Training in progress, step 1200, checkpoint
Browse files- checkpoint-1200/README.md +202 -0
- checkpoint-1200/adapter_config.json +37 -0
- checkpoint-1200/adapter_model.safetensors +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1200/global_step1200/mp_rank_00_model_states.pt +3 -0
- checkpoint-1200/latest +1 -0
- checkpoint-1200/rng_state_0.pth +3 -0
- checkpoint-1200/rng_state_1.pth +3 -0
- checkpoint-1200/rng_state_2.pth +3 -0
- checkpoint-1200/rng_state_3.pth +3 -0
- checkpoint-1200/rng_state_4.pth +3 -0
- checkpoint-1200/rng_state_5.pth +3 -0
- checkpoint-1200/rng_state_6.pth +3 -0
- checkpoint-1200/rng_state_7.pth +3 -0
- checkpoint-1200/scheduler.pt +3 -0
- checkpoint-1200/special_tokens_map.json +30 -0
- checkpoint-1200/tokenizer.json +0 -0
- checkpoint-1200/tokenizer_config.json +87 -0
- checkpoint-1200/trainer_state.json +2217 -0
- checkpoint-1200/training_args.bin +3 -0
- checkpoint-1200/zero_to_fp32.py +674 -0
checkpoint-1200/README.md
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: meta-llama/CodeLlama-7b-Instruct-hf
|
| 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-1200/adapter_config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "meta-llama/CodeLlama-7b-Instruct-hf",
|
| 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 |
+
"k_proj",
|
| 27 |
+
"gate_proj",
|
| 28 |
+
"o_proj",
|
| 29 |
+
"v_proj",
|
| 30 |
+
"down_proj",
|
| 31 |
+
"q_proj",
|
| 32 |
+
"up_proj"
|
| 33 |
+
],
|
| 34 |
+
"task_type": "CAUSAL_LM",
|
| 35 |
+
"use_dora": false,
|
| 36 |
+
"use_rslora": false
|
| 37 |
+
}
|
checkpoint-1200/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dcbe5c60e63a74bda1718c743a5e3df0a742995ea8da64d57c7081a949df9d74
|
| 3 |
+
size 40036488
|
checkpoint-1200/global_step1200/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 sha256:7fe1756db9562682483681fc0b74814220242d3f6209454540e84f712781bf10
|
| 3 |
+
size 29992112
|
checkpoint-1200/global_step1200/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 sha256:867114061f55d410aa6e1567bb30478daf9fbfa3ca73c1331b6f732a7f1bc257
|
| 3 |
+
size 29992112
|
checkpoint-1200/global_step1200/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:a5bd6fb50c7afbccc666127692b9bdc31ffcb9cead83ab93b7650447f388788e
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/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 sha256:7dfa9f5b34f0e8245f0e4fa21f59fb815dc89e69bb492ce4fed17715707cd99f
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/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:f806c3f50b2274130f854cf2511793e9fcd6af288f91c9b89e8c9a3828cc1c7e
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/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:f16464e7f764ca085cc6db7df81762dffb9e3f1d380f7a68f487546c517d0225
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/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 sha256:6d691553a58819fb3d35afe07f84e6ed80f5e1d4a7010cc1a942880661a4a671
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/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:b630835b37b54fd53a8ba06d8bf85156927ee01187deb1c59c1585460297a879
|
| 3 |
+
size 29992176
|
checkpoint-1200/global_step1200/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eddc4e65a3915dda6a83824c4edfc3eb2dfe1ccea338ba7245f8a8007da324db
|
| 3 |
+
size 40324204
|
checkpoint-1200/latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step1200
|
checkpoint-1200/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5504b8d722b425f58bab6aedf9a43fc8129b02036307d31c7a21e224d2412ace
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e555860fd7a2cfb8945f188f7232baf938ce622886881cc422b3eb0e7444eda4
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea0c272cf77c9504efaa077bfa8f9229d461c16d6641be0e57a7f20f9b761399
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:921e0812b510be6ea788fc2c6aa7541f3ff4eb1bb3dd7c230340a35d8e1e764b
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c3603a5403f0489f4be4d27720a0fa7e0fe0d08dbde5d58c1060cef37b9084d2
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b9f685b83b5545cc2db9c29e88184590e89acb7836b4bb92a6a1df01b4bf43f
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5843202dd0ab5bc42fd0b6fa35e7cc2dca365d38fb379a2faf93bf274ef023e6
|
| 3 |
+
size 15984
|
checkpoint-1200/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79f72ffc2afb7672fc32ddd050c69181c2c0c16f8eac79a352eecb064fb5a9c7
|
| 3 |
+
size 15984
|
checkpoint-1200/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35b4cc7927ca0f1b9b45cd00f72746408c82d953cb952c75e7569243d9fa3f0c
|
| 3 |
+
size 1064
|
checkpoint-1200/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"▁<PRE>",
|
| 4 |
+
"▁<MID>",
|
| 5 |
+
"▁<SUF>",
|
| 6 |
+
"▁<EOT>"
|
| 7 |
+
],
|
| 8 |
+
"bos_token": {
|
| 9 |
+
"content": "<s>",
|
| 10 |
+
"lstrip": false,
|
| 11 |
+
"normalized": false,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"single_word": false
|
| 14 |
+
},
|
| 15 |
+
"eos_token": {
|
| 16 |
+
"content": "</s>",
|
| 17 |
+
"lstrip": false,
|
| 18 |
+
"normalized": false,
|
| 19 |
+
"rstrip": false,
|
| 20 |
+
"single_word": false
|
| 21 |
+
},
|
| 22 |
+
"pad_token": "</s>",
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
checkpoint-1200/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1200/tokenizer_config.json
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"0": {
|
| 6 |
+
"content": "<unk>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"1": {
|
| 14 |
+
"content": "<s>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"2": {
|
| 22 |
+
"content": "</s>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"32007": {
|
| 30 |
+
"content": "▁<PRE>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"32008": {
|
| 38 |
+
"content": "▁<SUF>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"32009": {
|
| 46 |
+
"content": "▁<MID>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"32010": {
|
| 54 |
+
"content": "▁<EOT>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
}
|
| 61 |
+
},
|
| 62 |
+
"additional_special_tokens": [
|
| 63 |
+
"▁<PRE>",
|
| 64 |
+
"▁<MID>",
|
| 65 |
+
"▁<SUF>",
|
| 66 |
+
"▁<EOT>"
|
| 67 |
+
],
|
| 68 |
+
"bos_token": "<s>",
|
| 69 |
+
"chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if loop.index0 == 0 and system_message is defined %}{% set content = '<<SYS>>\n' + system_message + '\n<</SYS>>\n\n' + message['content'] %}{% endif %}{% if message['role'] == 'user' %}{{ '<s>' + '[INST] ' + content + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + content + ' ' + '</s>' }}{% endif %}{% endfor %}",
|
| 70 |
+
"clean_up_tokenization_spaces": false,
|
| 71 |
+
"eos_token": "</s>",
|
| 72 |
+
"eot_token": "▁<EOT>",
|
| 73 |
+
"extra_special_tokens": {},
|
| 74 |
+
"fill_token": "<FILL_ME>",
|
| 75 |
+
"legacy": null,
|
| 76 |
+
"middle_token": "▁<MID>",
|
| 77 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 78 |
+
"pad_token": "</s>",
|
| 79 |
+
"padding_side": "right",
|
| 80 |
+
"prefix_token": "▁<PRE>",
|
| 81 |
+
"sp_model_kwargs": {},
|
| 82 |
+
"split_special_tokens": false,
|
| 83 |
+
"suffix_token": "▁<SUF>",
|
| 84 |
+
"tokenizer_class": "CodeLlamaTokenizer",
|
| 85 |
+
"unk_token": "<unk>",
|
| 86 |
+
"use_default_system_prompt": false
|
| 87 |
+
}
|
checkpoint-1200/trainer_state.json
ADDED
|
@@ -0,0 +1,2217 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8698015765153574,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 1200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.007248346470961312,
|
| 13 |
+
"grad_norm": 0.07201674580574036,
|
| 14 |
+
"learning_rate": 4.999451708687114e-06,
|
| 15 |
+
"logits/chosen": -1.9793262481689453,
|
| 16 |
+
"logits/rejected": -2.5381760597229004,
|
| 17 |
+
"logps/chosen": -0.28126341104507446,
|
| 18 |
+
"logps/rejected": -0.3779803514480591,
|
| 19 |
+
"loss": 7.3904,
|
| 20 |
+
"rewards/accuracies": 0.550000011920929,
|
| 21 |
+
"rewards/chosen": -0.4218950867652893,
|
| 22 |
+
"rewards/margins": 0.14507544040679932,
|
| 23 |
+
"rewards/rejected": -0.5669704675674438,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.014496692941922623,
|
| 28 |
+
"grad_norm": 0.07562297582626343,
|
| 29 |
+
"learning_rate": 4.997807075247147e-06,
|
| 30 |
+
"logits/chosen": -2.0567643642425537,
|
| 31 |
+
"logits/rejected": -2.4989147186279297,
|
| 32 |
+
"logps/chosen": -0.27690139412879944,
|
| 33 |
+
"logps/rejected": -0.33544114232063293,
|
| 34 |
+
"loss": 7.3756,
|
| 35 |
+
"rewards/accuracies": 0.574999988079071,
|
| 36 |
+
"rewards/chosen": -0.41535210609436035,
|
| 37 |
+
"rewards/margins": 0.08780960738658905,
|
| 38 |
+
"rewards/rejected": -0.5031617283821106,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.021745039412883936,
|
| 43 |
+
"grad_norm": 0.09685570746660233,
|
| 44 |
+
"learning_rate": 4.9950668210706795e-06,
|
| 45 |
+
"logits/chosen": -2.10174298286438,
|
| 46 |
+
"logits/rejected": -2.378197431564331,
|
| 47 |
+
"logps/chosen": -0.26717427372932434,
|
| 48 |
+
"logps/rejected": -0.30565372109413147,
|
| 49 |
+
"loss": 7.451,
|
| 50 |
+
"rewards/accuracies": 0.4375,
|
| 51 |
+
"rewards/chosen": -0.4007614254951477,
|
| 52 |
+
"rewards/margins": 0.05771917849779129,
|
| 53 |
+
"rewards/rejected": -0.4584805369377136,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.028993385883845247,
|
| 58 |
+
"grad_norm": 0.08213861286640167,
|
| 59 |
+
"learning_rate": 4.9912321481237616e-06,
|
| 60 |
+
"logits/chosen": -2.1633317470550537,
|
| 61 |
+
"logits/rejected": -2.387866497039795,
|
| 62 |
+
"logps/chosen": -0.27634260058403015,
|
| 63 |
+
"logps/rejected": -0.37035584449768066,
|
| 64 |
+
"loss": 7.3892,
|
| 65 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 66 |
+
"rewards/chosen": -0.4145139157772064,
|
| 67 |
+
"rewards/margins": 0.14101983606815338,
|
| 68 |
+
"rewards/rejected": -0.555533766746521,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.03624173235480656,
|
| 73 |
+
"grad_norm": 0.08846044540405273,
|
| 74 |
