosen": -0.41262495517730713,
|
| 186 |
+
"eval_rewards/margins": 0.1354781985282898,
|
| 187 |
+
"eval_rewards/rejected": -0.5481031537055969,
|
| 188 |
+
"eval_runtime": 24.4286,
|
| 189 |
+
"eval_samples_per_second": 28.819,
|
| 190 |
+
"eval_steps_per_second": 3.602,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.10106810612151143,
|
| 195 |
+
"grad_norm": 0.07136944681406021,
|
| 196 |
+
"learning_rate": 4.933947257182901e-06,
|
| 197 |
+
"logits/chosen": 14.942098617553711,
|
| 198 |
+
"logits/rejected": 15.138586044311523,
|
| 199 |
+
"logps/chosen": -0.2860812246799469,
|
| 200 |
+
"logps/rejected": -0.36259371042251587,
|
| 201 |
+
"loss": 0.934,
|
| 202 |
+
"rewards/accuracies": 0.574999988079071,
|
| 203 |
+
"rewards/chosen": -0.42912182211875916,
|
| 204 |
+
"rewards/margins": 0.11476878076791763,
|
| 205 |
+
"rewards/rejected": -0.5438905954360962,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.11025611576892155,
|
| 210 |
+
"grad_norm": 0.07038908451795578,
|
| 211 |
+
"learning_rate": 4.921457902821578e-06,
|
| 212 |
+
"logits/chosen": 14.488851547241211,
|
| 213 |
+
"logits/rejected": 14.702054023742676,
|
| 214 |
+
"logps/chosen": -0.2662215232849121,
|
| 215 |
+
"logps/rejected": -0.3013685941696167,
|
| 216 |
+
"loss": 0.9202,
|
| 217 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 218 |
+
"rewards/chosen": -0.39933228492736816,
|
| 219 |
+
"rewards/margins": 0.05272058770060539,
|
| 220 |
+
"rewards/rejected": -0.45205289125442505,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11944412541633169,
|
| 225 |
+
"grad_norm": 0.06875801086425781,
|
| 226 |
+
"learning_rate": 4.907906416994146e-06,
|
| 227 |
+
"logits/chosen": 14.075657844543457,
|
| 228 |
+
"logits/rejected": 14.696513175964355,
|
| 229 |
+
"logps/chosen": -0.250360369682312,
|
| 230 |
+
"logps/rejected": -0.3504650592803955,
|
| 231 |
+
"loss": 0.9266,
|
| 232 |
+
"rewards/accuracies": 0.625,
|
| 233 |
+
"rewards/chosen": -0.375540554523468,
|
| 234 |
+
"rewards/margins": 0.15015706419944763,
|
| 235 |
+
"rewards/rejected": -0.5256975889205933,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.12863213506374183,
|
| 240 |
+
"grad_norm": 0.0984601378440857,
|
| 241 |
+
"learning_rate": 4.893298743830168e-06,
|
| 242 |
+
"logits/chosen": 13.738212585449219,
|
| 243 |
+
"logits/rejected": 14.311574935913086,
|
| 244 |
+
"logps/chosen": -0.26711025834083557,
|
| 245 |
+
"logps/rejected": -0.3587702810764313,
|
| 246 |
+
"loss": 0.9185,
|
| 247 |
+
"rewards/accuracies": 0.512499988079071,
|
| 248 |
+
"rewards/chosen": -0.40066537261009216,
|
| 249 |
+
"rewards/margins": 0.13749003410339355,
|
| 250 |
+
"rewards/rejected": -0.5381554365158081,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.13782014471115195,
|
| 255 |
+
"grad_norm": 0.10201425850391388,
|
| 256 |
+
"learning_rate": 4.8776412907378845e-06,
|
| 257 |
+
"logits/chosen": 13.7462797164917,
|
| 258 |
+
"logits/rejected": 14.230626106262207,
|
| 259 |
+
"logps/chosen": -0.25559619069099426,
|
| 260 |
+
"logps/rejected": -0.3708702623844147,
|
| 261 |
+
"loss": 0.9106,
|
| 262 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 263 |
+
"rewards/chosen": -0.3833943009376526,
|
| 264 |
+
"rewards/margins": 0.17291104793548584,
|
| 265 |
+
"rewards/rejected": -0.5563054084777832,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.13782014471115195,
|
| 270 |
+
"eval_logits/chosen": 13.458538055419922,
|
| 271 |
+
"eval_logits/rejected": 13.998083114624023,
|
| 272 |
+
"eval_logps/chosen": -0.2759075462818146,
|
| 273 |
+
"eval_logps/rejected": -0.3873325288295746,
|
| 274 |
+
"eval_loss": 0.9164085388183594,
|
| 275 |
+
"eval_rewards/accuracies": 0.5795454382896423,
|
| 276 |
+
"eval_rewards/chosen": -0.41386130452156067,
|
| 277 |
+
"eval_rewards/margins": 0.1671374887228012,
|
| 278 |
+
"eval_rewards/rejected": -0.5809988379478455,
|
| 279 |
+
"eval_runtime": 24.4393,
|
| 280 |
+
"eval_samples_per_second": 28.806,
|
| 281 |
+
"eval_steps_per_second": 3.601,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.14700815435856207,
|
| 286 |
+
"grad_norm": 0.11537656933069229,
|
| 287 |
+
"learning_rate": 4.860940925593703e-06,
|
| 288 |
+
"logits/chosen": 12.686149597167969,
|
| 289 |
+
"logits/rejected": 13.478736877441406,
|
| 290 |
+
"logps/chosen": -0.23941929638385773,
|
| 291 |
+
"logps/rejected": -0.3713286519050598,
|
| 292 |
+
"loss": 0.9094,
|
| 293 |
+
"rewards/accuracies": 0.625,
|
| 294 |
+
"rewards/chosen": -0.3591288924217224,
|
| 295 |
+
"rewards/margins": 0.1978640854358673,
|
| 296 |
+
"rewards/rejected": -0.5569929480552673,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.1561961640059722,
|
| 301 |
+
"grad_norm": 0.1196313351392746,
|
| 302 |
+
"learning_rate": 4.84320497372973e-06,
|
| 303 |
+
"logits/chosen": 13.221656799316406,
|
| 304 |
+
"logits/rejected": 13.317082405090332,
|
| 305 |
+
"logps/chosen": -0.3033878207206726,
|
| 306 |
+
"logps/rejected": -0.3784424960613251,
|
| 307 |
+
"loss": 0.9057,
|
| 308 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 309 |
+
"rewards/chosen": -0.4550817608833313,
|
| 310 |
+
"rewards/margins": 0.11258199065923691,
|
| 311 |
+
"rewards/rejected": -0.5676637887954712,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.16538417365338234,
|
| 316 |
+
"grad_norm": 0.18745549023151398,
|
| 317 |
+
"learning_rate": 4.824441214720629e-06,
|
| 318 |
+
"logits/chosen": 11.797627449035645,
|
| 319 |
+
"logits/rejected": 12.031414985656738,
|
| 320 |
+
"logps/chosen": -0.2746419608592987,
|
| 321 |
+
"logps/rejected": -0.3629845976829529,
|
| 322 |
+
"loss": 0.8954,
|
| 323 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 324 |
+
"rewards/chosen": -0.41196292638778687,
|
| 325 |
+
"rewards/margins": 0.13251398503780365,
|
| 326 |
+
"rewards/rejected": -0.5444768667221069,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.17457218330079247,
|
| 331 |
+
"grad_norm": 0.1806156188249588,
|
| 332 |
+
"learning_rate": 4.804657878971252e-06,
|
| 333 |
+
"logits/chosen": 10.275301933288574,
|
| 334 |
+
"logits/rejected": 10.937273025512695,
|
| 335 |
+
"logps/chosen": -0.2880379557609558,
|
| 336 |
+
"logps/rejected": -0.4154580533504486,
|
| 337 |
+
"loss": 0.8875,
|
| 338 |
+
"rewards/accuracies": 0.625,
|
| 339 |
+
"rewards/chosen": -0.43205690383911133,
|
| 340 |
+
"rewards/margins": 0.19113019108772278,
|
| 341 |
+
"rewards/rejected": -0.6231871247291565,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.1837601929482026,
|
| 346 |
+
"grad_norm": 0.1839464157819748,
|
| 347 |
+
"learning_rate": 4.783863644106502e-06,
|
| 348 |
+
"logits/chosen": 10.020039558410645,
|
| 349 |
+
"logits/rejected": 10.66059398651123,
|
| 350 |
+
"logps/chosen": -0.3136019706726074,
|
| 351 |
+
"logps/rejected": -0.4385503828525543,
|
| 352 |
+
"loss": 0.8647,
|
| 353 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 354 |
+
"rewards/chosen": -0.47040295600891113,
|
| 355 |
+
"rewards/margins": 0.18742261826992035,
|
| 356 |
+
"rewards/rejected": -0.6578255891799927,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.1837601929482026,
|
| 361 |
+
"eval_logits/chosen": 9.442557334899902,
|
| 362 |
+
"eval_logits/rejected": 10.053345680236816,
|
| 363 |
+
"eval_logps/chosen": -0.3080674409866333,
|
| 364 |
+
"eval_logps/rejected": -0.4899139702320099,
|
| 365 |
+
"eval_loss": 0.8702690005302429,
|
| 366 |
+
"eval_rewards/accuracies": 0.6931818127632141,
|
| 367 |
+
"eval_rewards/chosen": -0.46210116147994995,
|
| 368 |
+
"eval_rewards/margins": 0.27276986837387085,
|
| 369 |
+
"eval_rewards/rejected": -0.7348710894584656,
|
| 370 |
+
"eval_runtime": 24.4185,
|
| 371 |
+
"eval_samples_per_second": 28.831,
|
| 372 |
+
"eval_steps_per_second": 3.604,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.19294820259561274,
|
| 377 |
+
"grad_norm": 0.269613116979599,
|
| 378 |
+
"learning_rate": 4.762067631165049e-06,
|
| 379 |
+
"logits/chosen": 7.941342353820801,
|
| 380 |
+
"logits/rejected": 8.542920112609863,
|
| 381 |
+
"logps/chosen": -0.3083941638469696,
|
| 382 |
+
"logps/rejected": -0.5024437308311462,
|
| 383 |
+
"loss": 0.8471,
|
| 384 |
+
"rewards/accuracies": 0.612500011920929,
|
| 385 |
+
"rewards/chosen": -0.4625912606716156,
|
| 386 |
+
"rewards/margins": 0.29107433557510376,
|
| 387 |
+
"rewards/rejected": -0.7536656856536865,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.20213621224302286,
|
| 392 |
+
"grad_norm": 0.2640094459056854,
|
| 393 |
+
"learning_rate": 4.7392794005985324e-06,
|
| 394 |
+
"logits/chosen": 7.587499141693115,
|
| 395 |
+
"logits/rejected": 7.592519283294678,
|
| 396 |
+
"logps/chosen": -0.3381899893283844,
|
| 397 |
+
"logps/rejected": -0.48494213819503784,
|
| 398 |
+
"loss": 0.8427,
|
| 399 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 400 |
+
"rewards/chosen": -0.5072849988937378,
|
| 401 |
+
"rewards/margins": 0.22012826800346375,
|
| 402 |
+
"rewards/rejected": -0.7274132966995239,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.21132422189043298,
|
| 407 |
+
"grad_norm": 0.29708293080329895,
|
| 408 |
+
"learning_rate": 4.715508948078037e-06,
|
| 409 |
+
"logits/chosen": 6.250656604766846,
|
| 410 |
+
"logits/rejected": 6.7652716636657715,
|
| 411 |
+
"logps/chosen": -0.3644888997077942,
|
| 412 |
+
"logps/rejected": -0.5470594167709351,
|
| 413 |
+
"loss": 0.8201,
|
| 414 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 415 |
+
"rewards/chosen": -0.5467333793640137,
|
| 416 |
+
"rewards/margins": 0.2738557755947113,
|
| 417 |
+
"rewards/rejected": -0.8205891847610474,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.2205122315378431,
|
| 422 |
+
"grad_norm": 0.35299497842788696,
|
| 423 |
+
"learning_rate": 4.690766700109659e-06,
|
| 424 |
+
"logits/chosen": 4.6331706047058105,
|
| 425 |
+
"logits/rejected": 4.710076332092285,
|
| 426 |
+
"logps/chosen": -0.3634452223777771,
|
| 427 |
+
"logps/rejected": -0.7193974256515503,
|
| 428 |
+
"loss": 0.7877,
|
| 429 |
+
"rewards/accuracies": 0.675000011920929,
|
| 430 |
+
"rewards/chosen": -0.545167863368988,
|
| 431 |
+
"rewards/margins": 0.5339283347129822,
|
| 432 |
+
"rewards/rejected": -1.0790963172912598,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.22970024118525326,
|
| 437 |
+
"grad_norm": 0.4265730082988739,
|
| 438 |
+
"learning_rate": 4.665063509461098e-06,
|
| 439 |
+
"logits/chosen": 4.992984771728516,
|
| 440 |
+
"logits/rejected": 4.606354713439941,
|
| 441 |
+
"logps/chosen": -0.413116455078125,
|
| 442 |
+
"logps/rejected": -0.7104976177215576,
|
| 443 |
+
"loss": 0.7902,
|
| 444 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 445 |
+
"rewards/chosen": -0.6196746826171875,
|
| 446 |
+
"rewards/margins": 0.4460717737674713,
|
| 447 |
+
"rewards/rejected": -1.0657463073730469,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.22970024118525326,
|
| 452 |
+
"eval_logits/chosen": 4.127804279327393,
|
| 453 |
+
"eval_logits/rejected": 3.742251396179199,
|
| 454 |
+
"eval_logps/chosen": -0.420327365398407,
|
| 455 |
+
"eval_logps/rejected": -0.7902651429176331,
|
| 456 |
+
"eval_loss": 0.7682384252548218,
|
| 457 |
+
"eval_rewards/accuracies": 0.7159090638160706,
|
| 458 |
+
"eval_rewards/chosen": -0.6304910182952881,
|
| 459 |
+
"eval_rewards/margins": 0.5549066662788391,
|
| 460 |
+
"eval_rewards/rejected": -1.185397744178772,
|
| 461 |
+
"eval_runtime": 24.4318,
|
| 462 |
+
"eval_samples_per_second": 28.815,
|
| 463 |
+
"eval_steps_per_second": 3.602,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.23888825083266338,
|
| 468 |
+
"grad_norm": 0.7236106395721436,
|
| 469 |
+
"learning_rate": 4.638410650401267e-06,
|
| 470 |
+
"logits/chosen": 2.454423427581787,
|
| 471 |
+
"logits/rejected": 1.816563367843628,
|
| 472 |
+
"logps/chosen": -0.4492695927619934,
|
| 473 |
+
"logps/rejected": -0.8738088607788086,
|
| 474 |
+
"loss": 0.6911,
|
| 475 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 476 |
+
"rewards/chosen": -0.6739044189453125,
|
| 477 |
+
"rewards/margins": 0.6368088126182556,
|
| 478 |
+
"rewards/rejected": -1.3107131719589233,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2480762604800735,
|
| 483 |
+
"grad_norm": 0.5856125950813293,
|
| 484 |
+
"learning_rate": 4.610819813755038e-06,
|
| 485 |
+
"logits/chosen": 3.2105612754821777,
|
| 486 |
+
"logits/rejected": 2.5531132221221924,
|
| 487 |
+
"logps/chosen": -0.537078320980072,
|
| 488 |
+
"logps/rejected": -1.2025481462478638,
|
| 489 |
+
"loss": 0.6774,
|
| 490 |
+
"rewards/accuracies": 0.625,
|
| 491 |
+
"rewards/chosen": -0.80561763048172,
|
| 492 |
+
"rewards/margins": 0.9982045888900757,
|
| 493 |
+
"rewards/rejected": -1.8038222789764404,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.25726427012748365,
|
| 498 |
+
"grad_norm": 0.7396731972694397,
|
| 499 |
+
"learning_rate": 4.582303101775249e-06,
|
| 500 |
+
"logits/chosen": 2.0327231884002686,
|
| 501 |
+
"logits/rejected": 1.4601097106933594,
|
| 502 |
+
"logps/chosen": -0.47658151388168335,
|
| 503 |
+
"logps/rejected": -1.3808696269989014,
|
| 504 |
+
"loss": 0.628,
|
| 505 |
+
"rewards/accuracies": 0.6875,
|
| 506 |
+
"rewards/chosen": -0.7148722410202026,
|
| 507 |
+
"rewards/margins": 1.3564319610595703,
|
| 508 |
+
"rewards/rejected": -2.0713045597076416,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.2664522797748938,
|
| 513 |
+
"grad_norm": 2.412203550338745,
|
| 514 |
+
"learning_rate": 4.55287302283426e-06,
|
| 515 |
+
"logits/chosen": 1.2262591123580933,
|
| 516 |
+
"logits/rejected": 0.22599482536315918,
|
| 517 |
+
"logps/chosen": -0.5671601891517639,
|
| 518 |
+
"logps/rejected": -1.6760343313217163,
|
| 519 |
+
"loss": 0.5988,
|
| 520 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 521 |