+
"learning_rate": 4.986304738420684e-06,
|
| 75 |
+
"logits/chosen": -2.1402599811553955,
|
| 76 |
+
"logits/rejected": -2.4459526538848877,
|
| 77 |
+
"logps/chosen": -0.2535383999347687,
|
| 78 |
+
"logps/rejected": -0.3090876042842865,
|
| 79 |
+
"loss": 7.5171,
|
| 80 |
+
"rewards/accuracies": 0.512499988079071,
|
| 81 |
+
"rewards/chosen": -0.3803076148033142,
|
| 82 |
+
"rewards/margins": 0.08332376182079315,
|
| 83 |
+
"rewards/rejected": -0.4636313319206238,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.03624173235480656,
|
| 88 |
+
"eval_logits/chosen": -2.1165692806243896,
|
| 89 |
+
"eval_logits/rejected": -2.476428747177124,
|
| 90 |
+
"eval_logps/chosen": -0.2828062176704407,
|
| 91 |
+
"eval_logps/rejected": -0.3432886600494385,
|
| 92 |
+
"eval_loss": 0.9120001792907715,
|
| 93 |
+
"eval_rewards/accuracies": 0.5089285969734192,
|
| 94 |
+
"eval_rewards/chosen": -0.4242093861103058,
|
| 95 |
+
"eval_rewards/margins": 0.09072363376617432,
|
| 96 |
+
"eval_rewards/rejected": -0.5149329900741577,
|
| 97 |
+
"eval_runtime": 30.971,
|
| 98 |
+
"eval_samples_per_second": 28.801,
|
| 99 |
+
"eval_steps_per_second": 3.616,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.04349007882576787,
|
| 104 |
+
"grad_norm": 0.11753705143928528,
|
| 105 |
+
"learning_rate": 4.980286753286196e-06,
|
| 106 |
+
"logits/chosen": -2.0575368404388428,
|
| 107 |
+
"logits/rejected": -2.5456700325012207,
|
| 108 |
+
"logps/chosen": -0.24622221291065216,
|
| 109 |
+
"logps/rejected": -0.32402220368385315,
|
| 110 |
+
"loss": 7.3926,
|
| 111 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 112 |
+
"rewards/chosen": -0.36933332681655884,
|
| 113 |
+
"rewards/margins": 0.11669999361038208,
|
| 114 |
+
"rewards/rejected": -0.48603329062461853,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.05073842529672919,
|
| 119 |
+
"grad_norm": 0.09996571391820908,
|
| 120 |
+
"learning_rate": 4.973180832407471e-06,
|
| 121 |
+
"logits/chosen": -1.9278606176376343,
|
| 122 |
+
"logits/rejected": -2.4620182514190674,
|
| 123 |
+
"logps/chosen": -0.2596542239189148,
|
| 124 |
+
"logps/rejected": -0.3665880560874939,
|
| 125 |
+
"loss": 7.185,
|
| 126 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 127 |
+
"rewards/chosen": -0.3894812762737274,
|
| 128 |
+
"rewards/margins": 0.160400852560997,
|
| 129 |
+
"rewards/rejected": -0.5498821139335632,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.057986771767690494,
|
| 134 |
+
"grad_norm": 0.07287321239709854,
|
| 135 |
+
"learning_rate": 4.964990092676263e-06,
|
| 136 |
+
"logits/chosen": -2.078031063079834,
|
| 137 |
+
"logits/rejected": -2.461479663848877,
|
| 138 |
+
"logps/chosen": -0.24513795971870422,
|
| 139 |
+
"logps/rejected": -0.3448730707168579,
|
| 140 |
+
"loss": 7.365,
|
| 141 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 142 |
+
"rewards/chosen": -0.36770695447921753,
|
| 143 |
+
"rewards/margins": 0.14960262179374695,
|
| 144 |
+
"rewards/rejected": -0.5173095464706421,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.0652351182386518,
|
| 149 |
+
"grad_norm": 0.09656044095754623,
|
| 150 |
+
"learning_rate": 4.9557181268217225e-06,
|
| 151 |
+
"logits/chosen": -1.9979238510131836,
|
| 152 |
+
"logits/rejected": -2.4381277561187744,
|
| 153 |
+
"logps/chosen": -0.24047240614891052,
|
| 154 |
+
"logps/rejected": -0.3277527987957001,
|
| 155 |
+
"loss": 7.2664,
|
| 156 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 157 |
+
"rewards/chosen": -0.36070865392684937,
|
| 158 |
+
"rewards/margins": 0.13092057406902313,
|
| 159 |
+
"rewards/rejected": -0.4916292130947113,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.07248346470961312,
|
| 164 |
+
"grad_norm": 0.08125138282775879,
|
| 165 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 166 |
+
"logits/chosen": -1.8948112726211548,
|
| 167 |
+
"logits/rejected": -2.4755520820617676,
|
| 168 |
+
"logps/chosen": -0.20189261436462402,
|
| 169 |
+
"logps/rejected": -0.29732149839401245,
|
| 170 |
+
"loss": 7.2542,
|
| 171 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 172 |
+
"rewards/chosen": -0.30283889174461365,
|
| 173 |
+
"rewards/margins": 0.14314329624176025,
|
| 174 |
+
"rewards/rejected": -0.44598227739334106,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.07248346470961312,
|
| 179 |
+
"eval_logits/chosen": -2.12322735786438,
|
| 180 |
+
"eval_logits/rejected": -2.481174945831299,
|
| 181 |
+
"eval_logps/chosen": -0.2438412606716156,
|
| 182 |
+
"eval_logps/rejected": -0.3260033428668976,
|
| 183 |
+
"eval_loss": 0.891861081123352,
|
| 184 |
+
"eval_rewards/accuracies": 0.5625,
|
| 185 |
+
"eval_rewards/chosen": -0.3657619059085846,
|
| 186 |
+
"eval_rewards/margins": 0.12324309349060059,
|
| 187 |
+
"eval_rewards/rejected": -0.48900502920150757,
|
| 188 |
+
"eval_runtime": 30.3299,
|
| 189 |
+
"eval_samples_per_second": 29.41,
|
| 190 |
+
"eval_steps_per_second": 3.693,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.07973181118057443,
|
| 195 |
+
"grad_norm": 0.05962231010198593,
|
| 196 |
+
"learning_rate": 4.933947257182901e-06,
|
| 197 |
+
"logits/chosen": -2.1134355068206787,
|
| 198 |
+
"logits/rejected": -2.516538381576538,
|
| 199 |
+
"logps/chosen": -0.237023264169693,
|
| 200 |
+
"logps/rejected": -0.31476154923439026,
|
| 201 |
+
"loss": 7.1749,
|
| 202 |
+
"rewards/accuracies": 0.550000011920929,
|
| 203 |
+
"rewards/chosen": -0.3555349111557007,
|
| 204 |
+
"rewards/margins": 0.11660744249820709,
|
| 205 |
+
"rewards/rejected": -0.4721423089504242,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.08698015765153574,
|
| 210 |
+
"grad_norm": 0.06015922501683235,
|
| 211 |
+
"learning_rate": 4.921457902821578e-06,
|
| 212 |
+
"logits/chosen": -2.0215041637420654,
|
| 213 |
+
"logits/rejected": -2.4902031421661377,
|
| 214 |
+
"logps/chosen": -0.1890055537223816,
|
| 215 |
+
"logps/rejected": -0.3192065358161926,
|
| 216 |
+
"loss": 7.142,
|
| 217 |
+
"rewards/accuracies": 0.637499988079071,
|
| 218 |
+
"rewards/chosen": -0.2835083603858948,
|
| 219 |
+
"rewards/margins": 0.19530144333839417,
|
| 220 |
+
"rewards/rejected": -0.47880974411964417,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.09422850412249706,
|
| 225 |
+
"grad_norm": 0.06430571526288986,
|
| 226 |
+
"learning_rate": 4.907906416994146e-06,
|
| 227 |
+
"logits/chosen": -2.0684752464294434,
|
| 228 |
+
"logits/rejected": -2.510018825531006,
|
| 229 |
+
"logps/chosen": -0.2073744535446167,
|
| 230 |
+
"logps/rejected": -0.3121300935745239,
|
| 231 |
+
"loss": 7.1438,
|
| 232 |
+
"rewards/accuracies": 0.612500011920929,
|
| 233 |
+
"rewards/chosen": -0.31106168031692505,
|
| 234 |
+
"rewards/margins": 0.15713343024253845,
|
| 235 |
+
"rewards/rejected": -0.4681951403617859,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.10147685059345837,
|
| 240 |
+
"grad_norm": 0.08829955011606216,
|
| 241 |
+
"learning_rate": 4.893298743830168e-06,
|
| 242 |
+
"logits/chosen": -2.026458263397217,
|
| 243 |
+
"logits/rejected": -2.496157646179199,
|
| 244 |
+
"logps/chosen": -0.19946983456611633,
|
| 245 |
+
"logps/rejected": -0.32050156593322754,
|
| 246 |
+
"loss": 7.1196,
|
| 247 |
+
"rewards/accuracies": 0.625,
|
| 248 |
+
"rewards/chosen": -0.2992047667503357,
|
| 249 |
+
"rewards/margins": 0.18154758214950562,
|
| 250 |
+
"rewards/rejected": -0.4807523787021637,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.10872519706441967,
|
| 255 |
+
"grad_norm": 0.09773921221494675,
|
| 256 |
+
"learning_rate": 4.8776412907378845e-06,
|
| 257 |
+
"logits/chosen": -2.101029872894287,
|
| 258 |
+
"logits/rejected": -2.5849032402038574,
|
| 259 |
+
"logps/chosen": -0.18889756500720978,
|
| 260 |
+
"logps/rejected": -0.36427801847457886,
|
| 261 |
+
"loss": 7.1227,
|
| 262 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 263 |
+
"rewards/chosen": -0.28334635496139526,
|
| 264 |
+
"rewards/margins": 0.26307064294815063,
|
| 265 |
+
"rewards/rejected": -0.5464169979095459,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.10872519706441967,
|
| 270 |
+
"eval_logits/chosen": -2.1945455074310303,
|
| 271 |
+
"eval_logits/rejected": -2.559415578842163,
|
| 272 |
+
"eval_logps/chosen": -0.22209034860134125,
|
| 273 |
+
"eval_logps/rejected": -0.32476040720939636,
|
| 274 |
+
"eval_loss": 0.8760393261909485,
|
| 275 |
+
"eval_rewards/accuracies": 0.5803571343421936,
|
| 276 |
+
"eval_rewards/chosen": -0.33313554525375366,
|
| 277 |
+
"eval_rewards/margins": 0.15400508046150208,
|
| 278 |
+
"eval_rewards/rejected": -0.48714062571525574,
|
| 279 |
+
"eval_runtime": 30.3484,
|
| 280 |
+
"eval_samples_per_second": 29.392,
|
| 281 |
+
"eval_steps_per_second": 3.69,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.11597354353538099,
|
| 286 |
+
"grad_norm": 0.09142427891492844,
|
| 287 |
+
"learning_rate": 4.860940925593703e-06,
|
| 288 |
+
"logits/chosen": -2.235069751739502,
|
| 289 |
+
"logits/rejected": -2.5513346195220947,
|
| 290 |
+
"logps/chosen": -0.19598741829395294,
|
| 291 |
+
"logps/rejected": -0.2772120535373688,
|
| 292 |
+
"loss": 7.0272,
|
| 293 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 294 |
+
"rewards/chosen": -0.2939811050891876,
|
| 295 |
+
"rewards/margins": 0.12183700501918793,
|
| 296 |
+
"rewards/rejected": -0.41581812500953674,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.1232218900063423,
|
| 301 |
+
"grad_norm": 0.11735275387763977,
|
| 302 |
+
"learning_rate": 4.84320497372973e-06,
|
| 303 |
+
"logits/chosen": -2.154263496398926,
|
| 304 |
+
"logits/rejected": -2.512010335922241,
|
| 305 |
+
"logps/chosen": -0.1798369437456131,
|
| 306 |
+
"logps/rejected": -0.30910637974739075,
|
| 307 |
+
"loss": 7.1603,
|
| 308 |
+
"rewards/accuracies": 0.637499988079071,
|
| 309 |
+
"rewards/chosen": -0.26975542306900024,
|
| 310 |
+
"rewards/margins": 0.19390416145324707,
|
| 311 |
+
"rewards/rejected": -0.4636595845222473,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.1304702364773036,
|
| 316 |
+
"grad_norm": 0.09398588538169861,
|
| 317 |
+
"learning_rate": 4.824441214720629e-06,
|
| 318 |
+
"logits/chosen": -2.292147159576416,
|
| 319 |
+
"logits/rejected": -2.581425189971924,
|
| 320 |
+
"logps/chosen": -0.18498703837394714,
|
| 321 |
+
"logps/rejected": -0.31237050890922546,
|
| 322 |
+
"loss": 6.995,
|
| 323 |
+
"rewards/accuracies": 0.637499988079071,
|
| 324 |
+
"rewards/chosen": -0.2774805426597595,
|
| 325 |
+
"rewards/margins": 0.19107523560523987,
|
| 326 |
+
"rewards/rejected": -0.4685557782649994,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.13771858294826492,
|
| 331 |
+
"grad_norm": 0.13551996648311615,
|
| 332 |
+
"learning_rate": 4.804657878971252e-06,
|
| 333 |
+
"logits/chosen": -2.274120330810547,
|
| 334 |
+
"logits/rejected": -2.5797386169433594,
|
| 335 |
+
"logps/chosen": -0.21168990433216095,
|
| 336 |
+
"logps/rejected": -0.34927254915237427,
|
| 337 |
+
"loss": 6.9941,
|
| 338 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 339 |
+
"rewards/chosen": -0.317534863948822,
|
| 340 |
+
"rewards/margins": 0.20637397468090057,
|
| 341 |
+
"rewards/rejected": -0.5239088535308838,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.14496692941922623,
|
| 346 |
+
"grad_norm": 0.15515944361686707,
|
| 347 |
+
"learning_rate": 4.783863644106502e-06,
|
| 348 |
+
"logits/chosen": -2.2116379737854004,
|
| 349 |
+
"logits/rejected": -2.6693203449249268,
|
| 350 |
+
"logps/chosen": -0.1971816122531891,
|
| 351 |
+
"logps/rejected": -0.3498842120170593,
|
| 352 |
+
"loss": 7.0413,
|
| 353 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 354 |
+
"rewards/chosen": -0.2957724332809448,
|
| 355 |
+
"rewards/margins": 0.22905388474464417,
|
| 356 |
+
"rewards/rejected": -0.5248263478279114,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.14496692941922623,
|
| 361 |
+
"eval_logits/chosen": -2.3369696140289307,
|
| 362 |
+
"eval_logits/rejected": -2.7298672199249268,
|
| 363 |
+
"eval_logps/chosen": -0.22749511897563934,
|
| 364 |
+
"eval_logps/rejected": -0.3613782525062561,
|
| 365 |
+
"eval_loss": 0.8543878793716431,
|
| 366 |
+
"eval_rewards/accuracies": 0.5714285969734192,
|
| 367 |
+
"eval_rewards/chosen": -0.34124264121055603,
|
| 368 |
+
"eval_rewards/margins": 0.2008247673511505,
|
| 369 |
+
"eval_rewards/rejected": -0.5420674681663513,
|
| 370 |
+
"eval_runtime": 30.3397,
|
| 371 |
+
"eval_samples_per_second": 29.4,
|
| 372 |
+
"eval_steps_per_second": 3.692,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.15221527589018755,
|
| 377 |
+
"grad_norm": 0.13039635121822357,
|
| 378 |
+
"learning_rate": 4.762067631165049e-06,
|
| 379 |
+
"logits/chosen": -2.2575619220733643,
|
| 380 |
+
"logits/rejected": -2.755174160003662,
|
| 381 |
+
"logps/chosen": -0.199218288064003,
|
| 382 |
+
"logps/rejected": -0.33260637521743774,
|
| 383 |
+
"loss": 6.882,
|
| 384 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 385 |
+
"rewards/chosen": -0.2988274097442627,
|
| 386 |
+
"rewards/margins": 0.20008206367492676,
|
| 387 |
+
"rewards/rejected": -0.49890947341918945,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.15946362236114886,
|
| 392 |
+
"grad_norm": 0.15020275115966797,
|
| 393 |
+
"learning_rate": 4.7392794005985324e-06,
|
| 394 |
+
"logits/chosen": -2.2949752807617188,
|
| 395 |
+
"logits/rejected": -2.7288482189178467,
|
| 396 |
+
"logps/chosen": -0.21987763047218323,
|
| 397 |
+
"logps/rejected": -0.385539174079895,
|
| 398 |
+
"loss": 6.9158,
|
| 399 |
+
"rewards/accuracies": 0.699999988079071,
|
| 400 |
+
"rewards/chosen": -0.3298164904117584,
|
| 401 |
+
"rewards/margins": 0.24849233031272888,
|
| 402 |
+
"rewards/rejected": -0.5783087611198425,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.16671196883211017,
|
| 407 |
+
"grad_norm": 0.16104522347450256,
|
| 408 |
+
"learning_rate": 4.715508948078037e-06,
|
| 409 |
+
"logits/chosen": -2.1837964057922363,
|
| 410 |
+
"logits/rejected": -2.813563585281372,
|
| 411 |
+
"logps/chosen": -0.2069036215543747,
|
| 412 |
+
"logps/rejected": -0.3996518552303314,
|
| 413 |
+
"loss": 6.8503,
|
| 414 |
+
"rewards/accuracies": 0.699999988079071,
|
| 415 |
+
"rewards/chosen": -0.31035539507865906,
|
| 416 |
+
"rewards/margins": 0.28912240266799927,
|
| 417 |
+
"rewards/rejected": -0.5994777679443359,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.1739603153030715,
|
| 422 |
+
"grad_norm": 0.16533692181110382,
|
| 423 |
+
"learning_rate": 4.690766700109659e-06,
|
| 424 |
+
"logits/chosen": -2.314570188522339,
|
| 425 |
+
"logits/rejected": -2.8867621421813965,
|
| 426 |
+
"logps/chosen": -0.18964803218841553,
|
| 427 |
+
"logps/rejected": -0.3997463583946228,
|
| 428 |
+
"loss": 6.667,
|
| 429 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 430 |
+
"rewards/chosen": -0.2844720482826233,
|
| 431 |
+
"rewards/margins": 0.3151474893093109,
|
| 432 |
+
"rewards/rejected": -0.5996195077896118,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.1812086617740328,
|
| 437 |
+
"grad_norm": 0.24991220235824585,
|
| 438 |
+
"learning_rate": 4.665063509461098e-06,
|
| 439 |
+
"logits/chosen": -2.46189546585083,
|
| 440 |
+
"logits/rejected": -2.847446918487549,
|
| 441 |
+
"logps/chosen": -0.24189543724060059,
|
| 442 |
+
"logps/rejected": -0.4183991551399231,
|
| 443 |
+
"loss": 6.7146,
|
| 444 |
+
"rewards/accuracies": 0.5625,
|
| 445 |
+
"rewards/chosen": -0.3628431558609009,
|
| 446 |
+
"rewards/margins": 0.26475557684898376,
|
| 447 |
+
"rewards/rejected": -0.6275987029075623,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.1812086617740328,
|
| 452 |
+
"eval_logits/chosen": -2.5230772495269775,
|
| 453 |
+
"eval_logits/rejected": -2.9182300567626953,
|
| 454 |
+
"eval_logps/chosen": -0.2819940149784088,
|
| 455 |
+
"eval_logps/rejected": -0.48641347885131836,
|
| 456 |
+
"eval_loss": 0.8070082664489746,
|
| 457 |
+
"eval_rewards/accuracies": 0.5803571343421936,
|
| 458 |
+
"eval_rewards/chosen": -0.4229910671710968,
|
| 459 |
+
"eval_rewards/margins": 0.3066291809082031,
|
| 460 |
+
"eval_rewards/rejected": -0.7296201586723328,
|
| 461 |
+
"eval_runtime": 30.3523,
|
| 462 |
+
"eval_samples_per_second": 29.388,
|
| 463 |
+
"eval_steps_per_second": 3.69,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.18845700824499412,
|
| 468 |
+
"grad_norm": 0.3010155260562897,
|
| 469 |
+
"learning_rate": 4.638410650401267e-06,
|
| 470 |
+
"logits/chosen": -2.4178171157836914,
|
| 471 |
+
"logits/rejected": -2.914424419403076,
|
| 472 |
+
"logps/chosen": -0.24933743476867676,
|
| 473 |
+
"logps/rejected": -0.4757954478263855,
|
| 474 |
+
"loss": 6.5743,
|
| 475 |
+
"rewards/accuracies": 0.637499988079071,
|
| 476 |
+
"rewards/chosen": -0.3740061819553375,
|
| 477 |
+
"rewards/margins": 0.33968693017959595,
|
| 478 |
+
"rewards/rejected": -0.7136931419372559,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.19570535471595543,
|
| 483 |
+
"grad_norm": 0.3765665888786316,
|
| 484 |
+
"learning_rate": 4.610819813755038e-06,
|
| 485 |
+
"logits/chosen": -2.407275676727295,
|
| 486 |
+
"logits/rejected": -2.886014938354492,
|
| 487 |
+
"logps/chosen": -0.27837619185447693,
|
| 488 |
+
"logps/rejected": -0.6118712425231934,
|
| 489 |
+
"loss": 6.4931,
|
| 490 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 491 |
+
"rewards/chosen": -0.417564332485199,
|
| 492 |
+
"rewards/margins": 0.5002425909042358,
|
| 493 |
+
"rewards/rejected": -0.9178068041801453,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.20295370118691675,
|
| 498 |
+
"grad_norm": 0.29865291714668274,
|
| 499 |
+
"learning_rate": 4.582303101775249e-06,
|
| 500 |
+
"logits/chosen": -2.5404791831970215,
|
| 501 |
+
"logits/rejected": -2.861341953277588,
|
| 502 |
+
"logps/chosen": -0.3124132752418518,
|
| 503 |
+
"logps/rejected": -0.558597207069397,
|
| 504 |
+
"loss": 6.2272,
|
| 505 |
+
"rewards/accuracies": 0.550000011920929,
|
| 506 |
+
"rewards/chosen": -0.4686199128627777,
|
| 507 |
+
"rewards/margins": 0.3692759573459625,
|
| 508 |
+
"rewards/rejected": -0.8378958702087402,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.21020204765787803,
|
| 513 |
+
"grad_norm": 0.4311545789241791,
|
| 514 |
+
"learning_rate": 4.55287302283426e-06,
|
| 515 |
+
"logits/chosen": -2.393683910369873,
|
| 516 |
+
"logits/rejected": -2.833466053009033,
|
| 517 |
+
"logps/chosen": -0.3574284017086029,
|
| 518 |
+
"logps/rejected": -0.8571383357048035,
|
| 519 |
+
"loss": 5.7937,
|
| 520 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 521 |
+
"rewards/chosen": -0.5361425876617432,
|
| 522 |
+
"rewards/margins": 0.7495648264884949,
|
| 523 |
+
"rewards/rejected": -1.2857074737548828,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.21745039412883935,
|
| 528 |
+
"grad_norm": 0.7148597836494446,
|
| 529 |
+
"learning_rate": 4.522542485937369e-06,
|
| 530 |
+
"logits/chosen": -2.369452476501465,
|
| 531 |
+
"logits/rejected": -2.7431299686431885,
|
| 532 |
+
"logps/chosen": -0.46852874755859375,
|
| 533 |
+
"logps/rejected": -0.9673423767089844,
|
| 534 |
+
"loss": 5.741,
|
| 535 |
+
"rewards/accuracies": 0.6875,
|
| 536 |
+
"rewards/chosen": -0.7027931213378906,
|
| 537 |
+
"rewards/margins": 0.7482204437255859,
|
| 538 |
+
"rewards/rejected": -1.4510136842727661,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.21745039412883935,
|
| 543 |
+
"eval_logits/chosen": -2.5259616374969482,
|
| 544 |
+
"eval_logits/rejected": -2.8154189586639404,
|
| 545 |
+
"eval_logps/chosen": -0.5109023451805115,
|
| 546 |
+
"eval_logps/rejected": -0.9421015381813049,
|
| 547 |
+
"eval_loss": 0.6990054845809937,
|
| 548 |
+
"eval_rewards/accuracies": 0.5714285969734192,
|
| 549 |
+
"eval_rewards/chosen": -0.7663536071777344,
|
| 550 |
+
"eval_rewards/margins": 0.6467987895011902,
|
| 551 |
+
"eval_rewards/rejected": -1.4131524562835693,
|
| 552 |
+
"eval_runtime": 30.3501,
|
| 553 |
+
"eval_samples_per_second": 29.39,
|
| 554 |
+
"eval_steps_per_second": 3.69,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.22469874059980066,