+
"rewards/chosen": -0.8507402539253235,
|
| 522 |
+
"rewards/margins": 1.6633113622665405,
|
| 523 |
+
"rewards/rejected": -2.5140514373779297,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.2756402894223039,
|
| 528 |
+
"grad_norm": 1.0477895736694336,
|
| 529 |
+
"learning_rate": 4.522542485937369e-06,
|
| 530 |
+
"logits/chosen": 1.9952911138534546,
|
| 531 |
+
"logits/rejected": 1.1298446655273438,
|
| 532 |
+
"logps/chosen": -0.6152974367141724,
|
| 533 |
+
"logps/rejected": -2.128481388092041,
|
| 534 |
+
"loss": 0.5927,
|
| 535 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 536 |
+
"rewards/chosen": -0.9229460954666138,
|
| 537 |
+
"rewards/margins": 2.2697763442993164,
|
| 538 |
+
"rewards/rejected": -3.192722797393799,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.2756402894223039,
|
| 543 |
+
"eval_logits/chosen": 1.6342910528182983,
|
| 544 |
+
"eval_logits/rejected": 0.633538007736206,
|
| 545 |
+
"eval_logps/chosen": -0.606099545955658,
|
| 546 |
+
"eval_logps/rejected": -1.882785439491272,
|
| 547 |
+
"eval_loss": 0.5978505611419678,
|
| 548 |
+
"eval_rewards/accuracies": 0.7159090638160706,
|
| 549 |
+
"eval_rewards/chosen": -0.909149169921875,
|
| 550 |
+
"eval_rewards/margins": 1.9150291681289673,
|
| 551 |
+
"eval_rewards/rejected": -2.824178457260132,
|
| 552 |
+
"eval_runtime": 24.4299,
|
| 553 |
+
"eval_samples_per_second": 28.817,
|
| 554 |
+
"eval_steps_per_second": 3.602,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.284828299069714,
|
| 559 |
+
"grad_norm": 3.0767388343811035,
|
| 560 |
+
"learning_rate": 4.491324795060491e-06,
|
| 561 |
+
"logits/chosen": 1.0374902486801147,
|
| 562 |
+
"logits/rejected": 0.6220051646232605,
|
| 563 |
+
"logps/chosen": -0.6778531074523926,
|
| 564 |
+
"logps/rejected": -1.8290255069732666,
|
| 565 |
+
"loss": 0.5792,
|
| 566 |
+
"rewards/accuracies": 0.625,
|
| 567 |
+
"rewards/chosen": -1.0167796611785889,
|
| 568 |
+
"rewards/margins": 1.726758599281311,
|
| 569 |
+
"rewards/rejected": -2.7435383796691895,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.29401630871712414,
|
| 574 |
+
"grad_norm": 0.6015120148658752,
|
| 575 |
+
"learning_rate": 4.4592336433146e-06,
|
| 576 |
+
"logits/chosen": 1.0050956010818481,
|
| 577 |
+
"logits/rejected": -0.016118621453642845,
|
| 578 |
+
"logps/chosen": -0.6865260004997253,
|
| 579 |
+
"logps/rejected": -2.113417148590088,
|
| 580 |
+
"loss": 0.5384,
|
| 581 |
+
"rewards/accuracies": 0.762499988079071,
|
| 582 |
+
"rewards/chosen": -1.0297890901565552,
|
| 583 |
+
"rewards/margins": 2.140336513519287,
|
| 584 |
+
"rewards/rejected": -3.1701254844665527,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.30320431836453426,
|
| 589 |
+
"grad_norm": 0.7415631413459778,
|
| 590 |
+
"learning_rate": 4.426283106939474e-06,
|
| 591 |
+
"logits/chosen": 0.5536144971847534,
|
| 592 |
+
"logits/rejected": -0.1644023358821869,
|
| 593 |
+
"logps/chosen": -0.8181726336479187,
|
| 594 |
+
"logps/rejected": -2.581185817718506,
|
| 595 |
+
"loss": 0.5671,
|
| 596 |
+
"rewards/accuracies": 0.75,
|
| 597 |
+
"rewards/chosen": -1.2272589206695557,
|
| 598 |
+
"rewards/margins": 2.644519805908203,
|
| 599 |
+
"rewards/rejected": -3.871778964996338,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.3123923280119444,
|
| 604 |
+
"grad_norm": 0.8956871628761292,
|
| 605 |
+
"learning_rate": 4.3924876391293915e-06,
|
| 606 |
+
"logits/chosen": 1.6062015295028687,
|
| 607 |
+
"logits/rejected": 0.9243733286857605,
|
| 608 |
+
"logps/chosen": -0.8991573452949524,
|
| 609 |
+
"logps/rejected": -2.935060977935791,
|
| 610 |
+
"loss": 0.5124,
|
| 611 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 612 |
+
"rewards/chosen": -1.348736047744751,
|
| 613 |
+
"rewards/margins": 3.0538551807403564,
|
| 614 |
+
"rewards/rejected": -4.402591228485107,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.32158033765935456,
|
| 619 |
+
"grad_norm": 1.1822264194488525,
|
| 620 |
+
"learning_rate": 4.357862063693486e-06,
|
| 621 |
+
"logits/chosen": 1.8171085119247437,
|
| 622 |
+
"logits/rejected": 1.228049397468567,
|
| 623 |
+
"logps/chosen": -0.8822734951972961,
|
| 624 |
+
"logps/rejected": -2.4744174480438232,
|
| 625 |
+
"loss": 0.516,
|
| 626 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 627 |
+
"rewards/chosen": -1.3234103918075562,
|
| 628 |
+
"rewards/margins": 2.388216018676758,
|
| 629 |
+
"rewards/rejected": -3.7116265296936035,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.32158033765935456,
|
| 634 |
+
"eval_logits/chosen": 1.1368330717086792,
|
| 635 |
+
"eval_logits/rejected": 0.2795785665512085,
|
| 636 |
+
"eval_logps/chosen": -0.9485350251197815,
|
| 637 |
+
"eval_logps/rejected": -2.6484899520874023,
|
| 638 |
+
"eval_loss": 0.5133901238441467,
|
| 639 |
+
"eval_rewards/accuracies": 0.7727272510528564,
|
| 640 |
+
"eval_rewards/chosen": -1.4228025674819946,
|
| 641 |
+
"eval_rewards/margins": 2.5499324798583984,
|
| 642 |
+
"eval_rewards/rejected": -3.9727354049682617,
|
| 643 |
+
"eval_runtime": 24.4277,
|
| 644 |
+
"eval_samples_per_second": 28.82,
|
| 645 |
+
"eval_steps_per_second": 3.602,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.3307683473067647,
|
| 650 |
+
"grad_norm": 2.5278775691986084,
|
| 651 |
+
"learning_rate": 4.322421568553529e-06,
|
| 652 |
+
"logits/chosen": 1.1921958923339844,
|
| 653 |
+
"logits/rejected": 0.7565670013427734,
|
| 654 |
+
"logps/chosen": -1.4180412292480469,
|
| 655 |
+
"logps/rejected": -3.0890870094299316,
|
| 656 |
+
"loss": 0.4811,
|
| 657 |
+
"rewards/accuracies": 0.800000011920929,
|
| 658 |
+
"rewards/chosen": -2.1270618438720703,
|
| 659 |
+
"rewards/margins": 2.5065689086914062,
|
| 660 |
+
"rewards/rejected": -4.633630752563477,
|
| 661 |
+
"step": 360
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.3399563569541748,
|
| 665 |
+
"grad_norm": 1.7325788736343384,
|
| 666 |
+
"learning_rate": 4.286181699082008e-06,
|
| 667 |
+
"logits/chosen": 0.997096836566925,
|
| 668 |
+
"logits/rejected": 0.4399908483028412,
|
| 669 |
+
"logps/chosen": -1.9010308980941772,
|
| 670 |
+
"logps/rejected": -3.6025185585021973,
|
| 671 |
+
"loss": 0.4326,
|
| 672 |
+
"rewards/accuracies": 0.824999988079071,
|
| 673 |
+
"rewards/chosen": -2.851546287536621,
|
| 674 |
+
"rewards/margins": 2.552231550216675,
|
| 675 |
+
"rewards/rejected": -5.403778076171875,
|
| 676 |
+
"step": 370
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.34914436660158493,
|
| 680 |
+
"grad_norm": 4.608370304107666,
|
| 681 |
+
"learning_rate": 4.249158351283414e-06,
|
| 682 |
+
"logits/chosen": 1.0715999603271484,
|
| 683 |
+
"logits/rejected": 0.6113725900650024,
|
| 684 |
+
"logps/chosen": -2.4032845497131348,
|
| 685 |
+
"logps/rejected": -3.9940528869628906,
|
| 686 |
+
"loss": 0.4027,
|
| 687 |
+
"rewards/accuracies": 0.875,
|
| 688 |
+
"rewards/chosen": -3.6049270629882812,
|
| 689 |
+
"rewards/margins": 2.3861520290374756,
|
| 690 |
+
"rewards/rejected": -5.991078853607178,
|
| 691 |
+
"step": 380
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.35833237624899505,
|
| 695 |
+
"grad_norm": 4.600816249847412,
|
| 696 |
+
"learning_rate": 4.211367764821722e-06,
|
| 697 |
+
"logits/chosen": 1.4443682432174683,
|
| 698 |
+
"logits/rejected": 0.8617011904716492,
|
| 699 |
+
"logps/chosen": -2.349093198776245,
|
| 700 |
+
"logps/rejected": -3.9943203926086426,
|
| 701 |
+
"loss": 0.4118,
|
| 702 |
+
"rewards/accuracies": 0.8125,
|
| 703 |
+
"rewards/chosen": -3.5236401557922363,
|
| 704 |
+
"rewards/margins": 2.467839479446411,
|
| 705 |
+
"rewards/rejected": -5.991480350494385,
|
| 706 |
+
"step": 390
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.3675203858964052,
|
| 710 |
+
"grad_norm": 2.1458094120025635,
|
| 711 |
+
"learning_rate": 4.172826515897146e-06,
|
| 712 |
+
"logits/chosen": 1.3029582500457764,
|
| 713 |
+
"logits/rejected": 0.7705980539321899,
|
| 714 |
+
"logps/chosen": -2.192617416381836,
|
| 715 |
+
"logps/rejected": -3.8413078784942627,
|
| 716 |
+
"loss": 0.3557,
|
| 717 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 718 |
+
"rewards/chosen": -3.288925886154175,
|
| 719 |
+
"rewards/margins": 2.4730350971221924,
|
| 720 |
+
"rewards/rejected": -5.761960506439209,
|
| 721 |
+
"step": 400
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.3675203858964052,
|
| 725 |
+
"eval_logits/chosen": 0.8742353320121765,
|
| 726 |
+
"eval_logits/rejected": 0.14320053160190582,
|
| 727 |
+
"eval_logps/chosen": -2.0894505977630615,
|
| 728 |
+
"eval_logps/rejected": -4.20783805847168,
|
| 729 |
+
"eval_loss": 0.4082850515842438,
|
| 730 |
+
"eval_rewards/accuracies": 0.8863636255264282,
|
| 731 |
+
"eval_rewards/chosen": -3.1341757774353027,
|
| 732 |
+
"eval_rewards/margins": 3.1775810718536377,
|
| 733 |
+
"eval_rewards/rejected": -6.311756610870361,
|
| 734 |
+
"eval_runtime": 24.4316,
|
| 735 |
+
"eval_samples_per_second": 28.815,
|
| 736 |
+
"eval_steps_per_second": 3.602,
|
| 737 |
+
"step": 400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.3767083955438153,
|
| 741 |
+
"grad_norm": 2.5496387481689453,
|
| 742 |
+
"learning_rate": 4.133551509975264e-06,
|
| 743 |
+
"logits/chosen": 1.9024279117584229,
|
| 744 |
+
"logits/rejected": 1.4777928590774536,
|
| 745 |
+
"logps/chosen": -2.7541470527648926,
|
| 746 |
+
"logps/rejected": -4.542642116546631,
|
| 747 |
+
"loss": 0.4184,
|
| 748 |
+
"rewards/accuracies": 0.887499988079071,
|
| 749 |
+
"rewards/chosen": -4.13122034072876,
|
| 750 |
+
"rewards/margins": 2.6827428340911865,
|
| 751 |
+
"rewards/rejected": -6.813962459564209,
|
| 752 |
+
"step": 410
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.3858964051912255,
|
| 756 |
+
"grad_norm": 2.6187174320220947,
|
| 757 |
+
"learning_rate": 4.093559974371725e-06,
|
| 758 |
+
"logits/chosen": 1.9345887899398804,
|
| 759 |
+
"logits/rejected": 1.4047685861587524,
|
| 760 |
+
"logps/chosen": -2.389430522918701,
|
| 761 |
+
"logps/rejected": -4.439882755279541,
|
| 762 |
+
"loss": 0.4096,
|
| 763 |
+
"rewards/accuracies": 0.875,
|
| 764 |
+
"rewards/chosen": -3.5841457843780518,
|
| 765 |
+
"rewards/margins": 3.0756776332855225,
|
| 766 |
+
"rewards/rejected": -6.659823417663574,
|
| 767 |
+
"step": 420
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.3950844148386356,
|
| 771 |
+
"grad_norm": 3.762899398803711,
|
| 772 |
+
"learning_rate": 4.052869450695776e-06,
|
| 773 |
+
"logits/chosen": 1.2521915435791016,
|
| 774 |
+
"logits/rejected": 0.7238092422485352,
|
| 775 |
+
"logps/chosen": -2.5571534633636475,
|
| 776 |
+
"logps/rejected": -4.675185203552246,
|
| 777 |
+
"loss": 0.4083,
|
| 778 |
+
"rewards/accuracies": 0.862500011920929,
|
| 779 |
+
"rewards/chosen": -3.835730791091919,
|
| 780 |
+
"rewards/margins": 3.1770474910736084,
|
| 781 |
+
"rewards/rejected": -7.012777805328369,
|
| 782 |
+
"step": 430
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.4042724244860457,
|
| 786 |
+
"grad_norm": 3.2343404293060303,
|
| 787 |
+
"learning_rate": 4.011497787155938e-06,
|
| 788 |
+
"logits/chosen": 1.9626567363739014,
|
| 789 |
+
"logits/rejected": 1.3136894702911377,
|
| 790 |
+
"logps/chosen": -2.892967939376831,
|
| 791 |
+
"logps/rejected": -5.003688335418701,
|
| 792 |
+
"loss": 0.3783,
|
| 793 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 794 |
+
"rewards/chosen": -4.339452266693115,
|
| 795 |
+
"rewards/margins": 3.1660804748535156,
|
| 796 |
+
"rewards/rejected": -7.505532264709473,
|
| 797 |
+
"step": 440
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.41346043413345585,
|
| 801 |
+
"grad_norm": 3.2979824542999268,
|
| 802 |
+
"learning_rate": 3.969463130731183e-06,
|
| 803 |
+
"logits/chosen": 1.159234881401062,
|
| 804 |
+
"logits/rejected": 0.5396692752838135,
|
| 805 |
+
"logps/chosen": -2.7545204162597656,
|
| 806 |
+
"logps/rejected": -5.4179182052612305,
|
| 807 |
+
"loss": 0.347,
|
| 808 |
+
"rewards/accuracies": 0.9375,
|
| 809 |
+
"rewards/chosen": -4.131781101226807,
|
| 810 |
+
"rewards/margins": 3.9950966835021973,
|
| 811 |
+
"rewards/rejected": -8.126876831054688,
|
| 812 |
+
"step": 450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.41346043413345585,
|
| 816 |
+
"eval_logits/chosen": 0.9304068088531494,
|
| 817 |
+
"eval_logits/rejected": 0.30015668272972107,
|
| 818 |
+
"eval_logps/chosen": -2.485308885574341,
|
| 819 |
+
"eval_logps/rejected": -4.880238056182861,
|
| 820 |
+
"eval_loss": 0.36501914262771606,
|
| 821 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 822 |
+
"eval_rewards/chosen": -3.7279627323150635,
|
| 823 |
+
"eval_rewards/margins": 3.592393636703491,
|
| 824 |
+
"eval_rewards/rejected": -7.320356845855713,
|
| 825 |
+
"eval_runtime": 24.4234,
|
| 826 |
+
"eval_samples_per_second": 28.825,
|
| 827 |
+
"eval_steps_per_second": 3.603,
|
| 828 |
+
"step": 450
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.42264844378086597,
|
| 832 |
+
"grad_norm": 3.155860185623169,
|
| 833 |
+
"learning_rate": 3.92678391921108e-06,
|
| 834 |
+
"logits/chosen": 1.633522391319275,
|
| 835 |
+
"logits/rejected": 1.0575059652328491,
|
| 836 |
+
"logps/chosen": -2.7214303016662598,
|
| 837 |
+
"logps/rejected": -4.766176223754883,
|
| 838 |
+
"loss": 0.3689,
|
| 839 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 840 |
+
"rewards/chosen": -4.082144737243652,
|
| 841 |
+
"rewards/margins": 3.0671191215515137,
|
| 842 |
+
"rewards/rejected": -7.149264335632324,
|
| 843 |
+
"step": 460
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.4318364534282761,
|
| 847 |
+
"grad_norm": 2.9949357509613037,
|
| 848 |
+
"learning_rate": 3.88347887310836e-06,
|
| 849 |
+
"logits/chosen": 1.3139212131500244,
|
| 850 |
+