|
| 559 |
+
"grad_norm": 0.37795087695121765,
|
| 560 |
+
"learning_rate": 4.491324795060491e-06,
|
| 561 |
+
"logits/chosen": -2.5089354515075684,
|
| 562 |
+
"logits/rejected": -2.827730894088745,
|
| 563 |
+
"logps/chosen": -0.4125432074069977,
|
| 564 |
+
"logps/rejected": -1.0391974449157715,
|
| 565 |
+
"loss": 5.3005,
|
| 566 |
+
"rewards/accuracies": 0.625,
|
| 567 |
+
"rewards/chosen": -0.6188148260116577,
|
| 568 |
+
"rewards/margins": 0.9399812817573547,
|
| 569 |
+
"rewards/rejected": -1.5587961673736572,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.23194708707076198,
|
| 574 |
+
"grad_norm": 0.38493892550468445,
|
| 575 |
+
"learning_rate": 4.4592336433146e-06,
|
| 576 |
+
"logits/chosen": -2.3596529960632324,
|
| 577 |
+
"logits/rejected": -2.8088250160217285,
|
| 578 |
+
"logps/chosen": -0.5338706970214844,
|
| 579 |
+
"logps/rejected": -1.2868502140045166,
|
| 580 |
+
"loss": 5.2832,
|
| 581 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 582 |
+
"rewards/chosen": -0.8008060455322266,
|
| 583 |
+
"rewards/margins": 1.1294692754745483,
|
| 584 |
+
"rewards/rejected": -1.930275321006775,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.2391954335417233,
|
| 589 |
+
"grad_norm": 0.6237270832061768,
|
| 590 |
+
"learning_rate": 4.426283106939474e-06,
|
| 591 |
+
"logits/chosen": -2.5265612602233887,
|
| 592 |
+
"logits/rejected": -2.7830424308776855,
|
| 593 |
+
"logps/chosen": -0.6278705596923828,
|
| 594 |
+
"logps/rejected": -1.4297993183135986,
|
| 595 |
+
"loss": 5.1977,
|
| 596 |
+
"rewards/accuracies": 0.6875,
|
| 597 |
+
"rewards/chosen": -0.9418058395385742,
|
| 598 |
+
"rewards/margins": 1.2028930187225342,
|
| 599 |
+
"rewards/rejected": -2.1446988582611084,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.2464437800126846,
|
| 604 |
+
"grad_norm": 0.9095363020896912,
|
| 605 |
+
"learning_rate": 4.3924876391293915e-06,
|
| 606 |
+
"logits/chosen": -2.5918514728546143,
|
| 607 |
+
"logits/rejected": -2.8718667030334473,
|
| 608 |
+
"logps/chosen": -0.6412171125411987,
|
| 609 |
+
"logps/rejected": -1.508195161819458,
|
| 610 |
+
"loss": 4.7181,
|
| 611 |
+
"rewards/accuracies": 0.625,
|
| 612 |
+
"rewards/chosen": -0.9618256688117981,
|
| 613 |
+
"rewards/margins": 1.3004668951034546,
|
| 614 |
+
"rewards/rejected": -2.2622926235198975,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.2536921264836459,
|
| 619 |
+
"grad_norm": 0.6144569516181946,
|
| 620 |
+
"learning_rate": 4.357862063693486e-06,
|
| 621 |
+
"logits/chosen": -2.492586612701416,
|
| 622 |
+
"logits/rejected": -2.9216372966766357,
|
| 623 |
+
"logps/chosen": -0.7246330976486206,
|
| 624 |
+
"logps/rejected": -2.0466060638427734,
|
| 625 |
+
"loss": 4.6111,
|
| 626 |
+
"rewards/accuracies": 0.675000011920929,
|
| 627 |
+
"rewards/chosen": -1.0869497060775757,
|
| 628 |
+
"rewards/margins": 1.9829593896865845,
|
| 629 |
+
"rewards/rejected": -3.06990909576416,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.2536921264836459,
|
| 634 |
+
"eval_logits/chosen": -2.7258641719818115,
|
| 635 |
+
"eval_logits/rejected": -2.9883599281311035,
|
| 636 |
+
"eval_logps/chosen": -0.7502545118331909,
|
| 637 |
+
"eval_logps/rejected": -1.7796010971069336,
|
| 638 |
+
"eval_loss": 0.571855366230011,
|
| 639 |
+
"eval_rewards/accuracies": 0.5892857313156128,
|
| 640 |
+
"eval_rewards/chosen": -1.1253817081451416,
|
| 641 |
+
"eval_rewards/margins": 1.5440199375152588,
|
| 642 |
+
"eval_rewards/rejected": -2.6694016456604004,
|
| 643 |
+
"eval_runtime": 30.3197,
|
| 644 |
+
"eval_samples_per_second": 29.42,
|
| 645 |
+
"eval_steps_per_second": 3.694,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.2609404729546072,
|
| 650 |
+
"grad_norm": 0.6816007494926453,
|
| 651 |
+
"learning_rate": 4.322421568553529e-06,
|
| 652 |
+
"logits/chosen": -2.6618990898132324,
|
| 653 |
+
"logits/rejected": -2.9197638034820557,
|
| 654 |
+
"logps/chosen": -0.7783921360969543,
|
| 655 |
+
"logps/rejected": -1.9845831394195557,
|
| 656 |
+
"loss": 4.8991,
|
| 657 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 658 |
+
"rewards/chosen": -1.1675881147384644,
|
| 659 |
+
"rewards/margins": 1.8092864751815796,
|
| 660 |
+
"rewards/rejected": -2.976874589920044,
|
| 661 |
+
"step": 360
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.26818881942556855,
|
| 665 |
+
"grad_norm": 0.5186926126480103,
|
| 666 |
+
"learning_rate": 4.286181699082008e-06,
|
| 667 |
+
"logits/chosen": -2.690049171447754,
|
| 668 |
+
"logits/rejected": -2.9129927158355713,
|
| 669 |
+
"logps/chosen": -0.7160422205924988,
|
| 670 |
+
"logps/rejected": -2.097799777984619,
|
| 671 |
+
"loss": 4.6471,
|
| 672 |
+
"rewards/accuracies": 0.637499988079071,
|
| 673 |
+
"rewards/chosen": -1.0740633010864258,
|
| 674 |
+
"rewards/margins": 2.0726354122161865,
|
| 675 |
+
"rewards/rejected": -3.1466989517211914,
|
| 676 |
+
"step": 370
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.27543716589652983,
|
| 680 |
+
"grad_norm": 0.7122822403907776,
|
| 681 |
+
"learning_rate": 4.249158351283414e-06,
|
| 682 |
+
"logits/chosen": -2.6984095573425293,
|
| 683 |
+
"logits/rejected": -2.8411834239959717,
|
| 684 |
+
"logps/chosen": -0.724581778049469,
|
| 685 |
+
"logps/rejected": -2.1322455406188965,
|
| 686 |
+
"loss": 4.4326,
|
| 687 |
+
"rewards/accuracies": 0.637499988079071,
|
| 688 |
+
"rewards/chosen": -1.0868725776672363,
|
| 689 |
+
"rewards/margins": 2.1114957332611084,
|
| 690 |
+
"rewards/rejected": -3.198368549346924,
|
| 691 |
+
"step": 380
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.2826855123674912,
|
| 695 |
+
"grad_norm": 0.610661506652832,
|
| 696 |
+
"learning_rate": 4.211367764821722e-06,
|
| 697 |
+
"logits/chosen": -2.5932717323303223,
|
| 698 |
+
"logits/rejected": -2.855365514755249,
|
| 699 |
+
"logps/chosen": -0.6430361866950989,
|
| 700 |
+
"logps/rejected": -2.8457767963409424,
|
| 701 |
+
"loss": 4.3874,
|
| 702 |
+
"rewards/accuracies": 0.824999988079071,
|
| 703 |
+
"rewards/chosen": -0.9645543098449707,
|
| 704 |
+
"rewards/margins": 3.3041110038757324,
|
| 705 |
+
"rewards/rejected": -4.268665313720703,
|
| 706 |
+
"step": 390
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.28993385883845246,
|
| 710 |
+
"grad_norm": 0.627453088760376,
|
| 711 |
+
"learning_rate": 4.172826515897146e-06,
|
| 712 |
+
"logits/chosen": -2.46079683303833,
|
| 713 |
+
"logits/rejected": -2.839109182357788,
|
| 714 |
+
"logps/chosen": -0.8076715469360352,
|
| 715 |
+
"logps/rejected": -2.6943671703338623,
|
| 716 |
+
"loss": 4.2388,
|
| 717 |
+
"rewards/accuracies": 0.6875,
|
| 718 |
+
"rewards/chosen": -1.2115072011947632,
|
| 719 |
+
"rewards/margins": 2.8300435543060303,
|
| 720 |
+
"rewards/rejected": -4.041550636291504,
|
| 721 |
+
"step": 400
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.28993385883845246,
|
| 725 |
+
"eval_logits/chosen": -2.5736935138702393,
|
| 726 |
+
"eval_logits/rejected": -2.8549301624298096,
|
| 727 |
+
"eval_logps/chosen": -0.8027606010437012,
|
| 728 |
+
"eval_logps/rejected": -2.407195568084717,
|
| 729 |
+
"eval_loss": 0.49557629227638245,
|
| 730 |
+
"eval_rewards/accuracies": 0.6071428656578064,
|
| 731 |
+
"eval_rewards/chosen": -1.2041409015655518,
|
| 732 |
+
"eval_rewards/margins": 2.4066522121429443,
|
| 733 |
+
"eval_rewards/rejected": -3.6107935905456543,
|
| 734 |
+
"eval_runtime": 30.3144,
|
| 735 |
+
"eval_samples_per_second": 29.425,
|
| 736 |
+
"eval_steps_per_second": 3.695,
|
| 737 |
+
"step": 400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.2971822053094138,
|
| 741 |
+
"grad_norm": 0.582290768623352,
|
| 742 |
+
"learning_rate": 4.133551509975264e-06,
|
| 743 |
+
"logits/chosen": -2.5232419967651367,
|
| 744 |
+
"logits/rejected": -2.8139169216156006,
|
| 745 |
+
"logps/chosen": -0.7921724319458008,
|
| 746 |
+
"logps/rejected": -2.5887393951416016,
|
| 747 |
+
"loss": 3.9161,
|
| 748 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 749 |
+
"rewards/chosen": -1.1882586479187012,
|
| 750 |
+
"rewards/margins": 2.694850206375122,
|
| 751 |
+
"rewards/rejected": -3.883108615875244,
|
| 752 |
+
"step": 410
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.3044305517803751,
|
| 756 |
+
"grad_norm": 1.0708731412887573,
|
| 757 |
+
"learning_rate": 4.093559974371725e-06,
|
| 758 |
+
"logits/chosen": -2.402315616607666,
|
| 759 |
+
"logits/rejected": -2.6774837970733643,
|
| 760 |
+
"logps/chosen": -0.9347984194755554,
|
| 761 |
+
"logps/rejected": -3.4818978309631348,
|
| 762 |
+
"loss": 3.8304,
|
| 763 |
+
"rewards/accuracies": 0.762499988079071,
|
| 764 |
+
"rewards/chosen": -1.4021978378295898,
|
| 765 |
+
"rewards/margins": 3.8206489086151123,
|
| 766 |
+
"rewards/rejected": -5.222846508026123,
|
| 767 |
+
"step": 420
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.31167889825133643,
|
| 771 |
+
"grad_norm": 0.6126013398170471,
|
| 772 |
+
"learning_rate": 4.052869450695776e-06,
|
| 773 |
+
"logits/chosen": -2.4616169929504395,
|
| 774 |
+
"logits/rejected": -2.791163921356201,
|
| 775 |
+
"logps/chosen": -1.073099970817566,
|
| 776 |
+
"logps/rejected": -3.7586822509765625,
|
| 777 |
+
"loss": 3.7662,
|
| 778 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 779 |
+
"rewards/chosen": -1.609649896621704,
|
| 780 |
+
"rewards/margins": 4.028374671936035,
|
| 781 |
+
"rewards/rejected": -5.63802433013916,
|
| 782 |
+
"step": 430
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.3189272447222977,
|
| 786 |
+
"grad_norm": 0.6995553374290466,
|
| 787 |
+
"learning_rate": 4.011497787155938e-06,
|
| 788 |
+
"logits/chosen": -2.350310802459717,
|
| 789 |
+
"logits/rejected": -2.750123977661133,
|
| 790 |
+
"logps/chosen": -1.2352540493011475,
|
| 791 |
+
"logps/rejected": -4.180838584899902,
|
| 792 |
+
"loss": 3.4493,
|
| 793 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 794 |
+
"rewards/chosen": -1.852880835533142,
|
| 795 |
+
"rewards/margins": 4.418376922607422,