"logits/rejected": 0.5838541388511658,
|
| 851 |
+
"logps/chosen": -2.5324313640594482,
|
| 852 |
+
"logps/rejected": -4.993292808532715,
|
| 853 |
+
"loss": 0.3587,
|
| 854 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 855 |
+
"rewards/chosen": -3.7986464500427246,
|
| 856 |
+
"rewards/margins": 3.6912925243377686,
|
| 857 |
+
"rewards/rejected": -7.489938259124756,
|
| 858 |
+
"step": 470
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.4410244630756862,
|
| 862 |
+
"grad_norm": 2.4548208713531494,
|
| 863 |
+
"learning_rate": 3.839566987447492e-06,
|
| 864 |
+
"logits/chosen": 1.2342166900634766,
|
| 865 |
+
"logits/rejected": 0.8118699193000793,
|
| 866 |
+
"logps/chosen": -2.876863479614258,
|
| 867 |
+
"logps/rejected": -5.457588195800781,
|
| 868 |
+
"loss": 0.2922,
|
| 869 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 870 |
+
"rewards/chosen": -4.315295219421387,
|
| 871 |
+
"rewards/margins": 3.871086597442627,
|
| 872 |
+
"rewards/rejected": -8.186381340026855,
|
| 873 |
+
"step": 480
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.45021247272309634,
|
| 877 |
+
"grad_norm": 2.0953762531280518,
|
| 878 |
+
"learning_rate": 3.795067523432826e-06,
|
| 879 |
+
"logits/chosen": 1.8121936321258545,
|
| 880 |
+
"logits/rejected": 1.454637050628662,
|
| 881 |
+
"logps/chosen": -3.2022106647491455,
|
| 882 |
+
"logps/rejected": -5.47930908203125,
|
| 883 |
+
"loss": 0.3086,
|
| 884 |
+
"rewards/accuracies": 0.9375,
|
| 885 |
+
"rewards/chosen": -4.803316116333008,
|
| 886 |
+
"rewards/margins": 3.4156479835510254,
|
| 887 |
+
"rewards/rejected": -8.218963623046875,
|
| 888 |
+
"step": 490
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.4594004823705065,
|
| 892 |
+
"grad_norm": 2.4918301105499268,
|
| 893 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 894 |
+
"logits/chosen": 1.5386875867843628,
|
| 895 |
+
"logits/rejected": 1.2544763088226318,
|
| 896 |
+
"logps/chosen": -3.2174277305603027,
|
| 897 |
+
"logps/rejected": -6.012864112854004,
|
| 898 |
+
"loss": 0.3128,
|
| 899 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 900 |
+
"rewards/chosen": -4.826140880584717,
|
| 901 |
+
"rewards/margins": 4.193154335021973,
|
| 902 |
+
"rewards/rejected": -9.019296646118164,
|
| 903 |
+
"step": 500
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.4594004823705065,
|
| 907 |
+
"eval_logits/chosen": 0.8257808685302734,
|
| 908 |
+
"eval_logits/rejected": 0.27475622296333313,
|
| 909 |
+
"eval_logps/chosen": -2.755974054336548,
|
| 910 |
+
"eval_logps/rejected": -5.469714641571045,
|
| 911 |
+
"eval_loss": 0.32884541153907776,
|
| 912 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 913 |
+
"eval_rewards/chosen": -4.133961200714111,
|
| 914 |
+
"eval_rewards/margins": 4.070610523223877,
|
| 915 |
+
"eval_rewards/rejected": -8.204572677612305,
|
| 916 |
+
"eval_runtime": 24.4191,
|
| 917 |
+
"eval_samples_per_second": 28.83,
|
| 918 |
+
"eval_steps_per_second": 3.604,
|
| 919 |
+
"step": 500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.46858849201791664,
|
| 923 |
+
"grad_norm": 4.017474174499512,
|
| 924 |
+
"learning_rate": 3.7043841852542884e-06,
|
| 925 |
+
"logits/chosen": 1.8042447566986084,
|
| 926 |
+
"logits/rejected": 1.4390740394592285,
|
| 927 |
+
"logps/chosen": -3.4311797618865967,
|
| 928 |
+
"logps/rejected": -6.114380359649658,
|
| 929 |
+
"loss": 0.335,
|
| 930 |
+
"rewards/accuracies": 0.925000011920929,
|
| 931 |
+
"rewards/chosen": -5.1467695236206055,
|
| 932 |
+
"rewards/margins": 4.0248003005981445,
|
| 933 |
+
"rewards/rejected": -9.171568870544434,
|
| 934 |
+
"step": 510
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.47777650166532676,
|
| 938 |
+
"grad_norm": 5.516397953033447,
|
| 939 |
+
"learning_rate": 3.658240087799655e-06,
|
| 940 |
+
"logits/chosen": 1.1141811609268188,
|
| 941 |
+
"logits/rejected": 0.7766789197921753,
|
| 942 |
+
"logps/chosen": -3.1831612586975098,
|
| 943 |
+
"logps/rejected": -5.615653991699219,
|
| 944 |
+
"loss": 0.3306,
|
| 945 |
+
"rewards/accuracies": 0.887499988079071,
|
| 946 |
+
"rewards/chosen": -4.774742603302002,
|
| 947 |
+
"rewards/margins": 3.6487393379211426,
|
| 948 |
+
"rewards/rejected": -8.423480987548828,
|
| 949 |
+
"step": 520
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.4869645113127369,
|
| 953 |
+
"grad_norm": 4.1005635261535645,
|
| 954 |
+
"learning_rate": 3.611587947962319e-06,
|
| 955 |
+
"logits/chosen": 1.5796890258789062,
|
| 956 |
+
"logits/rejected": 1.1561863422393799,
|
| 957 |
+
"logps/chosen": -3.286179304122925,
|
| 958 |
+
"logps/rejected": -5.805339813232422,
|
| 959 |
+
"loss": 0.3151,
|
| 960 |
+
"rewards/accuracies": 0.9375,
|
| 961 |
+
"rewards/chosen": -4.929268836975098,
|
| 962 |
+
"rewards/margins": 3.778740644454956,
|
| 963 |
+
"rewards/rejected": -8.708009719848633,
|
| 964 |
+
"step": 530
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.496152520960147,
|
| 968 |
+
"grad_norm": 2.8616511821746826,
|
| 969 |
+
"learning_rate": 3.564448228912682e-06,
|
| 970 |
+
"logits/chosen": 0.6956934928894043,
|
| 971 |
+
"logits/rejected": 0.1268310248851776,
|
| 972 |
+
"logps/chosen": -2.8933937549591064,
|
| 973 |
+
"logps/rejected": -5.794272422790527,
|
| 974 |
+
"loss": 0.2995,
|
| 975 |
+
"rewards/accuracies": 0.9375,
|
| 976 |
+
"rewards/chosen": -4.340090274810791,
|
| 977 |
+
"rewards/margins": 4.351318836212158,
|
| 978 |
+
"rewards/rejected": -8.691408157348633,
|
| 979 |
+
"step": 540
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 0.5053405306075571,
|
| 983 |
+
"grad_norm": 2.237276315689087,
|
| 984 |
+
"learning_rate": 3.516841607689501e-06,
|
| 985 |
+
"logits/chosen": 1.7460235357284546,
|
| 986 |
+
"logits/rejected": 0.9990445375442505,
|
| 987 |
+
"logps/chosen": -2.860546827316284,
|
| 988 |
+
"logps/rejected": -5.517810821533203,
|
| 989 |
+
"loss": 0.3203,
|
| 990 |
+
"rewards/accuracies": 0.887499988079071,
|
| 991 |
+
"rewards/chosen": -4.290820121765137,
|
| 992 |
+
"rewards/margins": 3.985896348953247,
|
| 993 |
+
"rewards/rejected": -8.276716232299805,
|
| 994 |
+
"step": 550
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.5053405306075571,
|
| 998 |
+
"eval_logits/chosen": 0.8703196048736572,
|
| 999 |
+
"eval_logits/rejected": 0.3152187466621399,
|
| 1000 |
+
"eval_logps/chosen": -2.709866523742676,
|
| 1001 |
+
"eval_logps/rejected": -5.687611103057861,
|
| 1002 |
+
"eval_loss": 0.304200142621994,
|
| 1003 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 1004 |
+
"eval_rewards/chosen": -4.064799785614014,
|
| 1005 |
+
"eval_rewards/margins": 4.466617107391357,
|
| 1006 |
+
"eval_rewards/rejected": -8.531416893005371,
|
| 1007 |
+
"eval_runtime": 24.6083,
|
| 1008 |
+
"eval_samples_per_second": 28.608,
|
| 1009 |
+
"eval_steps_per_second": 3.576,
|
| 1010 |
+
"step": 550
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.5145285402549673,
|
| 1014 |
+
"grad_norm": 3.7983174324035645,
|
| 1015 |
+
"learning_rate": 3.4687889661302577e-06,
|
| 1016 |
+
"logits/chosen": 1.084142804145813,
|
| 1017 |
+
"logits/rejected": 0.9131366610527039,
|
| 1018 |
+
"logps/chosen": -3.4958484172821045,
|
| 1019 |
+
"logps/rejected": -7.143038749694824,
|
| 1020 |
+
"loss": 0.2728,
|
| 1021 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1022 |
+
"rewards/chosen": -5.243772506713867,
|
| 1023 |
+
"rewards/margins": 5.470786094665527,
|
| 1024 |
+
"rewards/rejected": -10.714558601379395,
|
| 1025 |
+
"step": 560
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.5237165499023774,
|
| 1029 |
+
"grad_norm": 2.398188829421997,
|
| 1030 |
+
"learning_rate": 3.4203113817116955e-06,
|
| 1031 |
+
"logits/chosen": 1.2777886390686035,
|
| 1032 |
+
"logits/rejected": 0.6827106475830078,
|
| 1033 |
+
"logps/chosen": -3.4014134407043457,
|
| 1034 |
+
"logps/rejected": -6.260494232177734,
|
| 1035 |
+
"loss": 0.2833,
|
| 1036 |
+
"rewards/accuracies": 0.9375,
|
| 1037 |
+
"rewards/chosen": -5.102120399475098,
|
| 1038 |
+
"rewards/margins": 4.288620948791504,
|
| 1039 |
+
"rewards/rejected": -9.390741348266602,
|
| 1040 |
+
"step": 570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.5329045595497875,
|
| 1044 |
+
"grad_norm": 4.97003173828125,
|
| 1045 |
+
"learning_rate": 3.3714301183045382e-06,
|
| 1046 |
+
"logits/chosen": 1.6978384256362915,
|
| 1047 |
+
"logits/rejected": 1.1092720031738281,
|
| 1048 |
+
"logps/chosen": -3.179011106491089,
|
| 1049 |
+
"logps/rejected": -6.802459716796875,
|
| 1050 |
+
"loss": 0.2942,
|
| 1051 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1052 |
+
"rewards/chosen": -4.768516540527344,
|
| 1053 |
+
"rewards/margins": 5.435172080993652,
|
| 1054 |
+
"rewards/rejected": -10.203688621520996,
|
| 1055 |
+
"step": 580
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.5420925691971976,
|
| 1059 |
+
"grad_norm": 4.482264995574951,
|
| 1060 |
+
"learning_rate": 3.3221666168464584e-06,
|
| 1061 |
+
"logits/chosen": 1.0245933532714844,
|
| 1062 |
+
"logits/rejected": 0.7797524929046631,
|
| 1063 |
+
"logps/chosen": -3.2393958568573,
|
| 1064 |
+
"logps/rejected": -5.916412353515625,
|
| 1065 |
+
"loss": 0.3347,
|
| 1066 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1067 |
+
"rewards/chosen": -4.859094142913818,
|
| 1068 |
+
"rewards/margins": 4.015524864196777,
|
| 1069 |
+
"rewards/rejected": -8.874618530273438,
|
| 1070 |
+
"step": 590
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.5512805788446078,
|
| 1074 |
+
"grad_norm": 3.913116931915283,
|
| 1075 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1076 |
+
"logits/chosen": 1.3320618867874146,
|
| 1077 |
+
"logits/rejected": 0.9762558937072754,
|
| 1078 |
+
"logps/chosen": -3.2751450538635254,
|
| 1079 |
+
"logps/rejected": -6.7808518409729,
|
| 1080 |
+
"loss": 0.2776,
|
| 1081 |
+
"rewards/accuracies": 0.875,
|
| 1082 |
+
"rewards/chosen": -4.912716865539551,
|
| 1083 |
+
"rewards/margins": 5.258560657501221,
|
| 1084 |
+
"rewards/rejected": -10.17127799987793,
|
| 1085 |
+
"step": 600
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.5512805788446078,
|
| 1089 |
+
"eval_logits/chosen": 0.9448758959770203,
|
| 1090 |
+
"eval_logits/rejected": 0.42817041277885437,
|
| 1091 |
+
"eval_logps/chosen": -2.835143804550171,
|
| 1092 |
+
"eval_logps/rejected": -6.087582588195801,
|
| 1093 |
+
"eval_loss": 0.28658536076545715,
|
| 1094 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 1095 |
+
"eval_rewards/chosen": -4.252715587615967,
|
| 1096 |
+
"eval_rewards/margins": 4.878659248352051,
|
| 1097 |
+
"eval_rewards/rejected": -9.13137435913086,
|
| 1098 |
+
"eval_runtime": 24.4325,
|
| 1099 |
+
"eval_samples_per_second": 28.814,
|
| 1100 |
+
"eval_steps_per_second": 3.602,
|
| 1101 |
+
"step": 600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.560468588492018,
|
| 1105 |
+
"grad_norm": 3.8335089683532715,
|
| 1106 |
+
"learning_rate": 3.222579492361179e-06,
|
| 1107 |
+
"logits/chosen": 1.295569658279419,
|
| 1108 |
+
"logits/rejected": 1.0852326154708862,
|
| 1109 |
+
"logps/chosen": -3.4870052337646484,
|
| 1110 |
+
"logps/rejected": -6.572986602783203,
|
| 1111 |
+
"loss": 0.3243,
|
| 1112 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1113 |
+
"rewards/chosen": -5.230508327484131,
|
| 1114 |
+
"rewards/margins": 4.628971576690674,
|
| 1115 |
+
"rewards/rejected": -9.859478950500488,
|
| 1116 |
+
"step": 610
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.569656598139428,
|
| 1120 |
+
"grad_norm": 3.095102071762085,
|
| 1121 |
+
"learning_rate": 3.1722995515381644e-06,
|
| 1122 |
+
"logits/chosen": 1.8339803218841553,
|
| 1123 |
+
"logits/rejected": 1.467551589012146,
|
| 1124 |
+
"logps/chosen": -3.6670470237731934,
|
| 1125 |
+
"logps/rejected": -6.880410194396973,
|
| 1126 |
+
"loss": 0.2615,
|
| 1127 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1128 |
+
"rewards/chosen": -5.500569820404053,
|
| 1129 |
+
"rewards/margins": 4.820044040679932,
|
| 1130 |
+
"rewards/rejected": -10.3206148147583,
|
| 1131 |
+
"step": 620
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.5788446077868382,
|
| 1135 |
+
"grad_norm": 4.381973743438721,
|
| 1136 |
+
"learning_rate": 3.121724717912138e-06,
|
| 1137 |
+
"logits/chosen": 2.390763521194458,
|
| 1138 |
+
"logits/rejected": 2.155505418777466,
|
| 1139 |
+
"logps/chosen": -3.3569788932800293,
|
| 1140 |
+
"logps/rejected": -6.319291114807129,
|
| 1141 |
+
"loss": 0.2649,
|
| 1142 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1143 |
+
"rewards/chosen": -5.035468101501465,
|
| 1144 |
+
"rewards/margins": 4.4434685707092285,
|
| 1145 |
+
"rewards/rejected": -9.478937149047852,
|
| 1146 |
+
"step": 630
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5880326174342483,
|
| 1150 |
+
"grad_norm": 3.9204964637756348,
|
| 1151 |
+
"learning_rate": 3.0708771752766397e-06,
|
| 1152 |
+
"logits/chosen": 1.4343383312225342,
|
| 1153 |
+
"logits/rejected": 1.1991338729858398,
|
| 1154 |
+
"logps/chosen": -3.654259443283081,
|
| 1155 |
+
"logps/rejected": -6.873109340667725,
|
| 1156 |
+
"loss": 0.2761,
|
| 1157 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1158 |
+
"rewards/chosen": -5.481389045715332,
|
| 1159 |
+
"rewards/margins": 4.828273773193359,
|
| 1160 |
+
"rewards/rejected": -10.309663772583008,
|
| 1161 |
+
"step": 640
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.5972206270816585,
|
| 1165 |
+
"grad_norm": 3.179067373275757,
|
| 1166 |
+
"learning_rate": 3.019779227044398e-06,
|
| 1167 |
+
"logits/chosen": 2.0368576049804688,
|
| 1168 |
+
"logits/rejected": 1.8037185668945312,
|
| 1169 |
+
"logps/chosen": -3.791682720184326,
|
| 1170 |
+
"logps/rejected": -6.912911415100098,
|
| 1171 |
+
"loss": 0.2739,
|
| 1172 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1173 |
+
"rewards/chosen": -5.68752384185791,
|
| 1174 |
+
"rewards/margins": 4.68184232711792,
|
| 1175 |
+
"rewards/rejected": -10.369367599487305,
|
| 1176 |
+
"step": 650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.5972206270816585,
|
| 1180 |
+
"eval_logits/chosen": 1.0889618396759033,
|
| 1181 |
+
"eval_logits/rejected": 0.5887767672538757,
|
| 1182 |
+
"eval_logps/chosen": -3.0191192626953125,
|
| 1183 |
+
"eval_logps/rejected": -6.388964653015137,
|
| 1184 |
+
"eval_loss": 0.27443525195121765,
|
| 1185 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 1186 |
+
"eval_rewards/chosen": -4.528679370880127,
|
| 1187 |
+
"eval_rewards/margins": 5.054768085479736,
|
| 1188 |
+
"eval_rewards/rejected": -9.583446502685547,
|
| 1189 |
+
"eval_runtime": 24.4235,
|
| 1190 |
+
"eval_samples_per_second": 28.825,
|
| 1191 |
+
"eval_steps_per_second": 3.603,
|
| 1192 |
+
"step": 650
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.6064086367290685,
|
| 1196 |
+
"grad_norm": 2.804940700531006,
|
| 1197 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 1198 |
+
"logits/chosen": 1.8347485065460205,
|
| 1199 |
+
"logits/rejected": 1.3636573553085327,
|
| 1200 |
+
"logps/chosen": -3.398799419403076,
|
| 1201 |
+
"logps/rejected": -6.337627410888672,
|
| 1202 |
+
"loss": 0.2678,
|
| 1203 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1204 |
+
"rewards/chosen": -5.098199367523193,
|
| 1205 |
+
"rewards/margins": 4.408241271972656,
|
| 1206 |
+
"rewards/rejected": -9.506441116333008,
|
| 1207 |
+
"step": 660
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.6155966463764787,
|
| 1211 |
+
"grad_norm": 3.7368969917297363,
|
| 1212 |
+
"learning_rate": 2.9169218667902562e-06,
|
| 1213 |
+
"logits/chosen": 1.6734691858291626,
|
| 1214 |
+
"logits/rejected": 1.0945308208465576,
|
| 1215 |
+
"logps/chosen": -3.354790210723877,
|
| 1216 |
+
"logps/rejected": -6.3691864013671875,
|
| 1217 |
+
"loss": 0.294,
|
| 1218 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1219 |
+
"rewards/chosen": -5.0321855545043945,
|
| 1220 |
+
"rewards/margins": 4.521594524383545,
|
| 1221 |
+
"rewards/rejected": -9.553780555725098,
|
| 1222 |
+
"step": 670
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.6247846560238888,
|
| 1226 |
+
"grad_norm": 4.483130931854248,
|
| 1227 |
+
"learning_rate": 2.8652075714060296e-06,
|
| 1228 |
+
"logits/chosen": 1.640228509902954,
|
| 1229 |
+
"logits/rejected": 1.1519067287445068,
|
| 1230 |
+
"logps/chosen": -3.4700570106506348,
|
| 1231 |
+
"logps/rejected": -6.6276445388793945,
|
| 1232 |
+
"loss": 0.23,
|
| 1233 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1234 |
+
"rewards/chosen": -5.205085277557373,
|
| 1235 |
+
"rewards/margins": 4.736380577087402,
|
| 1236 |
+
"rewards/rejected": -9.941465377807617,
|
| 1237 |
+
"step": 680
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.633972665671299,
|
| 1241 |
+
"grad_norm": 8.508203506469727,
|
| 1242 |
+
"learning_rate": 2.813333083910761e-06,
|
| 1243 |
+
"logits/chosen": 1.0501906871795654,
|
| 1244 |
+
"logits/rejected": 0.5691097974777222,
|
| 1245 |
+
"logps/chosen": -3.4982333183288574,
|
| 1246 |
+
"logps/rejected": -7.144225120544434,
|
| 1247 |
+
"loss": 0.2808,
|
| 1248 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1249 |
+
"rewards/chosen": -5.247350215911865,
|
| 1250 |
+
"rewards/margins": 5.468987464904785,
|
| 1251 |
+
"rewards/rejected": -10.716337203979492,
|
| 1252 |
+
"step": 690
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.6431606753187091,
|
| 1256 |
+
"grad_norm": 2.271857738494873,
|
| 1257 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1258 |
+
"logits/chosen": 1.2119544744491577,
|
| 1259 |
+
"logits/rejected": 0.7247776389122009,
|
| 1260 |
+
"logps/chosen": -4.096956253051758,
|
| 1261 |
+
"logps/rejected": -7.4536261558532715,
|
| 1262 |
+
"loss": 0.2808,
|
| 1263 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1264 |
+
"rewards/chosen": -6.145434379577637,
|
| 1265 |
+
"rewards/margins": 5.035003662109375,
|
| 1266 |
+
"rewards/rejected": -11.180438995361328,
|
| 1267 |
+
"step": 700
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 0.6431606753187091,
|
| 1271 |
+
"eval_logits/chosen": 1.085008978843689,
|
| 1272 |
+
"eval_logits/rejected": 0.5991834998130798,
|
| 1273 |
+
"eval_logps/chosen": -3.179896354675293,
|
| 1274 |
+
"eval_logps/rejected": -6.738409519195557,
|
| 1275 |
+
"eval_loss": 0.26070258021354675,
|
| 1276 |
+
"eval_rewards/accuracies": 0.9090909361839294,
|
| 1277 |
+
"eval_rewards/chosen": -4.7698445320129395,
|
| 1278 |
+
"eval_rewards/margins": 5.337769031524658,
|
| 1279 |
+
"eval_rewards/rejected": -10.107613563537598,
|
| 1280 |
+
"eval_runtime": 24.465,
|
| 1281 |
+
"eval_samples_per_second": 28.776,
|
| 1282 |
+
"eval_steps_per_second": 3.597,
|
| 1283 |
+
"step": 700
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.6523486849661192,
|
| 1287 |
+
"grad_norm": 4.058814525604248,
|
| 1288 |
+
"learning_rate": 2.70919460833079e-06,
|
| 1289 |
+
"logits/chosen": 1.2182409763336182,
|
| 1290 |
+
"logits/rejected": 1.0481547117233276,
|
| 1291 |
+
"logps/chosen": -3.532151699066162,
|
| 1292 |
+
"logps/rejected": -7.4123854637146,
|
| 1293 |
+
"loss": 0.2565,
|
| 1294 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1295 |
+
"rewards/chosen": -5.2982282638549805,
|
| 1296 |
+
"rewards/margins": 5.8203511238098145,
|
| 1297 |
+
"rewards/rejected": -11.118578910827637,
|
| 1298 |
+
"step": 710
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.6615366946135294,
|
| 1302 |
+
"grad_norm": 3.857574939727783,
|
| 1303 |
+
"learning_rate": 2.6569762988232838e-06,
|
| 1304 |
+
"logits/chosen": 0.8797380328178406,
|
| 1305 |
+
"logits/rejected": 0.6432709693908691,
|
| 1306 |
+
"logps/chosen": -3.3780128955841064,
|
| 1307 |
+
"logps/rejected": -6.922214508056641,
|
| 1308 |
+
"loss": 0.2321,
|
| 1309 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1310 |
+
"rewards/chosen": -5.067019462585449,
|
| 1311 |
+
"rewards/margins": 5.3163018226623535,
|
| 1312 |
+
"rewards/rejected": -10.383320808410645,
|
| 1313 |
+
"step": 720
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.6707247042609394,
|
| 1317 |
+
"grad_norm": 3.7125725746154785,
|
| 1318 |
+
"learning_rate": 2.604689134322999e-06,
|
| 1319 |
+
"logits/chosen": 1.447474479675293,
|
| 1320 |
+
"logits/rejected": 1.1114810705184937,
|
| 1321 |
+
"logps/chosen": -3.682945728302002,
|
| 1322 |
+
"logps/rejected": -7.297152519226074,
|
| 1323 |
+
"loss": 0.2812,
|
| 1324 |
+
"rewards/accuracies": 0.9375,
|
| 1325 |
+
"rewards/chosen": -5.524418830871582,
|
| 1326 |
+
"rewards/margins": 5.421310901641846,
|
| 1327 |
+
"rewards/rejected": -10.94572925567627,
|
| 1328 |
+
"step": 730
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.6799127139083496,
|
| 1332 |
+
"grad_norm": 2.026604652404785,
|
| 1333 |
+
"learning_rate": 2.5523560497083927e-06,
|
| 1334 |
+
"logits/chosen": 1.9674856662750244,
|
| 1335 |
+
"logits/rejected": 1.593400239944458,
|
| 1336 |
+
"logps/chosen": -3.3987834453582764,
|
| 1337 |
+
"logps/rejected": -6.970278263092041,
|
| 1338 |
+
"loss": 0.2603,
|
| 1339 |
+
"rewards/accuracies": 0.9375,
|
| 1340 |
+
"rewards/chosen": -5.098175048828125,
|
| 1341 |
+
"rewards/margins": 5.357242107391357,
|
| 1342 |
+
"rewards/rejected": -10.45541763305664,
|
| 1343 |
+
"step": 740
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.6891007235557597,
|
| 1347 |
+
"grad_norm": 2.357570171356201,
|
| 1348 |
+
"learning_rate": 2.5e-06,
|
| 1349 |
+
"logits/chosen": 1.9729163646697998,
|
| 1350 |
+
"logits/rejected": 1.8791675567626953,
|
| 1351 |
+
"logps/chosen": -3.3879425525665283,
|
| 1352 |
+
"logps/rejected": -6.712514400482178,
|
| 1353 |
+
"loss": 0.2348,
|
| 1354 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1355 |
+
"rewards/chosen": -5.081913948059082,
|
| 1356 |
+
"rewards/margins": 4.986858367919922,
|
| 1357 |
+
"rewards/rejected": -10.068772315979004,
|
| 1358 |
+
"step": 750
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.6891007235557597,
|
| 1362 |
+
"eval_logits/chosen": 1.1882920265197754,
|
| 1363 |
+
"eval_logits/rejected": 0.7092404961585999,
|
| 1364 |
+
"eval_logps/chosen": -2.855703592300415,
|
| 1365 |
+
"eval_logps/rejected": -6.557665824890137,
|
| 1366 |
+
"eval_loss": 0.25226154923439026,
|
| 1367 |
+
"eval_rewards/accuracies": 0.9090909361839294,
|
| 1368 |
+
"eval_rewards/chosen": -4.283555507659912,
|
| 1369 |
+
"eval_rewards/margins": 5.552942276000977,
|
| 1370 |
+
"eval_rewards/rejected": -9.836498260498047,
|
| 1371 |
+
"eval_runtime": 24.4236,
|
| 1372 |
+
"eval_samples_per_second": 28.825,
|
| 1373 |
+
"eval_steps_per_second": 3.603,
|
| 1374 |
+
"step": 750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.6982887332031699,
|
| 1378 |
+
"grad_norm": 3.1369588375091553,
|
| 1379 |
+
"learning_rate": 2.447643950291608e-06,
|
| 1380 |
+
"logits/chosen": 1.908418893814087,
|
| 1381 |
+
"logits/rejected": 1.637202262878418,
|
| 1382 |
+
"logps/chosen": -3.941102981567383,
|
| 1383 |
+
"logps/rejected": -7.562623023986816,
|
| 1384 |
+
"loss": 0.2412,
|
| 1385 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1386 |
+
"rewards/chosen": -5.911653995513916,
|
| 1387 |
+
"rewards/margins": 5.432280540466309,
|
| 1388 |
+
"rewards/rejected": -11.343935012817383,
|
| 1389 |
+
"step": 760
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.70747674285058,
|
| 1393 |
+
"grad_norm": 4.2312703132629395,
|
| 1394 |
+
"learning_rate": 2.3953108656770018e-06,
|
| 1395 |
+
"logits/chosen": 1.3125172853469849,
|
| 1396 |
+
"logits/rejected": 0.933386504650116,
|
| 1397 |
+
"logps/chosen": -3.5281143188476562,
|
| 1398 |
+
"logps/rejected": -7.116488456726074,
|
| 1399 |
+
"loss": 0.2528,
|
| 1400 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1401 |
+
"rewards/chosen": -5.292171001434326,
|
| 1402 |
+
"rewards/margins": 5.382561683654785,
|
| 1403 |
+
"rewards/rejected": -10.674734115600586,
|
| 1404 |
+
"step": 770
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.7166647524979901,
|
| 1408 |
+
"grad_norm": 5.6951799392700195,
|
| 1409 |
+
"learning_rate": 2.3430237011767166e-06,
|
| 1410 |
+
"logits/chosen": 2.295801877975464,
|
| 1411 |
+
"logits/rejected": 1.8810745477676392,
|
| 1412 |
+
"logps/chosen": -3.3887531757354736,
|
| 1413 |
+
"logps/rejected": -6.8857245445251465,
|
| 1414 |
+
"loss": 0.2672,
|
| 1415 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1416 |
+
"rewards/chosen": -5.083128929138184,
|
| 1417 |
+
"rewards/margins": 5.245457649230957,
|
| 1418 |
+
"rewards/rejected": -10.32858657836914,
|
| 1419 |
+
"step": 780
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.7258527621454003,
|
| 1423 |
+
"grad_norm": 3.734528064727783,
|
| 1424 |
+
"learning_rate": 2.290805391669212e-06,
|
| 1425 |
+
"logits/chosen": 1.6154979467391968,
|
| 1426 |
+
"logits/rejected": 1.1886816024780273,
|
| 1427 |
+
"logps/chosen": -3.379617214202881,
|
| 1428 |
+
"logps/rejected": -7.1727166175842285,
|
| 1429 |
+
"loss": 0.2315,
|
| 1430 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1431 |
+
"rewards/chosen": -5.069426536560059,
|
| 1432 |
+
"rewards/margins": 5.689648628234863,
|
| 1433 |
+
"rewards/rejected": -10.759074211120605,
|
| 1434 |
+
"step": 790
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.7350407717928104,
|
| 1438 |
+
"grad_norm": 3.2450063228607178,
|
| 1439 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1440 |
+
"logits/chosen": 1.9270827770233154,
|
| 1441 |
+
"logits/rejected": 1.5745903253555298,
|
| 1442 |
+
"logps/chosen": -3.9242210388183594,
|
| 1443 |
+
"logps/rejected": -7.254105567932129,
|
| 1444 |
+
"loss": 0.2301,
|
| 1445 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1446 |
+
"rewards/chosen": -5.886331081390381,
|
| 1447 |
+
"rewards/margins": 4.9948272705078125,
|
| 1448 |
+
"rewards/rejected": -10.881157875061035,
|
| 1449 |
+
"step": 800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.7350407717928104,
|
| 1453 |
+
"eval_logits/chosen": 1.228877067565918,
|
| 1454 |
+
"eval_logits/rejected": 0.7780414819717407,
|
| 1455 |
+
"eval_logps/chosen": -3.096149444580078,
|
| 1456 |
+
"eval_logps/rejected": -6.9255452156066895,
|
| 1457 |
+
"eval_loss": 0.23762211203575134,
|
| 1458 |
+
"eval_rewards/accuracies": 0.9090909361839294,
|
| 1459 |
+
"eval_rewards/chosen": -4.644224166870117,
|
| 1460 |
+
"eval_rewards/margins": 5.744093894958496,
|
| 1461 |
+
"eval_rewards/rejected": -10.388318061828613,
|
| 1462 |
+
"eval_runtime": 24.4202,
|
| 1463 |
+
"eval_samples_per_second": 28.829,
|
| 1464 |
+
"eval_steps_per_second": 3.604,
|
| 1465 |
+
"step": 800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.7442287814402205,
|
| 1469 |
+
"grad_norm": 3.6619718074798584,
|
| 1470 |
+
"learning_rate": 2.186666916089239e-06,
|
| 1471 |
+
"logits/chosen": 2.1697134971618652,
|
| 1472 |
+
"logits/rejected": 1.8535985946655273,
|
| 1473 |
+
"logps/chosen": -3.306549549102783,
|
| 1474 |
+
"logps/rejected": -6.570471286773682,
|
| 1475 |
+
"loss": 0.2239,
|
| 1476 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1477 |
+
"rewards/chosen": -4.959824562072754,
|
| 1478 |
+
"rewards/margins": 4.895882606506348,
|
| 1479 |
+
"rewards/rejected": -9.855707168579102,
|
| 1480 |
+
"step": 810
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.7534167910876306,
|
| 1484 |
+
"grad_norm": 3.9014956951141357,
|
| 1485 |
+
"learning_rate": 2.134792428593971e-06,
|
| 1486 |
+
"logits/chosen": 1.8228479623794556,
|
| 1487 |
+
"logits/rejected": 1.5440882444381714,
|
| 1488 |
+
"logps/chosen": -3.545902967453003,
|