|
| 796 |
+
"rewards/rejected": -6.271258354187012,
|
| 797 |
+
"step": 440
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.32617559119325906,
|
| 801 |
+
"grad_norm": 0.7922266721725464,
|
| 802 |
+
"learning_rate": 3.969463130731183e-06,
|
| 803 |
+
"logits/chosen": -2.368863344192505,
|
| 804 |
+
"logits/rejected": -2.7451913356781006,
|
| 805 |
+
"logps/chosen": -1.0907175540924072,
|
| 806 |
+
"logps/rejected": -4.085494041442871,
|
| 807 |
+
"loss": 4.0187,
|
| 808 |
+
"rewards/accuracies": 0.8125,
|
| 809 |
+
"rewards/chosen": -1.6360763311386108,
|
| 810 |
+
"rewards/margins": 4.4921650886535645,
|
| 811 |
+
"rewards/rejected": -6.128241062164307,
|
| 812 |
+
"step": 450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.32617559119325906,
|
| 816 |
+
"eval_logits/chosen": -2.44749116897583,
|
| 817 |
+
"eval_logits/rejected": -2.7382781505584717,
|
| 818 |
+
"eval_logps/chosen": -1.3378121852874756,
|
| 819 |
+
"eval_logps/rejected": -3.6518471240997314,
|
| 820 |
+
"eval_loss": 0.45260879397392273,
|
| 821 |
+
"eval_rewards/accuracies": 0.6428571343421936,
|
| 822 |
+
"eval_rewards/chosen": -2.006718158721924,
|
| 823 |
+
"eval_rewards/margins": 3.4710519313812256,
|
| 824 |
+
"eval_rewards/rejected": -5.4777703285217285,
|
| 825 |
+
"eval_runtime": 30.3269,
|
| 826 |
+
"eval_samples_per_second": 29.413,
|
| 827 |
+
"eval_steps_per_second": 3.693,
|
| 828 |
+
"step": 450
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.33342393766422035,
|
| 832 |
+
"grad_norm": 0.7150818109512329,
|
| 833 |
+
"learning_rate": 3.92678391921108e-06,
|
| 834 |
+
"logits/chosen": -2.4681639671325684,
|
| 835 |
+
"logits/rejected": -2.797795057296753,
|
| 836 |
+
"logps/chosen": -1.4199602603912354,
|
| 837 |
+
"logps/rejected": -4.342296600341797,
|
| 838 |
+
"loss": 3.5788,
|
| 839 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 840 |
+
"rewards/chosen": -2.1299405097961426,
|
| 841 |
+
"rewards/margins": 4.3835039138793945,
|
| 842 |
+
"rewards/rejected": -6.513444423675537,
|
| 843 |
+
"step": 460
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.34067228413518164,
|
| 847 |
+
"grad_norm": 1.260711431503296,
|
| 848 |
+
"learning_rate": 3.88347887310836e-06,
|
| 849 |
+
"logits/chosen": -2.4320380687713623,
|
| 850 |
+
"logits/rejected": -2.749352216720581,
|
| 851 |
+
"logps/chosen": -1.318566918373108,
|
| 852 |
+
"logps/rejected": -3.7736339569091797,
|
| 853 |
+
"loss": 3.5681,
|
| 854 |
+
"rewards/accuracies": 0.75,
|
| 855 |
+
"rewards/chosen": -1.977850317955017,
|
| 856 |
+
"rewards/margins": 3.682600498199463,
|
| 857 |
+
"rewards/rejected": -5.6604509353637695,
|
| 858 |
+
"step": 470
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.347920630606143,
|
| 862 |
+
"grad_norm": 0.9010512828826904,
|
| 863 |
+
"learning_rate": 3.839566987447492e-06,
|
| 864 |
+
"logits/chosen": -2.3840298652648926,
|
| 865 |
+
"logits/rejected": -2.7868971824645996,
|
| 866 |
+
"logps/chosen": -1.6358540058135986,
|
| 867 |
+
"logps/rejected": -5.068122386932373,
|
| 868 |
+
"loss": 3.2884,
|
| 869 |
+
"rewards/accuracies": 0.862500011920929,
|
| 870 |
+
"rewards/chosen": -2.4537811279296875,
|
| 871 |
+
"rewards/margins": 5.148402690887451,
|
| 872 |
+
"rewards/rejected": -7.602184295654297,
|
| 873 |
+
"step": 480
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.35516897707710426,
|
| 877 |
+
"grad_norm": 0.4977961778640747,
|
| 878 |
+
"learning_rate": 3.795067523432826e-06,
|
| 879 |
+
"logits/chosen": -2.4045886993408203,
|
| 880 |
+
"logits/rejected": -2.813812494277954,
|
| 881 |
+
"logps/chosen": -1.5761570930480957,
|
| 882 |
+
"logps/rejected": -5.026966094970703,
|
| 883 |
+
"loss": 3.5505,
|
| 884 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 885 |
+
"rewards/chosen": -2.3642354011535645,
|
| 886 |
+
"rewards/margins": 5.176213264465332,
|
| 887 |
+
"rewards/rejected": -7.5404486656188965,
|
| 888 |
+
"step": 490
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.3624173235480656,
|
| 892 |
+
"grad_norm": 0.9888324737548828,
|
| 893 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 894 |
+
"logits/chosen": -2.4877827167510986,
|
| 895 |
+
"logits/rejected": -2.7923107147216797,
|
| 896 |
+
"logps/chosen": -2.0706987380981445,
|
| 897 |
+
"logps/rejected": -5.011548042297363,
|
| 898 |
+
"loss": 3.3992,
|
| 899 |
+
"rewards/accuracies": 0.800000011920929,
|
| 900 |
+
"rewards/chosen": -3.106048107147217,
|
| 901 |
+
"rewards/margins": 4.411273002624512,
|
| 902 |
+
"rewards/rejected": -7.517321586608887,
|
| 903 |
+
"step": 500
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.3624173235480656,
|
| 907 |
+
"eval_logits/chosen": -2.5850253105163574,
|
| 908 |
+
"eval_logits/rejected": -2.898167133331299,
|
| 909 |
+
"eval_logps/chosen": -2.3814480304718018,
|
| 910 |
+
"eval_logps/rejected": -5.132807731628418,
|
| 911 |
+
"eval_loss": 0.40332508087158203,
|
| 912 |
+
"eval_rewards/accuracies": 0.7142857313156128,
|
| 913 |
+
"eval_rewards/chosen": -3.572171926498413,
|
| 914 |
+
"eval_rewards/margins": 4.127039432525635,
|
| 915 |
+
"eval_rewards/rejected": -7.699211597442627,
|
| 916 |
+
"eval_runtime": 30.324,
|
| 917 |
+
"eval_samples_per_second": 29.416,
|
| 918 |
+
"eval_steps_per_second": 3.693,
|
| 919 |
+
"step": 500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.3696656700190269,
|
| 923 |
+
"grad_norm": 0.9108109474182129,
|
| 924 |
+
"learning_rate": 3.7043841852542884e-06,
|
| 925 |
+
"logits/chosen": -2.4516501426696777,
|
| 926 |
+
"logits/rejected": -2.863548755645752,
|
| 927 |
+
"logps/chosen": -2.235166549682617,
|
| 928 |
+
"logps/rejected": -5.824828147888184,
|
| 929 |
+
"loss": 3.2045,
|
| 930 |
+
"rewards/accuracies": 0.824999988079071,
|
| 931 |
+
"rewards/chosen": -3.352750301361084,
|
| 932 |
+
"rewards/margins": 5.384493350982666,
|
| 933 |
+
"rewards/rejected": -8.737241744995117,
|
| 934 |
+
"step": 510
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.37691401648998824,
|
| 938 |
+
"grad_norm": 1.608782410621643,
|
| 939 |
+
"learning_rate": 3.658240087799655e-06,
|
| 940 |
+
"logits/chosen": -2.448782444000244,
|
| 941 |
+
"logits/rejected": -2.9055087566375732,
|
| 942 |
+
"logps/chosen": -2.345585584640503,
|
| 943 |
+
"logps/rejected": -5.665570259094238,
|
| 944 |
+
"loss": 3.2997,
|
| 945 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 946 |
+
"rewards/chosen": -3.518378496170044,
|
| 947 |
+
"rewards/margins": 4.979977130889893,
|
| 948 |
+
"rewards/rejected": -8.498355865478516,
|
| 949 |
+
"step": 520
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.3841623629609495,
|
| 953 |
+
"grad_norm": 2.4802284240722656,
|
| 954 |
+
"learning_rate": 3.611587947962319e-06,
|
| 955 |
+
"logits/chosen": -2.5400586128234863,
|
| 956 |
+
"logits/rejected": -2.9653382301330566,
|
| 957 |
+
"logps/chosen": -2.9243946075439453,
|
| 958 |
+
"logps/rejected": -7.067171573638916,
|
| 959 |
+
"loss": 2.7084,
|
| 960 |
+
"rewards/accuracies": 0.925000011920929,
|
| 961 |
+
"rewards/chosen": -4.386591911315918,
|
| 962 |
+
"rewards/margins": 6.214165210723877,
|
| 963 |
+
"rewards/rejected": -10.60075569152832,
|
| 964 |
+
"step": 530
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.39141070943191086,
|
| 968 |
+
"grad_norm": 1.4558824300765991,
|
| 969 |
+
"learning_rate": 3.564448228912682e-06,
|
| 970 |
+
"logits/chosen": -2.538511276245117,
|
| 971 |
+
"logits/rejected": -3.0383963584899902,
|
| 972 |
+
"logps/chosen": -3.83245849609375,
|
| 973 |
+
"logps/rejected": -7.150879859924316,
|
| 974 |
+
"loss": 2.85,
|
| 975 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 976 |
+
"rewards/chosen": -5.748687744140625,
|
| 977 |
+
"rewards/margins": 4.977631568908691,
|
| 978 |
+
"rewards/rejected": -10.726319313049316,
|
| 979 |
+
"step": 540
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 0.39865905590287215,
|
| 983 |
+
"grad_norm": 6.12661600112915,
|
| 984 |
+
"learning_rate": 3.516841607689501e-06,
|
| 985 |
+
"logits/chosen": -2.5623717308044434,
|
| 986 |
+
"logits/rejected": -2.930835485458374,
|
| 987 |
+
"logps/chosen": -3.9018759727478027,
|
| 988 |
+
"logps/rejected": -7.186791896820068,
|
| 989 |
+
"loss": 2.6406,
|
| 990 |
+
"rewards/accuracies": 0.875,
|
| 991 |
+
"rewards/chosen": -5.852814674377441,
|
| 992 |
+
"rewards/margins": 4.92737340927124,
|
| 993 |
+
"rewards/rejected": -10.78018856048584,
|
| 994 |
+
"step": 550
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.39865905590287215,
|
| 998 |
+
"eval_logits/chosen": -2.5746138095855713,
|
| 999 |
+
"eval_logits/rejected": -2.9623827934265137,
|
| 1000 |
+
"eval_logps/chosen": -4.4805707931518555,
|
| 1001 |
+
"eval_logps/rejected": -7.7552666664123535,
|
| 1002 |
+
"eval_loss": 0.30369508266448975,
|
| 1003 |
+
"eval_rewards/accuracies": 0.8660714030265808,
|
| 1004 |
+
"eval_rewards/chosen": -6.720856189727783,
|
| 1005 |
+
"eval_rewards/margins": 4.912044048309326,
|
| 1006 |
+
"eval_rewards/rejected": -11.632901191711426,
|
| 1007 |
+
"eval_runtime": 30.3366,
|
| 1008 |
+
"eval_samples_per_second": 29.403,
|
| 1009 |
+
"eval_steps_per_second": 3.692,
|
| 1010 |
+
"step": 550
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.4059074023738335,
|
| 1014 |
+
"grad_norm": 2.2883663177490234,
|
| 1015 |
+
"learning_rate": 3.4687889661302577e-06,
|
| 1016 |
+
"logits/chosen": -2.523510456085205,
|
| 1017 |
+
"logits/rejected": -3.0058722496032715,
|
| 1018 |
+
"logps/chosen": -4.60703706741333,
|
| 1019 |
+
"logps/rejected": -8.818490982055664,
|
| 1020 |
+
"loss": 2.1664,
|
| 1021 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1022 |
+
"rewards/chosen": -6.9105544090271,
|
| 1023 |
+
"rewards/margins": 6.3171820640563965,
|
| 1024 |
+
"rewards/rejected": -13.22773551940918,
|
| 1025 |
+
"step": 560
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.4131557488447948,
|
| 1029 |
+
"grad_norm": 3.3689115047454834,
|
| 1030 |
+
"learning_rate": 3.4203113817116955e-06,
|
| 1031 |