| 1489 |
+
"logps/rejected": -7.393707275390625,
|
| 1490 |
+
"loss": 0.2429,
|
| 1491 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1492 |
+
"rewards/chosen": -5.318853855133057,
|
| 1493 |
+
"rewards/margins": 5.7717061042785645,
|
| 1494 |
+
"rewards/rejected": -11.090560913085938,
|
| 1495 |
+
"step": 820
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.7626048007350408,
|
| 1499 |
+
"grad_norm": 4.266237735748291,
|
| 1500 |
+
"learning_rate": 2.0830781332097446e-06,
|
| 1501 |
+
"logits/chosen": 1.6186097860336304,
|
| 1502 |
+
"logits/rejected": 1.4948759078979492,
|
| 1503 |
+
"logps/chosen": -3.6410841941833496,
|
| 1504 |
+
"logps/rejected": -7.069480895996094,
|
| 1505 |
+
"loss": 0.2213,
|
| 1506 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1507 |
+
"rewards/chosen": -5.461626052856445,
|
| 1508 |
+
"rewards/margins": 5.142596244812012,
|
| 1509 |
+
"rewards/rejected": -10.604223251342773,
|
| 1510 |
+
"step": 830
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 0.771792810382451,
|
| 1514 |
+
"grad_norm": 2.3189163208007812,
|
| 1515 |
+
"learning_rate": 2.031546713535688e-06,
|
| 1516 |
+
"logits/chosen": 2.4443275928497314,
|
| 1517 |
+
"logits/rejected": 2.1987624168395996,
|
| 1518 |
+
"logps/chosen": -4.127468585968018,
|
| 1519 |
+
"logps/rejected": -8.15340805053711,
|
| 1520 |
+
"loss": 0.2525,
|
| 1521 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1522 |
+
"rewards/chosen": -6.191202640533447,
|
| 1523 |
+
"rewards/margins": 6.038909435272217,
|
| 1524 |
+
"rewards/rejected": -12.230112075805664,
|
| 1525 |
+
"step": 840
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.780980820029861,
|
| 1529 |
+
"grad_norm": 5.227460861206055,
|
| 1530 |
+
"learning_rate": 1.9802207729556023e-06,
|
| 1531 |
+
"logits/chosen": 2.0057969093322754,
|
| 1532 |
+
"logits/rejected": 1.7481991052627563,
|
| 1533 |
+
"logps/chosen": -3.890474796295166,
|
| 1534 |
+
"logps/rejected": -7.659049034118652,
|
| 1535 |
+
"loss": 0.2253,
|
| 1536 |
+
"rewards/accuracies": 0.9375,
|
| 1537 |
+
"rewards/chosen": -5.83571195602417,
|
| 1538 |
+
"rewards/margins": 5.65286111831665,
|
| 1539 |
+
"rewards/rejected": -11.48857307434082,
|
| 1540 |
+
"step": 850
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.780980820029861,
|
| 1544 |
+
"eval_logits/chosen": 1.2305774688720703,
|
| 1545 |
+
"eval_logits/rejected": 0.8189592957496643,
|
| 1546 |
+
"eval_logps/chosen": -3.205787420272827,
|
| 1547 |
+
"eval_logps/rejected": -7.230924606323242,
|
| 1548 |
+
"eval_loss": 0.23406219482421875,
|
| 1549 |
+
"eval_rewards/accuracies": 0.9090909361839294,
|
| 1550 |
+
"eval_rewards/chosen": -4.808681488037109,
|
| 1551 |
+
"eval_rewards/margins": 6.0377044677734375,
|
| 1552 |
+
"eval_rewards/rejected": -10.846386909484863,
|
| 1553 |
+
"eval_runtime": 24.5257,
|
| 1554 |
+
"eval_samples_per_second": 28.705,
|
| 1555 |
+
"eval_steps_per_second": 3.588,
|
| 1556 |
+
"step": 850
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.7901688296772712,
|
| 1560 |
+
"grad_norm": 5.743741512298584,
|
| 1561 |
+
"learning_rate": 1.9291228247233607e-06,
|
| 1562 |
+
"logits/chosen": 2.0627987384796143,
|
| 1563 |
+
"logits/rejected": 1.964015245437622,
|
| 1564 |
+
"logps/chosen": -3.810974597930908,
|
| 1565 |
+
"logps/rejected": -7.854369163513184,
|
| 1566 |
+
"loss": 0.254,
|
| 1567 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1568 |
+
"rewards/chosen": -5.716461658477783,
|
| 1569 |
+
"rewards/margins": 6.065093040466309,
|
| 1570 |
+
"rewards/rejected": -11.78155517578125,
|
| 1571 |
+
"step": 860
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.7993568393246813,
|
| 1575 |
+
"grad_norm": 3.256955623626709,
|
| 1576 |
+
"learning_rate": 1.8782752820878636e-06,
|
| 1577 |
+
"logits/chosen": 1.892846703529358,
|
| 1578 |
+
"logits/rejected": 1.5955915451049805,
|
| 1579 |
+
"logps/chosen": -4.009222507476807,
|
| 1580 |
+
"logps/rejected": -7.55028772354126,
|
| 1581 |
+
"loss": 0.2325,
|
| 1582 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1583 |
+
"rewards/chosen": -6.013833045959473,
|
| 1584 |
+
"rewards/margins": 5.3115973472595215,
|
| 1585 |
+
"rewards/rejected": -11.325429916381836,
|
| 1586 |
+
"step": 870
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.8085448489720914,
|
| 1590 |
+
"grad_norm": 3.6304337978363037,
|
| 1591 |
+
"learning_rate": 1.827700448461836e-06,
|
| 1592 |
+
"logits/chosen": 2.3684587478637695,
|
| 1593 |
+
"logits/rejected": 2.1417429447174072,
|
| 1594 |
+
"logps/chosen": -3.8222625255584717,
|
| 1595 |
+
"logps/rejected": -7.484266757965088,
|
| 1596 |
+
"loss": 0.2427,
|
| 1597 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1598 |
+
"rewards/chosen": -5.73339319229126,
|
| 1599 |
+
"rewards/margins": 5.493006229400635,
|
| 1600 |
+
"rewards/rejected": -11.226400375366211,
|
| 1601 |
+
"step": 880
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.8177328586195015,
|
| 1605 |
+
"grad_norm": 4.88602352142334,
|
| 1606 |
+
"learning_rate": 1.7774205076388207e-06,
|
| 1607 |
+
"logits/chosen": 1.8633663654327393,
|
| 1608 |
+
"logits/rejected": 1.6100108623504639,
|
| 1609 |
+
"logps/chosen": -3.4299614429473877,
|
| 1610 |
+
"logps/rejected": -7.589132785797119,
|
| 1611 |
+
"loss": 0.1988,
|
| 1612 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1613 |
+
"rewards/chosen": -5.144942283630371,
|
| 1614 |
+
"rewards/margins": 6.238757133483887,
|
| 1615 |
+
"rewards/rejected": -11.383699417114258,
|
| 1616 |
+
"step": 890
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.8269208682669117,
|
| 1620 |
+
"grad_norm": 3.657270669937134,
|
| 1621 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1622 |
+
"logits/chosen": 1.5545165538787842,
|
| 1623 |
+
"logits/rejected": 1.600629448890686,
|
| 1624 |
+
"logps/chosen": -3.981847047805786,
|
| 1625 |
+
"logps/rejected": -8.365175247192383,
|
| 1626 |
+
"loss": 0.2055,
|
| 1627 |
+
"rewards/accuracies": 0.9375,
|
| 1628 |
+
"rewards/chosen": -5.972771167755127,
|
| 1629 |
+
"rewards/margins": 6.5749921798706055,
|
| 1630 |
+
"rewards/rejected": -12.54776382446289,
|
| 1631 |
+
"step": 900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 0.8269208682669117,
|
| 1635 |
+
"eval_logits/chosen": 1.3828459978103638,
|
| 1636 |
+
"eval_logits/rejected": 0.9709804058074951,
|
| 1637 |
+
"eval_logps/chosen": -3.4689862728118896,
|
| 1638 |
+
"eval_logps/rejected": -7.613832473754883,
|
| 1639 |
+
"eval_loss": 0.2267654538154602,
|
| 1640 |
+
"eval_rewards/accuracies": 0.9090909361839294,
|
| 1641 |
+
"eval_rewards/chosen": -5.203479290008545,
|
| 1642 |
+
"eval_rewards/margins": 6.217269420623779,
|
| 1643 |
+
"eval_rewards/rejected": -11.420748710632324,
|
| 1644 |
+
"eval_runtime": 24.422,
|
| 1645 |
+
"eval_samples_per_second": 28.826,
|
| 1646 |
+
"eval_steps_per_second": 3.603,
|
| 1647 |
+
"step": 900
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.8361088779143218,
|
| 1651 |
+
"grad_norm": 7.899631500244141,
|
| 1652 |
+
"learning_rate": 1.677833383153542e-06,
|
| 1653 |
+
"logits/chosen": 1.8724586963653564,
|
| 1654 |
+
"logits/rejected": 1.4860353469848633,
|
| 1655 |
+
"logps/chosen": -3.703522205352783,
|
| 1656 |
+
"logps/rejected": -7.718357086181641,
|
| 1657 |
+
"loss": 0.2762,
|
| 1658 |
+
"rewards/accuracies": 0.875,
|
| 1659 |
+
"rewards/chosen": -5.5552825927734375,
|
| 1660 |
+
"rewards/margins": 6.02225399017334,
|
| 1661 |
+
"rewards/rejected": -11.577536582946777,
|
| 1662 |
+
"step": 910
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.8452968875617319,
|
| 1666 |
+
"grad_norm": 3.687998056411743,
|
| 1667 |
+
"learning_rate": 1.6285698816954626e-06,
|
| 1668 |
+
"logits/chosen": 3.0009334087371826,
|
| 1669 |
+
"logits/rejected": 2.3279061317443848,
|
| 1670 |
+
"logps/chosen": -4.25749397277832,
|
| 1671 |
+
"logps/rejected": -7.3263139724731445,
|
| 1672 |
+
"loss": 0.2202,
|
| 1673 |
+
"rewards/accuracies": 0.875,
|
| 1674 |
+
"rewards/chosen": -6.3862409591674805,
|
| 1675 |
+
"rewards/margins": 4.603229522705078,
|
| 1676 |
+
"rewards/rejected": -10.989469528198242,
|
| 1677 |
+
"step": 920
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.8544848972091421,
|
| 1681 |
+
"grad_norm": 3.9033992290496826,
|
| 1682 |
+
"learning_rate": 1.5796886182883053e-06,
|
| 1683 |
+
"logits/chosen": 2.3675689697265625,
|
| 1684 |
+
"logits/rejected": 2.2341208457946777,
|
| 1685 |
+
"logps/chosen": -4.182201385498047,
|
| 1686 |
+
"logps/rejected": -7.975285530090332,
|
| 1687 |
+
"loss": 0.1945,
|
| 1688 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1689 |
+
"rewards/chosen": -6.27330207824707,
|
| 1690 |
+
"rewards/margins": 5.6896257400512695,
|
| 1691 |
+
"rewards/rejected": -11.962926864624023,
|
| 1692 |
+
"step": 930
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 0.8636729068565522,
|
| 1696 |
+
"grad_norm": 7.484313488006592,
|
| 1697 |
+
"learning_rate": 1.5312110338697427e-06,
|
| 1698 |
+
"logits/chosen": 1.737408995628357,
|
| 1699 |
+
"logits/rejected": 1.6608823537826538,
|
| 1700 |
+
"logps/chosen": -4.052768707275391,
|
| 1701 |
+
"logps/rejected": -7.796560764312744,
|
| 1702 |
+
"loss": 0.2282,
|
| 1703 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1704 |
+
"rewards/chosen": -6.079153537750244,
|
| 1705 |
+
"rewards/margins": 5.615686893463135,
|
| 1706 |
+
"rewards/rejected": -11.694841384887695,
|
| 1707 |
+
"step": 940
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.8728609165039624,
|
| 1711 |
+
"grad_norm": 5.176153182983398,
|
| 1712 |
+
"learning_rate": 1.4831583923105e-06,
|
| 1713 |
+
"logits/chosen": 1.5573890209197998,
|
| 1714 |
+
"logits/rejected": 1.1255266666412354,
|
| 1715 |
+
"logps/chosen": -3.52099609375,
|
| 1716 |
+
"logps/rejected": -7.2554473876953125,
|
| 1717 |
+
"loss": 0.2434,
|
| 1718 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1719 |
+
"rewards/chosen": -5.281493663787842,
|
| 1720 |
+
"rewards/margins": 5.601677417755127,
|
| 1721 |
+
"rewards/rejected": -10.883171081542969,
|
| 1722 |
+
"step": 950
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.8728609165039624,
|
| 1726 |
+
"eval_logits/chosen": 1.3714543581008911,
|
| 1727 |
+
"eval_logits/rejected": 0.9610424041748047,
|
| 1728 |
+
"eval_logps/chosen": -3.4983131885528564,
|
| 1729 |
+
"eval_logps/rejected": -7.699583053588867,
|
| 1730 |
+
"eval_loss": 0.22121396660804749,
|
| 1731 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 1732 |
+
"eval_rewards/chosen": -5.247470378875732,
|
| 1733 |
+
"eval_rewards/margins": 6.301905632019043,
|
| 1734 |
+
"eval_rewards/rejected": -11.549375534057617,
|
| 1735 |
+
"eval_runtime": 24.4334,
|
| 1736 |
+
"eval_samples_per_second": 28.813,
|
| 1737 |
+
"eval_steps_per_second": 3.602,
|
| 1738 |
+
"step": 950
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.8820489261513724,
|
| 1742 |
+
"grad_norm": 3.092210292816162,
|
| 1743 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 1744 |
+
"logits/chosen": 1.5538525581359863,
|
| 1745 |
+
"logits/rejected": 1.2886962890625,
|
| 1746 |
+
"logps/chosen": -3.7919907569885254,
|
| 1747 |
+
"logps/rejected": -7.742431640625,
|
| 1748 |
+
"loss": 0.2628,
|
| 1749 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1750 |
+
"rewards/chosen": -5.687986373901367,
|
| 1751 |
+
"rewards/margins": 5.925662040710449,
|
| 1752 |
+
"rewards/rejected": -11.613648414611816,
|
| 1753 |
+
"step": 960
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.8912369357987826,
|
| 1757 |
+
"grad_norm": 2.258793830871582,
|
| 1758 |
+
"learning_rate": 1.388412052037682e-06,
|
| 1759 |
+
"logits/chosen": 1.9938386678695679,
|
| 1760 |
+
"logits/rejected": 1.7578411102294922,
|
| 1761 |
+
"logps/chosen": -3.902163028717041,
|
| 1762 |
+
"logps/rejected": -8.263092041015625,
|
| 1763 |
+
"loss": 0.1893,
|
| 1764 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1765 |
+
"rewards/chosen": -5.853244781494141,
|
| 1766 |
+
"rewards/margins": 6.541393280029297,
|
| 1767 |
+
"rewards/rejected": -12.394639015197754,
|
| 1768 |
+
"step": 970
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.9004249454461927,
|
| 1772 |
+
"grad_norm": 3.3003509044647217,
|
| 1773 |
+
"learning_rate": 1.3417599122003464e-06,
|
| 1774 |
+
"logits/chosen": 1.858659029006958,
|
| 1775 |
+
"logits/rejected": 1.7009245157241821,
|
| 1776 |
+
"logps/chosen": -3.995110034942627,
|
| 1777 |
+
"logps/rejected": -7.946316719055176,
|
| 1778 |
+
"loss": 0.2001,
|
| 1779 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1780 |
+
"rewards/chosen": -5.9926652908325195,
|
| 1781 |
+
"rewards/margins": 5.926809310913086,
|
| 1782 |
+
"rewards/rejected": -11.919473648071289,
|
| 1783 |
+
"step": 980
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 0.9096129550936028,
|
| 1787 |
+
"grad_norm": 5.714742183685303,
|
| 1788 |
+
"learning_rate": 1.2956158147457116e-06,
|
| 1789 |
+
"logits/chosen": 2.311112642288208,
|
| 1790 |
+
"logits/rejected": 1.819894790649414,
|
| 1791 |
+
"logps/chosen": -4.011856555938721,
|
| 1792 |
+
"logps/rejected": -8.002031326293945,
|
| 1793 |
+
"loss": 0.216,
|
| 1794 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1795 |
+
"rewards/chosen": -6.01778507232666,
|
| 1796 |
+
"rewards/margins": 5.985262870788574,
|
| 1797 |
+
"rewards/rejected": -12.003047943115234,
|
| 1798 |
+
"step": 990
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 0.918800964741013,
|
| 1802 |
+
"grad_norm": 2.5513203144073486,
|
| 1803 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1804 |
+
"logits/chosen": 2.042844295501709,
|
| 1805 |
+
"logits/rejected": 1.7135541439056396,
|
| 1806 |
+
"logps/chosen": -4.096378803253174,
|
| 1807 |
+