+
"logits/chosen": -2.592515230178833,
|
| 1032 |
+
"logits/rejected": -2.9502272605895996,
|
| 1033 |
+
"logps/chosen": -5.648528099060059,
|
| 1034 |
+
"logps/rejected": -10.127589225769043,
|
| 1035 |
+
"loss": 2.2641,
|
| 1036 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1037 |
+
"rewards/chosen": -8.472792625427246,
|
| 1038 |
+
"rewards/margins": 6.718589782714844,
|
| 1039 |
+
"rewards/rejected": -15.191381454467773,
|
| 1040 |
+
"step": 570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.42040409531575607,
|
| 1044 |
+
"grad_norm": 4.080602169036865,
|
| 1045 |
+
"learning_rate": 3.3714301183045382e-06,
|
| 1046 |
+
"logits/chosen": -2.418156862258911,
|
| 1047 |
+
"logits/rejected": -3.0541069507598877,
|
| 1048 |
+
"logps/chosen": -5.756854057312012,
|
| 1049 |
+
"logps/rejected": -11.639761924743652,
|
| 1050 |
+
"loss": 2.3099,
|
| 1051 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1052 |
+
"rewards/chosen": -8.63528060913086,
|
| 1053 |
+
"rewards/margins": 8.82436466217041,
|
| 1054 |
+
"rewards/rejected": -17.459644317626953,
|
| 1055 |
+
"step": 580
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.4276524417867174,
|
| 1059 |
+
"grad_norm": 1.4410984516143799,
|
| 1060 |
+
"learning_rate": 3.3221666168464584e-06,
|
| 1061 |
+
"logits/chosen": -2.506753921508789,
|
| 1062 |
+
"logits/rejected": -3.029228687286377,
|
| 1063 |
+
"logps/chosen": -6.3516435623168945,
|
| 1064 |
+
"logps/rejected": -11.839346885681152,
|
| 1065 |
+
"loss": 2.0248,
|
| 1066 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1067 |
+
"rewards/chosen": -9.527464866638184,
|
| 1068 |
+
"rewards/margins": 8.231555938720703,
|
| 1069 |
+
"rewards/rejected": -17.75901985168457,
|
| 1070 |
+
"step": 590
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.4349007882576787,
|
| 1074 |
+
"grad_norm": 3.3688838481903076,
|
| 1075 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1076 |
+
"logits/chosen": -2.4242749214172363,
|
| 1077 |
+
"logits/rejected": -2.9992825984954834,
|
| 1078 |
+
"logps/chosen": -6.2571916580200195,
|
| 1079 |
+
"logps/rejected": -12.356736183166504,
|
| 1080 |
+
"loss": 2.0191,
|
| 1081 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1082 |
+
"rewards/chosen": -9.385787010192871,
|
| 1083 |
+
"rewards/margins": 9.149316787719727,
|
| 1084 |
+
"rewards/rejected": -18.535104751586914,
|
| 1085 |
+
"step": 600
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.4349007882576787,
|
| 1089 |
+
"eval_logits/chosen": -2.5597894191741943,
|
| 1090 |
+
"eval_logits/rejected": -2.9679033756256104,
|
| 1091 |
+
"eval_logps/chosen": -6.51260232925415,
|
| 1092 |
+
"eval_logps/rejected": -11.510099411010742,
|
| 1093 |
+
"eval_loss": 0.22690872848033905,
|
| 1094 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1095 |
+
"eval_rewards/chosen": -9.768902778625488,
|
| 1096 |
+
"eval_rewards/margins": 7.496245384216309,
|
| 1097 |
+
"eval_rewards/rejected": -17.26515007019043,
|
| 1098 |
+
"eval_runtime": 30.2947,
|
| 1099 |
+
"eval_samples_per_second": 29.444,
|
| 1100 |
+
"eval_steps_per_second": 3.697,
|
| 1101 |
+
"step": 600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.44214913472864004,
|
| 1105 |
+
"grad_norm": 7.5001959800720215,
|
| 1106 |
+
"learning_rate": 3.222579492361179e-06,
|
| 1107 |
+
"logits/chosen": -2.594581127166748,
|
| 1108 |
+
"logits/rejected": -3.0206668376922607,
|
| 1109 |
+
"logps/chosen": -7.0802435874938965,
|
| 1110 |
+
"logps/rejected": -12.667734146118164,
|
| 1111 |
+
"loss": 1.6936,
|
| 1112 |
+
"rewards/accuracies": 0.9375,
|
| 1113 |
+
"rewards/chosen": -10.620367050170898,
|
| 1114 |
+
"rewards/margins": 8.381233215332031,
|
| 1115 |
+
"rewards/rejected": -19.001598358154297,
|
| 1116 |
+
"step": 610
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.4493974811996013,
|
| 1120 |
+
"grad_norm": 2.639087677001953,
|
| 1121 |
+
"learning_rate": 3.1722995515381644e-06,
|
| 1122 |
+
"logits/chosen": -2.511791229248047,
|
| 1123 |
+
"logits/rejected": -3.0638930797576904,
|
| 1124 |
+
"logps/chosen": -7.493863105773926,
|
| 1125 |
+
"logps/rejected": -14.954449653625488,
|
| 1126 |
+
"loss": 1.6215,
|
| 1127 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1128 |
+
"rewards/chosen": -11.24079418182373,
|
| 1129 |
+
"rewards/margins": 11.19088363647461,
|
| 1130 |
+
"rewards/rejected": -22.431676864624023,
|
| 1131 |
+
"step": 620
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.45664582767056267,
|
| 1135 |
+
"grad_norm": 5.2657270431518555,
|
| 1136 |
+
"learning_rate": 3.121724717912138e-06,
|
| 1137 |
+
"logits/chosen": -2.5814051628112793,
|
| 1138 |
+
"logits/rejected": -2.963531017303467,
|
| 1139 |
+
"logps/chosen": -8.267634391784668,
|
| 1140 |
+
"logps/rejected": -14.717096328735352,
|
| 1141 |
+
"loss": 1.7936,
|
| 1142 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1143 |
+
"rewards/chosen": -12.40145206451416,
|
| 1144 |
+
"rewards/margins": 9.6741943359375,
|
| 1145 |
+
"rewards/rejected": -22.075647354125977,
|
| 1146 |
+
"step": 630
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.46389417414152395,
|
| 1150 |
+
"grad_norm": 8.628580093383789,
|
| 1151 |
+
"learning_rate": 3.0708771752766397e-06,
|
| 1152 |
+
"logits/chosen": -2.436100482940674,
|
| 1153 |
+
"logits/rejected": -2.9183783531188965,
|
| 1154 |
+
"logps/chosen": -7.733956336975098,
|
| 1155 |
+
"logps/rejected": -15.915423393249512,
|
| 1156 |
+
"loss": 1.7631,
|
| 1157 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1158 |
+
"rewards/chosen": -11.600934028625488,
|
| 1159 |
+
"rewards/margins": 12.272205352783203,
|
| 1160 |
+
"rewards/rejected": -23.873138427734375,
|
| 1161 |
+
"step": 640
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.4711425206124853,
|
| 1165 |
+
"grad_norm": 3.072305679321289,
|
| 1166 |
+
"learning_rate": 3.019779227044398e-06,
|
| 1167 |
+
"logits/chosen": -2.4763903617858887,
|
| 1168 |
+
"logits/rejected": -2.9726059436798096,
|
| 1169 |
+
"logps/chosen": -6.810072422027588,
|
| 1170 |
+
"logps/rejected": -13.924298286437988,
|
| 1171 |
+
"loss": 1.5531,
|
| 1172 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1173 |
+
"rewards/chosen": -10.215107917785645,
|
| 1174 |
+
"rewards/margins": 10.671339988708496,
|
| 1175 |
+
"rewards/rejected": -20.88644790649414,
|
| 1176 |
+
"step": 650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.4711425206124853,
|
| 1180 |
+
"eval_logits/chosen": -2.459520101547241,
|
| 1181 |
+
"eval_logits/rejected": -2.852581024169922,
|
| 1182 |
+
"eval_logps/chosen": -6.569087028503418,
|
| 1183 |
+
"eval_logps/rejected": -12.969863891601562,
|
| 1184 |
+
"eval_loss": 0.21141496300697327,
|
| 1185 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1186 |
+
"eval_rewards/chosen": -9.853631019592285,
|
| 1187 |
+
"eval_rewards/margins": 9.601165771484375,
|
| 1188 |
+
"eval_rewards/rejected": -19.454797744750977,
|
| 1189 |
+
"eval_runtime": 30.32,
|
| 1190 |
+
"eval_samples_per_second": 29.42,
|
| 1191 |
+
"eval_steps_per_second": 3.694,
|
| 1192 |
+
"step": 650
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.4783908670834466,
|
| 1196 |
+
"grad_norm": 3.4517831802368164,
|
| 1197 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 1198 |
+
"logits/chosen": -2.5354456901550293,
|
| 1199 |
+
"logits/rejected": -2.840172290802002,
|
| 1200 |
+
"logps/chosen": -7.238469123840332,
|
| 1201 |
+
"logps/rejected": -14.787511825561523,
|
| 1202 |
+
"loss": 1.3669,
|
| 1203 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1204 |
+
"rewards/chosen": -10.857705116271973,
|
| 1205 |
+
"rewards/margins": 11.323565483093262,
|
| 1206 |
+
"rewards/rejected": -22.181270599365234,
|
| 1207 |
+
"step": 660
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.4856392135544079,
|
| 1211 |
+
"grad_norm": 8.935027122497559,
|
| 1212 |
+
"learning_rate": 2.9169218667902562e-06,
|
| 1213 |
+
"logits/chosen": -2.47978138923645,
|
| 1214 |
+
"logits/rejected": -2.9299492835998535,
|
| 1215 |
+
"logps/chosen": -7.6814703941345215,
|
| 1216 |
+
"logps/rejected": -15.477340698242188,
|
| 1217 |
+
"loss": 1.8781,
|
| 1218 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1219 |
+
"rewards/chosen": -11.522205352783203,
|
| 1220 |
+
"rewards/margins": 11.693807601928711,
|
| 1221 |
+
"rewards/rejected": -23.216012954711914,
|
| 1222 |
+
"step": 670
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.4928875600253692,
|
| 1226 |
+
"grad_norm": 1.7096459865570068,
|
| 1227 |
+
"learning_rate": 2.8652075714060296e-06,
|
| 1228 |
+
"logits/chosen": -2.4414560794830322,
|
| 1229 |
+
"logits/rejected": -2.864112377166748,
|
| 1230 |
+
"logps/chosen": -8.507112503051758,
|
| 1231 |
+
"logps/rejected": -15.909780502319336,
|
| 1232 |
+
"loss": 1.2831,
|
| 1233 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1234 |
+
"rewards/chosen": -12.76066780090332,
|
| 1235 |
+
"rewards/margins": 11.10400104522705,
|
| 1236 |
+
"rewards/rejected": -23.864669799804688,
|
| 1237 |
+
"step": 680
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.5001359064963306,
|
| 1241 |
+
"grad_norm": 3.814954996109009,
|
| 1242 |
+
"learning_rate": 2.813333083910761e-06,
|
| 1243 |
+
"logits/chosen": -2.4312682151794434,
|
| 1244 |
+
"logits/rejected": -2.993825674057007,
|
| 1245 |
+
"logps/chosen": -8.516946792602539,
|
| 1246 |
+
"logps/rejected": -17.835708618164062,
|
| 1247 |
+
"loss": 1.0306,
|
| 1248 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1249 |
+
"rewards/chosen": -12.775420188903809,
|
| 1250 |
+
"rewards/margins": 13.978144645690918,
|
| 1251 |
+
"rewards/rejected": -26.75356674194336,
|
| 1252 |
+
"step": 690
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.5073842529672918,
|
| 1256 |
+
"grad_norm": 6.554664611816406,
|
| 1257 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1258 |
+
"logits/chosen": -2.458709239959717,
|
| 1259 |
+
"logits/rejected": -2.9223692417144775,
|
| 1260 |
+
"logps/chosen": -9.111552238464355,