"logps/rejected": -8.864072799682617,
|
| 1808 |
+
"loss": 0.1812,
|
| 1809 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1810 |
+
"rewards/chosen": -6.144567489624023,
|
| 1811 |
+
"rewards/margins": 7.151541709899902,
|
| 1812 |
+
"rewards/rejected": -13.296109199523926,
|
| 1813 |
+
"step": 1000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.918800964741013,
|
| 1817 |
+
"eval_logits/chosen": 1.4344538450241089,
|
| 1818 |
+
"eval_logits/rejected": 1.0740816593170166,
|
| 1819 |
+
"eval_logps/chosen": -3.56337833404541,
|
| 1820 |
+
"eval_logps/rejected": -7.854743957519531,
|
| 1821 |
+
"eval_loss": 0.21498693525791168,
|
| 1822 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 1823 |
+
"eval_rewards/chosen": -5.345067977905273,
|
| 1824 |
+
"eval_rewards/margins": 6.437047004699707,
|
| 1825 |
+
"eval_rewards/rejected": -11.782115936279297,
|
| 1826 |
+
"eval_runtime": 24.4225,
|
| 1827 |
+
"eval_samples_per_second": 28.826,
|
| 1828 |
+
"eval_steps_per_second": 3.603,
|
| 1829 |
+
"step": 1000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.9279889743884231,
|
| 1833 |
+
"grad_norm": 3.213252067565918,
|
| 1834 |
+
"learning_rate": 1.204932476567175e-06,
|
| 1835 |
+
"logits/chosen": 2.064485788345337,
|
| 1836 |
+
"logits/rejected": 1.9654451608657837,
|
| 1837 |
+
"logps/chosen": -3.7592079639434814,
|
| 1838 |
+
"logps/rejected": -8.004045486450195,
|
| 1839 |
+
"loss": 0.2113,
|
| 1840 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1841 |
+
"rewards/chosen": -5.638812065124512,
|
| 1842 |
+
"rewards/margins": 6.367256164550781,
|
| 1843 |
+
"rewards/rejected": -12.00606918334961,
|
| 1844 |
+
"step": 1010
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.9371769840358333,
|
| 1848 |
+
"grad_norm": 3.000185012817383,
|
| 1849 |
+
"learning_rate": 1.160433012552508e-06,
|
| 1850 |
+
"logits/chosen": 1.9040180444717407,
|
| 1851 |
+
"logits/rejected": 1.8388664722442627,
|
| 1852 |
+
"logps/chosen": -4.3835272789001465,
|
| 1853 |
+
"logps/rejected": -8.310102462768555,
|
| 1854 |
+
"loss": 0.2414,
|
| 1855 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1856 |
+
"rewards/chosen": -6.575291633605957,
|
| 1857 |
+
"rewards/margins": 5.889863014221191,
|
| 1858 |
+
"rewards/rejected": -12.465154647827148,
|
| 1859 |
+
"step": 1020
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.9463649936832433,
|
| 1863 |
+
"grad_norm": 2.7824106216430664,
|
| 1864 |
+
"learning_rate": 1.11652112689164e-06,
|
| 1865 |
+
"logits/chosen": 1.4278347492218018,
|
| 1866 |
+
"logits/rejected": 1.4083877801895142,
|
| 1867 |
+
"logps/chosen": -3.6480088233947754,
|
| 1868 |
+
"logps/rejected": -8.00098705291748,
|
| 1869 |
+
"loss": 0.1763,
|
| 1870 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1871 |
+
"rewards/chosen": -5.472012519836426,
|
| 1872 |
+
"rewards/margins": 6.5294671058654785,
|
| 1873 |
+
"rewards/rejected": -12.001480102539062,
|
| 1874 |
+
"step": 1030
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.9555530033306535,
|
| 1878 |
+
"grad_norm": 5.236849784851074,
|
| 1879 |
+
"learning_rate": 1.073216080788921e-06,
|
| 1880 |
+
"logits/chosen": 2.1868062019348145,
|
| 1881 |
+
"logits/rejected": 1.9401241540908813,
|
| 1882 |
+
"logps/chosen": -4.442656993865967,
|
| 1883 |
+
"logps/rejected": -8.857598304748535,
|
| 1884 |
+
"loss": 0.2158,
|
| 1885 |
+
"rewards/accuracies": 0.9375,
|
| 1886 |
+
"rewards/chosen": -6.663985252380371,
|
| 1887 |
+
"rewards/margins": 6.622411251068115,
|
| 1888 |
+
"rewards/rejected": -13.286396980285645,
|
| 1889 |
+
"step": 1040
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.9647410129780636,
|
| 1893 |
+
"grad_norm": 3.4044084548950195,
|
| 1894 |
+
"learning_rate": 1.0305368692688175e-06,
|
| 1895 |
+
"logits/chosen": 1.5827648639678955,
|
| 1896 |
+
"logits/rejected": 1.7356208562850952,
|
| 1897 |
+
"logps/chosen": -3.988180637359619,
|
| 1898 |
+
"logps/rejected": -8.252123832702637,
|
| 1899 |
+
"loss": 0.1967,
|
| 1900 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1901 |
+
"rewards/chosen": -5.982270240783691,
|
| 1902 |
+
"rewards/margins": 6.3959150314331055,
|
| 1903 |
+
"rewards/rejected": -12.378186225891113,
|
| 1904 |
+
"step": 1050
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.9647410129780636,
|
| 1908 |
+
"eval_logits/chosen": 1.4267497062683105,
|
| 1909 |
+
"eval_logits/rejected": 1.0997689962387085,
|
| 1910 |
+
"eval_logps/chosen": -3.4779577255249023,
|
| 1911 |
+
"eval_logps/rejected": -7.853654384613037,
|
| 1912 |
+
"eval_loss": 0.20923514664173126,
|
| 1913 |
+
"eval_rewards/accuracies": 0.9318181872367859,
|
| 1914 |
+
"eval_rewards/chosen": -5.216937065124512,
|
| 1915 |
+
"eval_rewards/margins": 6.563545227050781,
|
| 1916 |
+
"eval_rewards/rejected": -11.780481338500977,
|
| 1917 |
+
"eval_runtime": 24.446,
|
| 1918 |
+
"eval_samples_per_second": 28.798,
|
| 1919 |
+
"eval_steps_per_second": 3.6,
|
| 1920 |
+
"step": 1050
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.9739290226254738,
|
| 1924 |
+
"grad_norm": 7.323760032653809,
|
| 1925 |
+
"learning_rate": 9.88502212844063e-07,
|
| 1926 |
+
"logits/chosen": 1.8404476642608643,
|
| 1927 |
+
"logits/rejected": 1.5995383262634277,
|
| 1928 |
+
"logps/chosen": -3.8108818531036377,
|
| 1929 |
+
"logps/rejected": -8.686467170715332,
|
| 1930 |
+
"loss": 0.2182,
|
| 1931 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1932 |
+
"rewards/chosen": -5.716322898864746,
|
| 1933 |
+
"rewards/margins": 7.313376426696777,
|
| 1934 |
+
"rewards/rejected": -13.029699325561523,
|
| 1935 |
+
"step": 1060
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.9831170322728839,
|
| 1939 |
+
"grad_norm": 2.989016056060791,
|
| 1940 |
+
"learning_rate": 9.471305493042243e-07,
|
| 1941 |
+
"logits/chosen": 2.1675870418548584,
|
| 1942 |
+
"logits/rejected": 2.089310646057129,
|
| 1943 |
+
"logps/chosen": -3.7525386810302734,
|
| 1944 |
+
"logps/rejected": -8.541536331176758,
|
| 1945 |
+
"loss": 0.2056,
|
| 1946 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1947 |
+
"rewards/chosen": -5.628808498382568,
|
| 1948 |
+
"rewards/margins": 7.183494567871094,
|
| 1949 |
+
"rewards/rejected": -12.81230354309082,
|
| 1950 |
+
"step": 1070
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 0.992305041920294,
|
| 1954 |
+
"grad_norm": 5.650535583496094,
|
| 1955 |
+
"learning_rate": 9.064400256282757e-07,
|
| 1956 |
+
"logits/chosen": 2.612128496170044,
|
| 1957 |
+
"logits/rejected": 2.198878765106201,
|
| 1958 |
+
"logps/chosen": -4.110289096832275,
|
| 1959 |
+
"logps/rejected": -8.18535327911377,
|
| 1960 |
+
"loss": 0.1974,
|
| 1961 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1962 |
+
"rewards/chosen": -6.165433406829834,
|
| 1963 |
+
"rewards/margins": 6.11259651184082,
|
| 1964 |
+
"rewards/rejected": -12.278030395507812,
|
| 1965 |
+
"step": 1080
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 1.000918800964741,
|
| 1969 |
+
"grad_norm": 3.090592861175537,
|
| 1970 |
+
"learning_rate": 8.664484900247363e-07,
|
| 1971 |
+
"logits/chosen": 2.4326796531677246,
|
| 1972 |
+
"logits/rejected": 2.0719056129455566,
|
| 1973 |
+
"logps/chosen": -4.295161247253418,
|
| 1974 |
+
"logps/rejected": -8.057024955749512,
|
| 1975 |
+
"loss": 0.1982,
|
| 1976 |
+
"rewards/accuracies": 0.9333333373069763,
|
| 1977 |
+
"rewards/chosen": -6.442741394042969,
|
| 1978 |
+
"rewards/margins": 5.642796039581299,
|
| 1979 |
+
"rewards/rejected": -12.085537910461426,
|
| 1980 |
+
"step": 1090
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 1.0101068106121511,
|
| 1984 |
+
"grad_norm": 3.5762991905212402,
|
| 1985 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1986 |
+
"logits/chosen": 1.808143973350525,
|
| 1987 |
+
"logits/rejected": 1.3669540882110596,
|
| 1988 |
+
"logps/chosen": -4.025475978851318,
|
| 1989 |
+
"logps/rejected": -8.578283309936523,
|
| 1990 |
+
"loss": 0.2476,
|
| 1991 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1992 |
+
"rewards/chosen": -6.038214683532715,
|
| 1993 |
+
"rewards/margins": 6.829211235046387,
|
| 1994 |
+
"rewards/rejected": -12.867425918579102,
|
| 1995 |
+
"step": 1100
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.0101068106121511,
|
| 1999 |
+
"eval_logits/chosen": 1.437761902809143,
|
| 2000 |
+
"eval_logits/rejected": 1.1129838228225708,
|
| 2001 |
+
"eval_logps/chosen": -3.463768720626831,
|
| 2002 |
+
"eval_logps/rejected": -7.876412868499756,
|
| 2003 |
+
"eval_loss": 0.2056656777858734,
|
| 2004 |
+
"eval_rewards/accuracies": 0.9318181872367859,
|
| 2005 |
+
"eval_rewards/chosen": -5.195653915405273,
|
| 2006 |
+
"eval_rewards/margins": 6.6189656257629395,
|
| 2007 |
+
"eval_rewards/rejected": -11.814618110656738,
|
| 2008 |
+
"eval_runtime": 24.4283,
|
| 2009 |
+
"eval_samples_per_second": 28.819,
|
| 2010 |
+
"eval_steps_per_second": 3.602,
|
| 2011 |
+
"step": 1100
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 1.0192948202595613,
|
| 2015 |
+
"grad_norm": 3.994568109512329,
|
| 2016 |
+
"learning_rate": 7.886322351782782e-07,
|
| 2017 |
+
"logits/chosen": 2.0599985122680664,
|
| 2018 |
+
"logits/rejected": 1.7133548259735107,
|
| 2019 |
+
"logps/chosen": -4.200375080108643,
|
| 2020 |
+
"logps/rejected": -8.241968154907227,
|
| 2021 |
+
"loss": 0.1984,
|
| 2022 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2023 |
+
"rewards/chosen": -6.300562381744385,
|
| 2024 |
+
"rewards/margins": 6.062388896942139,
|
| 2025 |
+
"rewards/rejected": -12.36295223236084,
|
| 2026 |
+
"step": 1110
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 1.0284828299069715,
|
| 2030 |
+
"grad_norm": 4.166400909423828,
|
| 2031 |
+
"learning_rate": 7.508416487165862e-07,
|
| 2032 |
+
"logits/chosen": 1.7248255014419556,
|
| 2033 |
+
"logits/rejected": 1.351927399635315,
|
| 2034 |
+
"logps/chosen": -3.682168960571289,
|
| 2035 |
+
"logps/rejected": -8.27522087097168,
|
| 2036 |
+
"loss": 0.186,
|
| 2037 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2038 |
+
"rewards/chosen": -5.523253440856934,
|
| 2039 |
+
"rewards/margins": 6.889578342437744,
|
| 2040 |
+
"rewards/rejected": -12.412832260131836,
|
| 2041 |
+
"step": 1120
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 1.0376708395543814,
|
| 2045 |
+
"grad_norm": 4.695723533630371,
|
| 2046 |
+
"learning_rate": 7.138183009179922e-07,
|
| 2047 |
+
"logits/chosen": 2.1033718585968018,
|
| 2048 |
+
"logits/rejected": 1.9854538440704346,
|
| 2049 |
+
"logps/chosen": -4.065175533294678,
|
| 2050 |
+
"logps/rejected": -8.255699157714844,
|
| 2051 |
+
"loss": 0.1932,
|
| 2052 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2053 |
+
"rewards/chosen": -6.097762584686279,
|
| 2054 |
+
"rewards/margins": 6.285785675048828,
|
| 2055 |
+
"rewards/rejected": -12.38354778289795,
|
| 2056 |
+
"step": 1130
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 1.0468588492017916,
|
| 2060 |
+
"grad_norm": 4.183388710021973,
|
| 2061 |
+
"learning_rate": 6.775784314464717e-07,
|
| 2062 |
+
"logits/chosen": 2.612067699432373,
|
| 2063 |
+
"logits/rejected": 2.1869616508483887,
|
| 2064 |
+
"logps/chosen": -3.6629364490509033,
|
| 2065 |
+
"logps/rejected": -8.047903060913086,
|
| 2066 |
+
"loss": 0.2186,
|
| 2067 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2068 |
+
"rewards/chosen": -5.494403839111328,
|
| 2069 |
+
"rewards/margins": 6.577451229095459,
|
| 2070 |
+
"rewards/rejected": -12.071855545043945,
|
| 2071 |
+
"step": 1140
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 1.0560468588492018,
|
| 2075 |
+
"grad_norm": 5.9020514488220215,
|
| 2076 |
+
"learning_rate": 6.421379363065142e-07,
|
| 2077 |
+
"logits/chosen": 1.2938798666000366,
|
| 2078 |
+
"logits/rejected": 1.1452168226242065,
|
| 2079 |
+
"logps/chosen": -4.144273281097412,
|
| 2080 |
+
"logps/rejected": -8.718961715698242,
|
| 2081 |
+
"loss": 0.2127,
|
| 2082 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2083 |
+
"rewards/chosen": -6.216409206390381,
|
| 2084 |
+
"rewards/margins": 6.862032890319824,
|
| 2085 |
+
"rewards/rejected": -13.078442573547363,
|
| 2086 |
+
"step": 1150
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 1.0560468588492018,
|
| 2090 |
+
"eval_logits/chosen": 1.4493238925933838,
|
| 2091 |
+
"eval_logits/rejected": 1.128269076347351,
|
| 2092 |
+
"eval_logps/chosen": -3.538313150405884,
|
| 2093 |
+
"eval_logps/rejected": -7.989621162414551,
|
| 2094 |
+
"eval_loss": 0.20314906537532806,
|
| 2095 |
+
"eval_rewards/accuracies": 0.9318181872367859,
|
| 2096 |
+
"eval_rewards/chosen": -5.307469367980957,
|
| 2097 |
+
"eval_rewards/margins": 6.676960468292236,
|
| 2098 |
+
"eval_rewards/rejected": -11.984430313110352,
|
| 2099 |
+
"eval_runtime": 24.6248,
|
| 2100 |
+
"eval_samples_per_second": 28.589,
|
| 2101 |
+
"eval_steps_per_second": 3.574,
|
| 2102 |
+
"step": 1150
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 1.065234868496612,
|
| 2106 |
+
"grad_norm": 3.6220788955688477,
|
| 2107 |
+
"learning_rate": 6.075123608706093e-07,
|
| 2108 |
+
"logits/chosen": 1.9196579456329346,
|
| 2109 |
+
"logits/rejected": 1.8552544116973877,
|
| 2110 |
+
"logps/chosen": -3.56549334526062,
|
| 2111 |
+
"logps/rejected": -8.362249374389648,
|
| 2112 |
+
"loss": 0.2277,
|
| 2113 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2114 |
+
"rewards/chosen": -5.348240852355957,
|
| 2115 |
+
"rewards/margins": 7.195133209228516,
|
| 2116 |
+
"rewards/rejected": -12.543375015258789,
|
| 2117 |
+
"step": 1160
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 1.0744228781440222,
|
| 2121 |
+
"grad_norm": 3.9738340377807617,
|
| 2122 |
+
"learning_rate": 5.737168930605272e-07,
|
| 2123 |
+
"logits/chosen": 2.693675994873047,