|
| 1261 |
+
"logps/rejected": -19.193201065063477,
|
| 1262 |
+
"loss": 1.1941,
|
| 1263 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1264 |
+
"rewards/chosen": -13.667327880859375,
|
| 1265 |
+
"rewards/margins": 15.122472763061523,
|
| 1266 |
+
"rewards/rejected": -28.7898006439209,
|
| 1267 |
+
"step": 700
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 0.5073842529672918,
|
| 1271 |
+
"eval_logits/chosen": -2.5367431640625,
|
| 1272 |
+
"eval_logits/rejected": -2.9102938175201416,
|
| 1273 |
+
"eval_logps/chosen": -10.25200366973877,
|
| 1274 |
+
"eval_logps/rejected": -17.488019943237305,
|
| 1275 |
+
"eval_loss": 0.1846819519996643,
|
| 1276 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1277 |
+
"eval_rewards/chosen": -15.378005027770996,
|
| 1278 |
+
"eval_rewards/margins": 10.854025840759277,
|
| 1279 |
+
"eval_rewards/rejected": -26.232030868530273,
|
| 1280 |
+
"eval_runtime": 30.3487,
|
| 1281 |
+
"eval_samples_per_second": 29.392,
|
| 1282 |
+
"eval_steps_per_second": 3.69,
|
| 1283 |
+
"step": 700
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.5146325994382531,
|
| 1287 |
+
"grad_norm": 11.16737174987793,
|
| 1288 |
+
"learning_rate": 2.70919460833079e-06,
|
| 1289 |
+
"logits/chosen": -2.520266056060791,
|
| 1290 |
+
"logits/rejected": -2.9030609130859375,
|
| 1291 |
+
"logps/chosen": -10.487688064575195,
|
| 1292 |
+
"logps/rejected": -18.03569984436035,
|
| 1293 |
+
"loss": 1.7604,
|
| 1294 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1295 |
+
"rewards/chosen": -15.731532096862793,
|
| 1296 |
+
"rewards/margins": 11.322016716003418,
|
| 1297 |
+
"rewards/rejected": -27.053552627563477,
|
| 1298 |
+
"step": 710
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.5218809459092144,
|
| 1302 |
+
"grad_norm": 8.75973129272461,
|
| 1303 |
+
"learning_rate": 2.6569762988232838e-06,
|
| 1304 |
+
"logits/chosen": -2.427156686782837,
|
| 1305 |
+
"logits/rejected": -2.9073309898376465,
|
| 1306 |
+
"logps/chosen": -8.983491897583008,
|
| 1307 |
+
"logps/rejected": -18.08041000366211,
|
| 1308 |
+
"loss": 1.2509,
|
| 1309 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1310 |
+
"rewards/chosen": -13.475237846374512,
|
| 1311 |
+
"rewards/margins": 13.64537525177002,
|
| 1312 |
+
"rewards/rejected": -27.1206111907959,
|
| 1313 |
+
"step": 720
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.5291292923801758,
|
| 1317 |
+
"grad_norm": 3.2157235145568848,
|
| 1318 |
+
"learning_rate": 2.604689134322999e-06,
|
| 1319 |
+
"logits/chosen": -2.4182469844818115,
|
| 1320 |
+
"logits/rejected": -2.941476821899414,
|
| 1321 |
+
"logps/chosen": -8.248749732971191,
|
| 1322 |
+
"logps/rejected": -16.914398193359375,
|
| 1323 |
+
"loss": 1.2837,
|
| 1324 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1325 |
+
"rewards/chosen": -12.373125076293945,
|
| 1326 |
+
"rewards/margins": 12.99847412109375,
|
| 1327 |
+
"rewards/rejected": -25.371599197387695,
|
| 1328 |
+
"step": 730
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.5363776388511371,
|
| 1332 |
+
"grad_norm": 3.8172707557678223,
|
| 1333 |
+
"learning_rate": 2.5523560497083927e-06,
|
| 1334 |
+
"logits/chosen": -2.4228386878967285,
|
| 1335 |
+
"logits/rejected": -2.9412994384765625,
|
| 1336 |
+
"logps/chosen": -7.712450981140137,
|
| 1337 |
+
"logps/rejected": -16.946636199951172,
|
| 1338 |
+
"loss": 1.2981,
|
| 1339 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1340 |
+
"rewards/chosen": -11.56867790222168,
|
| 1341 |
+
"rewards/margins": 13.851278305053711,
|
| 1342 |
+
"rewards/rejected": -25.41995620727539,
|
| 1343 |
+
"step": 740
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.5436259853220984,
|
| 1347 |
+
"grad_norm": 3.1835663318634033,
|
| 1348 |
+
"learning_rate": 2.5e-06,
|
| 1349 |
+
"logits/chosen": -2.413264036178589,
|
| 1350 |
+
"logits/rejected": -2.8962299823760986,
|
| 1351 |
+
"logps/chosen": -7.1312408447265625,
|
| 1352 |
+
"logps/rejected": -17.016965866088867,
|
| 1353 |
+
"loss": 1.1198,
|
| 1354 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1355 |
+
"rewards/chosen": -10.69686222076416,
|
| 1356 |
+
"rewards/margins": 14.828587532043457,
|
| 1357 |
+
"rewards/rejected": -25.52545166015625,
|
| 1358 |
+
"step": 750
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.5436259853220984,
|
| 1362 |
+
"eval_logits/chosen": -2.482361316680908,
|
| 1363 |
+
"eval_logits/rejected": -2.858593702316284,
|
| 1364 |
+
"eval_logps/chosen": -8.353605270385742,
|
| 1365 |
+
"eval_logps/rejected": -16.015188217163086,
|
| 1366 |
+
"eval_loss": 0.15141016244888306,
|
| 1367 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1368 |
+
"eval_rewards/chosen": -12.530407905578613,
|
| 1369 |
+
"eval_rewards/margins": 11.4923734664917,
|
| 1370 |
+
"eval_rewards/rejected": -24.022781372070312,
|
| 1371 |
+
"eval_runtime": 30.3199,
|
| 1372 |
+
"eval_samples_per_second": 29.42,
|
| 1373 |
+
"eval_steps_per_second": 3.694,
|
| 1374 |
+
"step": 750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.5508743317930597,
|
| 1378 |
+
"grad_norm": 3.896941661834717,
|
| 1379 |
+
"learning_rate": 2.447643950291608e-06,
|
| 1380 |
+
"logits/chosen": -2.4703681468963623,
|
| 1381 |
+
"logits/rejected": -2.8748607635498047,
|
| 1382 |
+
"logps/chosen": -8.088781356811523,
|
| 1383 |
+
"logps/rejected": -16.520170211791992,
|
| 1384 |
+
"loss": 1.1826,
|
| 1385 |
+
"rewards/accuracies": 0.9375,
|
| 1386 |
+
"rewards/chosen": -12.133172035217285,
|
| 1387 |
+
"rewards/margins": 12.647082328796387,
|
| 1388 |
+
"rewards/rejected": -24.780254364013672,
|
| 1389 |
+
"step": 760
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.5581226782640211,
|
| 1393 |
+
"grad_norm": 7.569993495941162,
|
| 1394 |
+
"learning_rate": 2.3953108656770018e-06,
|
| 1395 |
+
"logits/chosen": -2.452991485595703,
|
| 1396 |
+
"logits/rejected": -2.8747894763946533,
|
| 1397 |
+
"logps/chosen": -9.35241413116455,
|
| 1398 |
+
"logps/rejected": -18.470417022705078,
|
| 1399 |
+
"loss": 1.2527,
|
| 1400 |
+
"rewards/accuracies": 1.0,
|
| 1401 |
+
"rewards/chosen": -14.028620719909668,
|
| 1402 |
+
"rewards/margins": 13.67700481414795,
|
| 1403 |
+
"rewards/rejected": -27.70562744140625,
|
| 1404 |
+
"step": 770
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.5653710247349824,
|
| 1408 |
+
"grad_norm": 3.0607998371124268,
|
| 1409 |
+
"learning_rate": 2.3430237011767166e-06,
|
| 1410 |
+
"logits/chosen": -2.4571075439453125,
|
| 1411 |
+
"logits/rejected": -2.8483026027679443,
|
| 1412 |
+
"logps/chosen": -7.992903232574463,
|
| 1413 |
+
"logps/rejected": -17.579792022705078,
|
| 1414 |
+
"loss": 0.7915,
|
| 1415 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1416 |
+
"rewards/chosen": -11.989354133605957,
|
| 1417 |
+
"rewards/margins": 14.380337715148926,
|
| 1418 |
+
"rewards/rejected": -26.36968994140625,
|
| 1419 |
+
"step": 780
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.5726193712059436,
|
| 1423 |
+
"grad_norm": 6.031089782714844,
|
| 1424 |
+
"learning_rate": 2.290805391669212e-06,
|
| 1425 |
+
"logits/chosen": -2.3782265186309814,
|
| 1426 |
+
"logits/rejected": -2.7803268432617188,
|
| 1427 |
+
"logps/chosen": -8.120704650878906,
|
| 1428 |
+
"logps/rejected": -17.844463348388672,
|
| 1429 |
+
"loss": 0.9819,
|
| 1430 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1431 |
+
"rewards/chosen": -12.18105697631836,
|
| 1432 |
+
"rewards/margins": 14.585641860961914,
|
| 1433 |
+
"rewards/rejected": -26.766698837280273,
|
| 1434 |
+
"step": 790
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.5798677176769049,
|
| 1438 |
+
"grad_norm": 4.0991363525390625,
|
| 1439 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1440 |
+
"logits/chosen": -2.3170902729034424,
|
| 1441 |
+
"logits/rejected": -2.82861328125,
|
| 1442 |
+
"logps/chosen": -7.7450361251831055,
|
| 1443 |
+
"logps/rejected": -17.050933837890625,
|
| 1444 |
+
"loss": 1.0712,
|
| 1445 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1446 |
+
"rewards/chosen": -11.6175537109375,
|
| 1447 |
+
"rewards/margins": 13.958849906921387,
|
| 1448 |
+
"rewards/rejected": -25.576404571533203,
|
| 1449 |
+
"step": 800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.5798677176769049,
|
| 1453 |
+
"eval_logits/chosen": -2.3983051776885986,
|
| 1454 |
+
"eval_logits/rejected": -2.7665843963623047,
|
| 1455 |
+
"eval_logps/chosen": -8.15335464477539,
|
| 1456 |
+
"eval_logps/rejected": -16.25174331665039,
|
| 1457 |
+
"eval_loss": 0.12996967136859894,
|
| 1458 |
+
"eval_rewards/accuracies": 0.9464285969734192,
|
| 1459 |
+
"eval_rewards/chosen": -12.230032920837402,
|
| 1460 |
+
"eval_rewards/margins": 12.14758586883545,
|
| 1461 |
+
"eval_rewards/rejected": -24.377620697021484,
|
| 1462 |
+
"eval_runtime": 30.3735,
|
| 1463 |
+
"eval_samples_per_second": 29.368,
|
| 1464 |
+
"eval_steps_per_second": 3.687,
|
| 1465 |
+
"step": 800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.5871160641478663,
|
| 1469 |
+
"grad_norm": 5.751565456390381,
|
| 1470 |
+
"learning_rate": 2.186666916089239e-06,
|
| 1471 |
+
"logits/chosen": -2.3348138332366943,
|
| 1472 |
+
"logits/rejected": -2.769927501678467,
|
| 1473 |
+
"logps/chosen": -8.014350891113281,
|
| 1474 |
+
"logps/rejected": -16.63930892944336,
|
| 1475 |
+
"loss": 1.2619,
|
| 1476 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1477 |
+
"rewards/chosen": -12.021525382995605,
|
| 1478 |
+
"rewards/margins": 12.93743896484375,
|
| 1479 |
+
"rewards/rejected": -24.958965301513672,
|
| 1480 |
+
"step": 810
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.5943644106188276,
|
| 1484 |
+
"grad_norm": 7.957540512084961,
|
| 1485 |
+
"learning_rate": 2.134792428593971e-06,
|
| 1486 |
+
"logits/chosen": -2.327152967453003,
|
| 1487 |
+
"logits/rejected": -2.796149253845215,
|
| 1488 |
+
"logps/chosen": -7.6543426513671875,
|
| 1489 |
+
"logps/rejected": -16.770008087158203,
|
| 1490 |
+
"loss": 1.0105,
|