|
| 2124 |
+
"logits/rejected": 2.596327543258667,
|
| 2125 |
+
"logps/chosen": -3.953673839569092,
|
| 2126 |
+
"logps/rejected": -8.083026885986328,
|
| 2127 |
+
"loss": 0.2156,
|
| 2128 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2129 |
+
"rewards/chosen": -5.930511474609375,
|
| 2130 |
+
"rewards/margins": 6.194031238555908,
|
| 2131 |
+
"rewards/rejected": -12.124540328979492,
|
| 2132 |
+
"step": 1170
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 1.083610887791432,
|
| 2136 |
+
"grad_norm": 3.9082953929901123,
|
| 2137 |
+
"learning_rate": 5.407663566854008e-07,
|
| 2138 |
+
"logits/chosen": 2.457482099533081,
|
| 2139 |
+
"logits/rejected": 2.206406354904175,
|
| 2140 |
+
"logps/chosen": -4.224593162536621,
|
| 2141 |
+
"logps/rejected": -8.104263305664062,
|
| 2142 |
+
"loss": 0.1883,
|
| 2143 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2144 |
+
"rewards/chosen": -6.336889266967773,
|
| 2145 |
+
"rewards/margins": 5.8195061683654785,
|
| 2146 |
+
"rewards/rejected": -12.15639591217041,
|
| 2147 |
+
"step": 1180
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 1.0927988974388423,
|
| 2151 |
+
"grad_norm": 2.5964362621307373,
|
| 2152 |
+
"learning_rate": 5.086752049395094e-07,
|
| 2153 |
+
"logits/chosen": 1.9967896938323975,
|
| 2154 |
+
"logits/rejected": 1.618037462234497,
|
| 2155 |
+
"logps/chosen": -3.5683434009552,
|
| 2156 |
+
"logps/rejected": -7.307412624359131,
|
| 2157 |
+
"loss": 0.1798,
|
| 2158 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2159 |
+
"rewards/chosen": -5.352514743804932,
|
| 2160 |
+
"rewards/margins": 5.608603477478027,
|
| 2161 |
+
"rewards/rejected": -10.961118698120117,
|
| 2162 |
+
"step": 1190
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 1.1019869070862525,
|
| 2166 |
+
"grad_norm": 2.499810218811035,
|
| 2167 |
+
"learning_rate": 4.774575140626317e-07,
|
| 2168 |
+
"logits/chosen": 1.7958850860595703,
|
| 2169 |
+
"logits/rejected": 1.480818748474121,
|
| 2170 |
+
"logps/chosen": -3.8547580242156982,
|
| 2171 |
+
"logps/rejected": -8.144584655761719,
|
| 2172 |
+
"loss": 0.2005,
|
| 2173 |
+
"rewards/accuracies": 0.9375,
|
| 2174 |
+
"rewards/chosen": -5.7821364402771,
|
| 2175 |
+
"rewards/margins": 6.4347405433654785,
|
| 2176 |
+
"rewards/rejected": -12.216877937316895,
|
| 2177 |
+
"step": 1200
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 1.1019869070862525,
|
| 2181 |
+
"eval_logits/chosen": 1.4727997779846191,
|
| 2182 |
+
"eval_logits/rejected": 1.1707344055175781,
|
| 2183 |
+
"eval_logps/chosen": -3.532540798187256,
|
| 2184 |
+
"eval_logps/rejected": -8.002986907958984,
|
| 2185 |
+
"eval_loss": 0.2008148729801178,
|
| 2186 |
+
"eval_rewards/accuracies": 0.9204545617103577,
|
| 2187 |
+
"eval_rewards/chosen": -5.298811435699463,
|
| 2188 |
+
"eval_rewards/margins": 6.705668926239014,
|
| 2189 |
+
"eval_rewards/rejected": -12.004480361938477,
|
| 2190 |
+
"eval_runtime": 24.695,
|
| 2191 |
+
"eval_samples_per_second": 28.508,
|
| 2192 |
+
"eval_steps_per_second": 3.563,
|
| 2193 |
+
"step": 1200
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.1111749167336626,
|
| 2197 |
+
"grad_norm": 3.8940417766571045,
|
| 2198 |
+
"learning_rate": 4.4712697716573994e-07,
|
| 2199 |
+
"logits/chosen": 2.203734874725342,
|
| 2200 |
+
"logits/rejected": 2.0729122161865234,
|
| 2201 |
+
"logps/chosen": -4.096261501312256,
|
| 2202 |
+
"logps/rejected": -7.813987731933594,
|
| 2203 |
+
"loss": 0.1796,
|
| 2204 |
+
"rewards/accuracies": 0.9375,
|
| 2205 |
+
"rewards/chosen": -6.144392490386963,
|
| 2206 |
+
"rewards/margins": 5.5765886306762695,
|
| 2207 |
+
"rewards/rejected": -11.720980644226074,
|
| 2208 |
+
"step": 1210
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 1.1203629263810726,
|
| 2212 |
+
"grad_norm": 4.197968006134033,
|
| 2213 |
+
"learning_rate": 4.1769689822475147e-07,
|
| 2214 |
+
"logits/chosen": 1.5070486068725586,
|
| 2215 |
+
"logits/rejected": 1.5961544513702393,
|
| 2216 |
+
"logps/chosen": -3.433948040008545,
|
| 2217 |
+
"logps/rejected": -8.33233642578125,
|
| 2218 |
+
"loss": 0.1742,
|
| 2219 |
+
"rewards/accuracies": 0.9375,
|
| 2220 |
+
"rewards/chosen": -5.1509222984313965,
|
| 2221 |
+
"rewards/margins": 7.3475823402404785,
|
| 2222 |
+
"rewards/rejected": -12.498504638671875,
|
| 2223 |
+
"step": 1220
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 1.1295509360284828,
|
| 2227 |
+
"grad_norm": 3.927893877029419,
|
| 2228 |
+
"learning_rate": 3.891801862449629e-07,
|
| 2229 |
+
"logits/chosen": 2.136690616607666,
|
| 2230 |
+
"logits/rejected": 2.030285120010376,
|
| 2231 |
+
"logps/chosen": -4.768858909606934,
|
| 2232 |
+
"logps/rejected": -8.714016914367676,
|
| 2233 |
+
"loss": 0.1811,
|
| 2234 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2235 |
+
"rewards/chosen": -7.153287410736084,
|
| 2236 |
+
"rewards/margins": 5.917738914489746,
|
| 2237 |
+
"rewards/rejected": -13.071026802062988,
|
| 2238 |
+
"step": 1230
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 1.138738945675893,
|
| 2242 |
+
"grad_norm": 3.148071765899658,
|
| 2243 |
+
"learning_rate": 3.615893495987335e-07,
|
| 2244 |
+
"logits/chosen": 2.084843873977661,
|
| 2245 |
+
"logits/rejected": 2.249072551727295,
|
| 2246 |
+
"logps/chosen": -4.252139091491699,
|
| 2247 |
+
"logps/rejected": -8.39255428314209,
|
| 2248 |
+
"loss": 0.1942,
|
| 2249 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2250 |
+
"rewards/chosen": -6.378207683563232,
|
| 2251 |
+
"rewards/margins": 6.210625648498535,
|
| 2252 |
+
"rewards/rejected": -12.588833808898926,
|
| 2253 |
+
"step": 1240
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 1.1479269553233031,
|
| 2257 |
+
"grad_norm": 1.6102039813995361,
|
| 2258 |
+
"learning_rate": 3.3493649053890325e-07,
|
| 2259 |
+
"logits/chosen": 2.5606529712677,
|
| 2260 |
+
"logits/rejected": 2.4942524433135986,
|
| 2261 |
+
"logps/chosen": -4.308623313903809,
|
| 2262 |
+
"logps/rejected": -8.170819282531738,
|
| 2263 |
+
"loss": 0.2088,
|
| 2264 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 2265 |
+
"rewards/chosen": -6.4629340171813965,
|
| 2266 |
+
"rewards/margins": 5.793294429779053,
|
| 2267 |
+
"rewards/rejected": -12.256229400634766,
|
| 2268 |
+
"step": 1250
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 1.1479269553233031,
|
| 2272 |
+
"eval_logits/chosen": 1.4887795448303223,
|
| 2273 |
+
"eval_logits/rejected": 1.181876540184021,
|
| 2274 |
+
"eval_logps/chosen": -3.5690500736236572,
|
| 2275 |
+
"eval_logps/rejected": -8.063680648803711,
|
| 2276 |
+
"eval_loss": 0.20090900361537933,
|
| 2277 |
+
"eval_rewards/accuracies": 0.9318181872367859,
|
| 2278 |
+
"eval_rewards/chosen": -5.353575229644775,
|
| 2279 |
+
"eval_rewards/margins": 6.741945266723633,
|
| 2280 |
+
"eval_rewards/rejected": -12.09552001953125,
|
| 2281 |
+
"eval_runtime": 24.433,
|
| 2282 |
+
"eval_samples_per_second": 28.813,
|
| 2283 |
+
"eval_steps_per_second": 3.602,
|
| 2284 |
+
"step": 1250
|
| 2285 |
+
}
|
| 2286 |
+
],
|
| 2287 |
+
"logging_steps": 10,
|
| 2288 |
+
"max_steps": 1500,
|
| 2289 |
+
"num_input_tokens_seen": 0,
|
| 2290 |
+
"num_train_epochs": 2,
|
| 2291 |
+
"save_steps": 50,
|
| 2292 |
+
"stateful_callbacks": {
|
| 2293 |
+
"TrainerControl": {
|
| 2294 |
+
"args": {
|
| 2295 |
+
"should_epoch_stop": false,
|
| 2296 |
+
"should_evaluate": false,
|
| 2297 |
+
"should_log": false,
|
| 2298 |
+
"should_save": true,
|
| 2299 |
+
"should_training_stop": false
|
| 2300 |
+
},
|
| 2301 |
+
"attributes": {}
|
| 2302 |
+
}
|
| 2303 |
+
},
|
| 2304 |
+
"total_flos": 3.0347258909373235e+18,
|
| 2305 |
+
"train_batch_size": 1,
|
| 2306 |
+
"trial_name": null,
|
| 2307 |
+
"trial_params": null
|
| 2308 |
+
}
|
checkpoint-1250/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35e758dd0e4a754b8adb96ec29d9d5e6d185c2cfb0f0a27add29180a481d3863
|
| 3 |
+
size 7224
|
checkpoint-1250/zero_to_fp32.py
ADDED
|
@@ -0,0 +1,674 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python
|
| 2 |
+
|
| 3 |
+
# Copyright (c) Microsoft Corporation.
|
| 4 |
+
# SPDX-License-Identifier: Apache-2.0
|
| 5 |
+
|
| 6 |
+
# DeepSpeed Team
|
| 7 |
+
|
| 8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
| 9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
| 10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
| 11 |
+
# application.
|
| 12 |
+
#
|
| 13 |
+
# example:
|
| 14 |
+
# python zero_to_fp32.py . output_dir/
|
| 15 |
+
# or
|
| 16 |
+
# python zero_to_fp32.py . output_dir/ --safe_serialization
|
| 17 |
+
|
| 18 |
+
import argparse
|
| 19 |
+
import torch
|
| 20 |
+
import glob
|
| 21 |
+
import math
|
| 22 |
+
import os
|
| 23 |
+
import re
|
| 24 |
+
import json
|
| 25 |
+
from tqdm import tqdm
|
| 26 |
+
from collections import OrderedDict
|
| 27 |
+
from dataclasses import dataclass
|
| 28 |
+
|
| 29 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 30 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
| 31 |
+
from deepspeed.utils import logger
|
| 32 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 33 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 34 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
@dataclass
|
| 38 |
+
class zero_model_state:
|
| 39 |
+
buffers: dict()
|
| 40 |
+
param_shapes: dict()
|
| 41 |
+
shared_params: list
|
| 42 |
+
ds_version: int
|
| 43 |
+
frozen_param_shapes: dict()
|
| 44 |
+
frozen_param_fragments: dict()
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
debug = 0
|
| 48 |
+
|
| 49 |
+
# load to cpu
|
| 50 |
+
device = torch.device('cpu')
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
def atoi(text):
|
| 54 |
+
return int(text) if text.isdigit() else text
|
| 55 |
+
|
| 56 |
+
|
| 57 |
+
def natural_keys(text):
|
| 58 |
+
'''
|
| 59 |
+
alist.sort(key=natural_keys) sorts in human order
|
| 60 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 61 |
+
(See Toothy's implementation in the comments)
|
| 62 |
+
'''
|
| 63 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 67 |
+
if not os.path.isdir(checkpoint_dir):
|
| 68 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 69 |
+
|
| 70 |
+
# there should be only one file
|
| 71 |
+
if zero_stage <= 2:
|
| 72 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 73 |
+
elif zero_stage == 3:
|
| 74 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 75 |
+
|
| 76 |
+
if not os.path.exists(file):
|
| 77 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 78 |
+
|
| 79 |
+
return file
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 83 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 84 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 85 |
+
|
| 86 |
+
if len(ckpt_files) == 0:
|
| 87 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 88 |
+
|
| 89 |
+
return ckpt_files
|
| 90 |
+
|
| 91 |
+
|
| 92 |
+
def get_optim_files(checkpoint_dir):
|
| 93 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
def get_model_state_files(checkpoint_dir):
|
| 97 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 98 |
+
|
| 99 |
+
|
| 100 |
+
def parse_model_states(files):
|
| 101 |
+
zero_model_states = []
|
| 102 |
+
for file in files:
|
| 103 |
+
state_dict = torch.load(file, map_location=device)
|
| 104 |
+
|
| 105 |
+
if BUFFER_NAMES not in state_dict:
|
| 106 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
| 107 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
| 108 |
+
if debug:
|
| 109 |
+
print("Found buffers:", buffer_names)
|
| 110 |
+
|
| 111 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 112 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 113 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
| 114 |
+
|
| 115 |
+
# collect parameters that are included in param_shapes
|
| 116 |
+
param_names = []
|
| 117 |
+
for s in param_shapes:
|
| 118 |
+
for name in s.keys():
|
| 119 |
+
param_names.append(name)
|
| 120 |
+
|
| 121 |
+
# update with frozen parameters
|
| 122 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 123 |
+
if frozen_param_shapes is not None:
|
| 124 |
+
if debug:
|
| 125 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 126 |
+
param_names += list(frozen_param_shapes.keys())
|
| 127 |
+
|
| 128 |
+
# handle shared params
|
| 129 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 130 |
+
|
| 131 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
| 132 |
+
|
| 133 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 134 |
+
|
| 135 |
+
z_model_state = zero_model_state(buffers=buffers,
|
| 136 |
+
param_shapes=param_shapes,
|
| 137 |
+
shared_params=shared_params,
|
| 138 |
+
ds_version=ds_version,
|
| 139 |
+
frozen_param_shapes=frozen_param_shapes,
|
| 140 |
+
frozen_param_fragments=frozen_param_fragments)
|
| 141 |
+
zero_model_states.append(z_model_state)
|
| 142 |
+
|
| 143 |
+
return zero_model_states
|
| 144 |
+
|
| 145 |
+
|
| 146 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
| 147 |
+
total_files = len(files)
|
| 148 |
+
state_dicts = []
|
| 149 |
+
for f in files:
|
| 150 |
+
state_dict = torch.load(f, map_location=device)
|
| 151 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 152 |
+
# and also handle the case where it was already removed by another helper script
|
| 153 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 154 |
+
state_dicts.append(state_dict)
|
| 155 |
+
|
| 156 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 157 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 158 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 159 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 160 |
+
|
| 161 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 162 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 163 |
+
# use the max of the partition_count to get the dp world_size.
|
| 164 |
+
|
| 165 |
+
if type(world_size) is list:
|
| 166 |
+
world_size = max(world_size)
|
| 167 |
+
|
| 168 |
+
if world_size != total_files:
|
| 169 |
+
raise ValueError(
|
| 170 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 171 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 172 |
+
)
|
| 173 |
+
|
| 174 |
+
# the groups are named differently in each stage
|
| 175 |
+
if zero_stage <= 2:
|
| 176 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 177 |
+
elif zero_stage == 3:
|
| 178 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
| 179 |
+
else:
|
| 180 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 181 |
+
|
| 182 |
+
if zero_stage <= 2:
|
| 183 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 184 |
+
elif zero_stage == 3:
|
| 185 |
+
# if there is more than one param group, there will be multiple flattened tensors - one
|
| 186 |
+
# flattened tensor per group - for simplicity merge them into a single tensor
|
| 187 |
+
#
|
| 188 |
+
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
| 189 |
+
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
| 190 |
+
|
| 191 |
+
fp32_flat_groups = [
|
| 192 |
+
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
| 193 |
+
]
|
| 194 |
+
|
| 195 |
+
return zero_stage, world_size, fp32_flat_groups
|
| 196 |
+
|
| 197 |
+
|
| 198 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
|
| 199 |
+
"""
|
| 200 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 201 |
+
|
| 202 |
+
Args:
|
| 203 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 204 |
+
|
| 205 |
+
"""
|
| 206 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 207 |
+
|
| 208 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 209 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 210 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 211 |
+
|
| 212 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 213 |
+
|
| 214 |
+
zero_model_states = parse_model_states(model_files)
|
| 215 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 216 |
+
|
| 217 |
+
if zero_stage <= 2:
|
| 218 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 219 |
+
exclude_frozen_parameters)
|
| 220 |
+
elif zero_stage == 3:
|
| 221 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 222 |
+
exclude_frozen_parameters)
|
| 223 |
+
|
| 224 |
+
|
| 225 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 226 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 227 |
+
return
|
| 228 |
+
|
| 229 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 230 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 231 |
+
|
| 232 |
+
if debug:
|
| 233 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 234 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 235 |
+
|
| 236 |
+
wanted_params = len(frozen_param_shapes)
|
| 237 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 238 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 239 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 240 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 241 |
+
|
| 242 |
+
total_params = 0
|
| 243 |
+
total_numel = 0
|
| 244 |
+
for name, shape in frozen_param_shapes.items():
|
| 245 |
+
total_params += 1
|
| 246 |
+
unpartitioned_numel = shape.numel()
|
| 247 |
+
total_numel += unpartitioned_numel
|
| 248 |
+
|
| 249 |
+
state_dict[name] = frozen_param_fragments[name]
|
| 250 |
+
|
| 251 |
+
if debug:
|
| 252 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 253 |
+
|
| 254 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 255 |
+
|
| 256 |
+
|
| 257 |
+
def _has_callable(obj, fn):
|
| 258 |
+
attr = getattr(obj, fn, None)
|
| 259 |
+
return callable(attr)
|
| 260 |
+
|
| 261 |
+
|
| 262 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 263 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 264 |
+
|
| 265 |
+
# Reconstruction protocol:
|
| 266 |
+
#
|
| 267 |
+
# XXX: document this
|
| 268 |
+
|
| 269 |
+
if debug:
|
| 270 |
+
for i in range(world_size):
|
| 271 |
+
for j in range(len(fp32_flat_groups[0])):
|
| 272 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 273 |
+
|
| 274 |
+
# XXX: memory usage doubles here (zero2)
|
| 275 |
+
num_param_groups = len(fp32_flat_groups[0])
|
| 276 |
+
merged_single_partition_of_fp32_groups = []
|
| 277 |
+
for i in range(num_param_groups):
|
| 278 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 279 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 280 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 281 |
+
avail_numel = sum(
|
| 282 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 283 |
+
|
| 284 |
+
if debug:
|
| 285 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 286 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 287 |
+
# not asserting if there is a mismatch due to possible padding
|
| 288 |
+
print(f"Have {avail_numel} numels to process.")
|
| 289 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 290 |
+
|
| 291 |
+
# params
|
| 292 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 293 |
+
# out-of-core computing solution
|
| 294 |
+
total_numel = 0
|
| 295 |
+
total_params = 0
|
| 296 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 297 |
+
offset = 0
|
| 298 |
+
avail_numel = full_single_fp32_vector.numel()
|
| 299 |
+
for name, shape in shapes.items():
|
| 300 |
+
|
| 301 |
+
unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
|
| 302 |
+
total_numel += unpartitioned_numel
|
| 303 |
+
total_params += 1
|
| 304 |
+
|
| 305 |
+
if debug:
|
| 306 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 307 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 308 |
+
offset += unpartitioned_numel
|
| 309 |
+
|
| 310 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 311 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 312 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 313 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
| 314 |
+
align_to = 2 * world_size
|
| 315 |
+
|
| 316 |
+
def zero2_align(x):
|
| 317 |
+
return align_to * math.ceil(x / align_to)
|
| 318 |
+
|
| 319 |
+
if debug:
|
| 320 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 321 |
+
|
| 322 |
+
offset = zero2_align(offset)
|
| 323 |
+
avail_numel = zero2_align(avail_numel)
|
| 324 |
+
|
| 325 |
+
if debug:
|
| 326 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 327 |
+
|
| 328 |
+
# Sanity check
|
| 329 |
+
if offset != avail_numel:
|
| 330 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 331 |
+
|
| 332 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 333 |
+
|
| 334 |
+
|
| 335 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 336 |
+
exclude_frozen_parameters):
|
| 337 |
+
state_dict = OrderedDict()
|
| 338 |
+
|
| 339 |
+
# buffers
|
| 340 |
+
buffers = zero_model_states[0].buffers
|
| 341 |
+
state_dict.update(buffers)
|
| 342 |
+
if debug:
|
| 343 |
+
print(f"added {len(buffers)} buffers")
|
| 344 |
+
|
| 345 |
+
if not exclude_frozen_parameters:
|
| 346 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 347 |
+
|
| 348 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 349 |
+
|
| 350 |
+
# recover shared parameters
|
| 351 |
+
for pair in zero_model_states[0].shared_params:
|
| 352 |
+
if pair[1] in state_dict:
|
| 353 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 354 |
+
|
| 355 |
+
return state_dict
|
| 356 |
+
|
| 357 |
+
|
| 358 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 359 |
+
remainder = unpartitioned_numel % world_size
|
| 360 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
| 361 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 362 |
+
return partitioned_numel, padding_numel
|
| 363 |
+
|
| 364 |
+
|
| 365 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 366 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 367 |
+
return
|
| 368 |
+
|
| 369 |
+
if debug:
|
| 370 |
+
for i in range(world_size):
|
| 371 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 372 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 373 |
+
|
| 374 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 375 |
+
wanted_params = len(frozen_param_shapes)
|
| 376 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 377 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 378 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 379 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 380 |
+
|
| 381 |
+
total_params = 0
|
| 382 |
+
total_numel = 0
|
| 383 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 384 |
+
total_params += 1
|
| 385 |
+
unpartitioned_numel = shape.numel()
|
| 386 |
+
total_numel += unpartitioned_numel
|
| 387 |
+
|
| 388 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 389 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 390 |
+
|
| 391 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 392 |
+
|
| 393 |
+
if debug:
|
| 394 |
+
print(
|
| 395 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 396 |
+
)
|
| 397 |
+
|
| 398 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 399 |
+
|
| 400 |
+
|
| 401 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 402 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 403 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 404 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 405 |
+
# param, re-consolidating each param, while dealing with padding if any
|
| 406 |
+
|
| 407 |
+
# merge list of dicts, preserving order
|
| 408 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 409 |
+
|
| 410 |
+
if debug:
|
| 411 |
+
for i in range(world_size):
|
| 412 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 413 |
+
|
| 414 |
+
wanted_params = len(param_shapes)
|
| 415 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 416 |
+
# not asserting if there is a mismatch due to possible padding
|
| 417 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 418 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 419 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 420 |
+
|
| 421 |
+
# params
|
| 422 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 423 |
+
# out-of-core computing solution
|
| 424 |
+
offset = 0
|
| 425 |
+
total_numel = 0
|
| 426 |
+
total_params = 0
|
| 427 |
+
for name, shape in tqdm(param_shapes.items(), desc='Gathering Sharded Weights'):
|
| 428 |
+
unpartitioned_numel = shape.numel()
|
| 429 |
+
total_numel += unpartitioned_numel
|
| 430 |
+
total_params += 1
|
| 431 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 432 |
+
|
| 433 |
+
if debug:
|
| 434 |
+
print(
|
| 435 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 436 |
+
)
|
| 437 |
+
|
| 438 |
+
# XXX: memory usage doubles here
|
| 439 |
+
state_dict[name] = torch.cat(
|
| 440 |
+
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
| 441 |
+
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 442 |
+
offset += partitioned_numel
|
| 443 |
+
|
| 444 |
+
offset *= world_size
|
| 445 |
+
|
| 446 |
+
# Sanity check
|
| 447 |
+
if offset != avail_numel:
|
| 448 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 449 |
+
|
| 450 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 451 |
+
|
| 452 |
+
|
| 453 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 454 |
+
exclude_frozen_parameters):
|
| 455 |
+
state_dict = OrderedDict()
|
| 456 |
+
|
| 457 |
+
# buffers
|
| 458 |
+
buffers = zero_model_states[0].buffers
|
| 459 |
+
state_dict.update(buffers)
|
| 460 |
+
if debug:
|
| 461 |
+
print(f"added {len(buffers)} buffers")
|
| 462 |
+
|
| 463 |
+
if not exclude_frozen_parameters:
|
| 464 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 465 |
+
|
| 466 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 467 |
+
|
| 468 |
+
# recover shared parameters
|
| 469 |
+
for pair in zero_model_states[0].shared_params:
|
| 470 |
+
if pair[1] in state_dict:
|
| 471 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 472 |
+
|
| 473 |
+
return state_dict
|
| 474 |
+
|
| 475 |
+
|
| 476 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
|
| 477 |
+
"""
|
| 478 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 479 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 480 |
+
via a model hub.
|
| 481 |
+
|
| 482 |
+
Args:
|
| 483 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 484 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
|
| 485 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 486 |
+
|
| 487 |
+
Returns:
|
| 488 |
+
- pytorch ``state_dict``
|
| 489 |
+
|
| 490 |
+
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
| 491 |
+
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 492 |
+
the checkpoint.
|
| 493 |
+
|
| 494 |
+
A typical usage might be ::
|
| 495 |
+
|
| 496 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 497 |
+
# do the training and checkpoint saving
|
| 498 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 499 |
+
model = model.cpu() # move to cpu
|
| 500 |
+
model.load_state_dict(state_dict)
|
| 501 |
+
# submit to model hub or save the model to share with others
|
| 502 |
+
|
| 503 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 504 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 505 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 506 |
+
|
| 507 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 508 |
+
|
| 509 |
+
"""
|
| 510 |
+
if tag is None:
|
| 511 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 512 |
+
if os.path.isfile(latest_path):
|
| 513 |
+
with open(latest_path, 'r') as fd:
|
| 514 |
+
tag = fd.read().strip()
|
| 515 |
+
else:
|
| 516 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 517 |
+
|
| 518 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 519 |
+
|
| 520 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
| 521 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 522 |
+
|
| 523 |
+
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
|
| 524 |
+
|
| 525 |
+
|
| 526 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
|
| 527 |
+
output_dir,
|
| 528 |
+
max_shard_size="5GB",
|
| 529 |
+
safe_serialization=False,
|
| 530 |
+
tag=None,
|
| 531 |
+
exclude_frozen_parameters=False):
|
| 532 |
+
"""
|
| 533 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 534 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 535 |
+
|
| 536 |
+
Args:
|
| 537 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 538 |
+
- ``output_dir``: directory to the pytorch fp32 state_dict output files
|
| 539 |
+
- ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
|
| 540 |
+
- ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
|
| 541 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 542 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 543 |
+
"""
|
| 544 |
+
# Dependency pre-check
|
| 545 |
+
if safe_serialization:
|
| 546 |
+
try:
|
| 547 |
+
from safetensors.torch import save_file
|
| 548 |
+
except ImportError:
|
| 549 |
+
print('If you want to use `safe_serialization`, please `pip install safetensors`')
|
| 550 |
+
raise
|
| 551 |
+
if max_shard_size is not None:
|
| 552 |
+
try:
|
| 553 |
+
from huggingface_hub import split_torch_state_dict_into_shards
|
| 554 |
+
except ImportError:
|
| 555 |
+
print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
|
| 556 |
+
raise
|
| 557 |
+
|
| 558 |
+
# Convert zero checkpoint to state_dict
|
| 559 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
|
| 560 |
+
|
| 561 |
+
# Shard the model if it is too big.
|
| 562 |
+
weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
|
| 563 |
+
if max_shard_size is not None:
|
| 564 |
+
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
|
| 565 |
+
state_dict_split = split_torch_state_dict_into_shards(state_dict,
|
| 566 |
+
filename_pattern=filename_pattern,
|
| 567 |
+
max_shard_size=max_shard_size)
|
| 568 |
+
else:
|
| 569 |
+
from collections import namedtuple
|
| 570 |
+
StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
|
| 571 |
+
state_dict_split = StateDictSplit(is_sharded=False,
|
| 572 |
+
filename_to_tensors={weights_name: list(state_dict.keys())})
|
| 573 |
+
|
| 574 |
+
# Save the model
|
| 575 |
+
filename_to_tensors = state_dict_split.filename_to_tensors.items()
|
| 576 |
+
for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
|
| 577 |
+
shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
|
| 578 |
+
output_path = os.path.join(output_dir, shard_file)
|
| 579 |
+
if safe_serialization:
|
| 580 |
+
save_file(shard, output_path, metadata={"format": "pt"})
|
| 581 |
+
else:
|
| 582 |
+
torch.save(shard, output_path)
|
| 583 |
+
|
| 584 |
+
# Save index if sharded
|
| 585 |
+
if state_dict_split.is_sharded:
|
| 586 |
+
index = {
|
| 587 |
+
"metadata": state_dict_split.metadata,
|
| 588 |
+
"weight_map": state_dict_split.tensor_to_filename,
|
| 589 |
+
}
|
| 590 |
+
save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
|
| 591 |
+
save_index_file = os.path.join(output_dir, save_index_file)
|
| 592 |
+
with open(save_index_file, "w", encoding="utf-8") as f:
|
| 593 |
+
content = json.dumps(index, indent=2, sort_keys=True) + "\n"
|
| 594 |
+
f.write(content)
|
| 595 |
+
|
| 596 |
+
|
| 597 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 598 |
+
"""
|
| 599 |
+
1. Put the provided model to cpu
|
| 600 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 601 |
+
3. Load it into the provided model
|
| 602 |
+
|
| 603 |
+
Args:
|
| 604 |
+
- ``model``: the model object to update
|
| 605 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 606 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 607 |
+
|
| 608 |
+
Returns:
|
| 609 |
+
- ``model`: modified model
|
| 610 |
+
|
| 611 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 612 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 613 |
+
conveniently placed for you in the checkpoint folder.
|
| 614 |
+
|
| 615 |
+
A typical usage might be ::
|
| 616 |
+
|
| 617 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 618 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 619 |
+
# submit to model hub or save the model to share with others
|
| 620 |
+
|
| 621 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 622 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 623 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 624 |
+
|
| 625 |
+
"""
|
| 626 |
+
logger.info(f"Extracting fp32 weights")
|
| 627 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 628 |
+
|
| 629 |
+
logger.info(f"Overwriting model with fp32 weights")
|
| 630 |
+
model = model.cpu()
|
| 631 |
+
model.load_state_dict(state_dict, strict=False)
|
| 632 |
+
|
| 633 |
+
return model
|
| 634 |
+
|
| 635 |
+
|
| 636 |
+
if __name__ == "__main__":
|
| 637 |
+
parser = argparse.ArgumentParser()
|
| 638 |
+
parser.add_argument("checkpoint_dir",
|
| 639 |
+
type=str,
|
| 640 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 641 |
+
parser.add_argument("output_dir",
|
| 642 |
+
type=str,
|
| 643 |
+
help="directory to the pytorch fp32 state_dict output files"
|
| 644 |
+
"(e.g. path/checkpoint-12-output/)")
|
| 645 |
+
parser.add_argument(
|
| 646 |
+
"--max_shard_size",
|
| 647 |
+
type=str,
|
| 648 |
+
default="5GB",
|
| 649 |
+
help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
|
| 650 |
+
"lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
|
| 651 |
+
"We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
|
| 652 |
+
"without CPU OOM issues.")
|
| 653 |
+
parser.add_argument(
|
| 654 |
+
"--safe_serialization",
|
| 655 |
+
default=False,
|
| 656 |
+
action='store_true',
|
| 657 |
+
help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
|
| 658 |
+
parser.add_argument("-t",
|
| 659 |
+
"--tag",
|
| 660 |
+
type=str,
|
| 661 |
+
default=None,
|
| 662 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 663 |
+
parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
|
| 664 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 665 |
+
args = parser.parse_args()
|
| 666 |
+
|
| 667 |
+
debug = args.debug
|
| 668 |
+
|
| 669 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
|
| 670 |
+
args.output_dir,
|
| 671 |
+
max_shard_size=args.max_shard_size,
|
| 672 |
+
safe_serialization=args.safe_serialization,
|
| 673 |
+
tag=args.tag,
|
| 674 |
+
exclude_frozen_parameters=args.exclude_frozen_parameters)
|