| 1491 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1492 |
+
"rewards/chosen": -11.481512069702148,
|
| 1493 |
+
"rewards/margins": 13.673500061035156,
|
| 1494 |
+
"rewards/rejected": -25.155010223388672,
|
| 1495 |
+
"step": 820
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.6016127570897889,
|
| 1499 |
+
"grad_norm": 3.2892773151397705,
|
| 1500 |
+
"learning_rate": 2.0830781332097446e-06,
|
| 1501 |
+
"logits/chosen": -2.3687503337860107,
|
| 1502 |
+
"logits/rejected": -2.8093888759613037,
|
| 1503 |
+
"logps/chosen": -9.138994216918945,
|
| 1504 |
+
"logps/rejected": -18.48851776123047,
|
| 1505 |
+
"loss": 0.8295,
|
| 1506 |
+
"rewards/accuracies": 0.9375,
|
| 1507 |
+
"rewards/chosen": -13.708491325378418,
|
| 1508 |
+
"rewards/margins": 14.024286270141602,
|
| 1509 |
+
"rewards/rejected": -27.732776641845703,
|
| 1510 |
+
"step": 830
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 0.6088611035607502,
|
| 1514 |
+
"grad_norm": 3.2489373683929443,
|
| 1515 |
+
"learning_rate": 2.031546713535688e-06,
|
| 1516 |
+
"logits/chosen": -2.454373836517334,
|
| 1517 |
+
"logits/rejected": -2.787046432495117,
|
| 1518 |
+
"logps/chosen": -8.555706977844238,
|
| 1519 |
+
"logps/rejected": -17.804542541503906,
|
| 1520 |
+
"loss": 0.9398,
|
| 1521 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1522 |
+
"rewards/chosen": -12.833561897277832,
|
| 1523 |
+
"rewards/margins": 13.873250961303711,
|
| 1524 |
+
"rewards/rejected": -26.70681381225586,
|
| 1525 |
+
"step": 840
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.6161094500317115,
|
| 1529 |
+
"grad_norm": 4.858041286468506,
|
| 1530 |
+
"learning_rate": 1.9802207729556023e-06,
|
| 1531 |
+
"logits/chosen": -2.3334898948669434,
|
| 1532 |
+
"logits/rejected": -2.799832344055176,
|
| 1533 |
+
"logps/chosen": -8.557367324829102,
|
| 1534 |
+
"logps/rejected": -18.578689575195312,
|
| 1535 |
+
"loss": 0.8919,
|
| 1536 |
+
"rewards/accuracies": 1.0,
|
| 1537 |
+
"rewards/chosen": -12.836050033569336,
|
| 1538 |
+
"rewards/margins": 15.031984329223633,
|
| 1539 |
+
"rewards/rejected": -27.868032455444336,
|
| 1540 |
+
"step": 850
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.6161094500317115,
|
| 1544 |
+
"eval_logits/chosen": -2.394418716430664,
|
| 1545 |
+
"eval_logits/rejected": -2.768528461456299,
|
| 1546 |
+
"eval_logps/chosen": -8.943707466125488,
|
| 1547 |
+
"eval_logps/rejected": -17.889659881591797,
|
| 1548 |
+
"eval_loss": 0.11857374012470245,
|
| 1549 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 1550 |
+
"eval_rewards/chosen": -13.415560722351074,
|
| 1551 |
+
"eval_rewards/margins": 13.418928146362305,
|
| 1552 |
+
"eval_rewards/rejected": -26.834491729736328,
|
| 1553 |
+
"eval_runtime": 30.3214,
|
| 1554 |
+
"eval_samples_per_second": 29.418,
|
| 1555 |
+
"eval_steps_per_second": 3.694,
|
| 1556 |
+
"step": 850
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.6233577965026729,
|
| 1560 |
+
"grad_norm": 6.336078643798828,
|
| 1561 |
+
"learning_rate": 1.9291228247233607e-06,
|
| 1562 |
+
"logits/chosen": -2.3998782634735107,
|
| 1563 |
+
"logits/rejected": -2.8122310638427734,
|
| 1564 |
+
"logps/chosen": -8.910730361938477,
|
| 1565 |
+
"logps/rejected": -18.417646408081055,
|
| 1566 |
+
"loss": 0.9668,
|
| 1567 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1568 |
+
"rewards/chosen": -13.366094589233398,
|
| 1569 |
+
"rewards/margins": 14.26037883758545,
|
| 1570 |
+
"rewards/rejected": -27.626474380493164,
|
| 1571 |
+
"step": 860
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.6306061429736342,
|
| 1575 |
+
"grad_norm": 12.80254077911377,
|
| 1576 |
+
"learning_rate": 1.8782752820878636e-06,
|
| 1577 |
+
"logits/chosen": -2.3551506996154785,
|
| 1578 |
+
"logits/rejected": -2.8078038692474365,
|
| 1579 |
+
"logps/chosen": -7.851692199707031,
|
| 1580 |
+
"logps/rejected": -18.801013946533203,
|
| 1581 |
+
"loss": 0.623,
|
| 1582 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1583 |
+
"rewards/chosen": -11.77753734588623,
|
| 1584 |
+
"rewards/margins": 16.42398452758789,
|
| 1585 |
+
"rewards/rejected": -28.201519012451172,
|
| 1586 |
+
"step": 870
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.6378544894445954,
|
| 1590 |
+
"grad_norm": 3.489668369293213,
|
| 1591 |
+
"learning_rate": 1.827700448461836e-06,
|
| 1592 |
+
"logits/chosen": -2.4149603843688965,
|
| 1593 |
+
"logits/rejected": -2.790621519088745,
|
| 1594 |
+
"logps/chosen": -8.617798805236816,
|
| 1595 |
+
"logps/rejected": -19.069250106811523,
|
| 1596 |
+
"loss": 0.8222,
|
| 1597 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1598 |
+
"rewards/chosen": -12.926698684692383,
|
| 1599 |
+
"rewards/margins": 15.677175521850586,
|
| 1600 |
+
"rewards/rejected": -28.6038761138916,
|
| 1601 |
+
"step": 880
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.6451028359155567,
|
| 1605 |
+
"grad_norm": 1.9104561805725098,
|
| 1606 |
+
"learning_rate": 1.7774205076388207e-06,
|
| 1607 |
+
"logits/chosen": -2.3620147705078125,
|
| 1608 |
+
"logits/rejected": -2.8059632778167725,
|
| 1609 |
+
"logps/chosen": -9.715843200683594,
|
| 1610 |
+
"logps/rejected": -21.32562828063965,
|
| 1611 |
+
"loss": 0.7366,
|
| 1612 |
+
"rewards/accuracies": 1.0,
|
| 1613 |
+
"rewards/chosen": -14.573763847351074,
|
| 1614 |
+
"rewards/margins": 17.414676666259766,
|
| 1615 |
+
"rewards/rejected": -31.98844337463379,
|
| 1616 |
+
"step": 890
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.6523511823865181,
|
| 1620 |
+
"grad_norm": 9.600937843322754,
|
| 1621 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1622 |
+
"logits/chosen": -2.49873948097229,
|
| 1623 |
+
"logits/rejected": -2.825862169265747,
|
| 1624 |
+
"logps/chosen": -10.357137680053711,
|
| 1625 |
+
"logps/rejected": -19.829532623291016,
|
| 1626 |
+
"loss": 1.1936,
|
| 1627 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1628 |
+
"rewards/chosen": -15.53570556640625,
|
| 1629 |
+
"rewards/margins": 14.208595275878906,
|
| 1630 |
+
"rewards/rejected": -29.744298934936523,
|
| 1631 |
+
"step": 900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 0.6523511823865181,
|
| 1635 |
+
"eval_logits/chosen": -2.4111170768737793,
|
| 1636 |
+
"eval_logits/rejected": -2.7982914447784424,
|
| 1637 |
+
"eval_logps/chosen": -10.249353408813477,
|
| 1638 |
+
"eval_logps/rejected": -20.502059936523438,
|
| 1639 |
+
"eval_loss": 0.11673425137996674,
|
| 1640 |
+
"eval_rewards/accuracies": 0.9464285969734192,
|
| 1641 |
+
"eval_rewards/chosen": -15.374032020568848,
|
| 1642 |
+
"eval_rewards/margins": 15.379058837890625,
|
| 1643 |
+
"eval_rewards/rejected": -30.75309181213379,
|
| 1644 |
+
"eval_runtime": 30.3049,
|
| 1645 |
+
"eval_samples_per_second": 29.434,
|
| 1646 |
+
"eval_steps_per_second": 3.696,
|
| 1647 |
+
"step": 900
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.6595995288574794,
|
| 1651 |
+
"grad_norm": 4.28416633605957,
|
| 1652 |
+
"learning_rate": 1.677833383153542e-06,
|
| 1653 |
+
"logits/chosen": -2.348818063735962,
|
| 1654 |
+
"logits/rejected": -2.8150997161865234,
|
| 1655 |
+
"logps/chosen": -9.674580574035645,
|
| 1656 |
+
"logps/rejected": -21.100772857666016,
|
| 1657 |
+
"loss": 0.6817,
|
| 1658 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1659 |
+
"rewards/chosen": -14.511869430541992,
|
| 1660 |
+
"rewards/margins": 17.1392879486084,
|
| 1661 |
+
"rewards/rejected": -31.65115737915039,
|
| 1662 |
+
"step": 910
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.6668478753284407,
|
| 1666 |
+
"grad_norm": 6.507571220397949,
|
| 1667 |
+
"learning_rate": 1.6285698816954626e-06,
|
| 1668 |
+
"logits/chosen": -2.3171777725219727,
|
| 1669 |
+
"logits/rejected": -2.7714550495147705,
|
| 1670 |
+
"logps/chosen": -8.589642524719238,
|
| 1671 |
+
"logps/rejected": -20.672466278076172,
|
| 1672 |
+
"loss": 0.8751,
|
| 1673 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1674 |
+
"rewards/chosen": -12.8844633102417,
|
| 1675 |
+
"rewards/margins": 18.124235153198242,
|
| 1676 |
+
"rewards/rejected": -31.008697509765625,
|
| 1677 |
+
"step": 920
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.674096221799402,
|
| 1681 |
+
"grad_norm": 5.07767915725708,
|
| 1682 |
+
"learning_rate": 1.5796886182883053e-06,
|
| 1683 |
+
"logits/chosen": -2.374326705932617,
|
| 1684 |
+
"logits/rejected": -2.7454559803009033,
|
| 1685 |
+
"logps/chosen": -9.055803298950195,
|
| 1686 |
+
"logps/rejected": -21.033706665039062,
|
| 1687 |
+
"loss": 0.8639,
|
| 1688 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1689 |
+
"rewards/chosen": -13.583703994750977,
|
| 1690 |
+
"rewards/margins": 17.966854095458984,
|
| 1691 |
+
"rewards/rejected": -31.550561904907227,
|
| 1692 |
+
"step": 930
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 0.6813445682703633,
|
| 1696 |
+
"grad_norm": 4.192989349365234,
|
| 1697 |
+
"learning_rate": 1.5312110338697427e-06,
|
| 1698 |
+
"logits/chosen": -2.4058878421783447,
|
| 1699 |
+
"logits/rejected": -2.81569766998291,
|
| 1700 |
+
"logps/chosen": -10.162097930908203,
|
| 1701 |
+
"logps/rejected": -21.776790618896484,
|
| 1702 |
+
"loss": 0.8294,
|
| 1703 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1704 |
+
"rewards/chosen": -15.243146896362305,
|
| 1705 |
+
"rewards/margins": 17.422039031982422,
|
| 1706 |
+
"rewards/rejected": -32.665184020996094,
|
| 1707 |
+
"step": 940
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.6885929147413247,
|
| 1711 |
+
"grad_norm": 6.396577835083008,
|
| 1712 |
+
"learning_rate": 1.4831583923105e-06,
|
| 1713 |
+
"logits/chosen": -2.4091625213623047,
|
| 1714 |
+
"logits/rejected": -2.7940189838409424,
|
| 1715 |
+
"logps/chosen": -9.696913719177246,
|
| 1716 |
+
"logps/rejected": -20.56863784790039,
|
| 1717 |
+
"loss": 0.9113,
|
| 1718 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1719 |
+
"rewards/chosen": -14.545370101928711,
|
| 1720 |
+
"rewards/margins": 16.30758285522461,
|
| 1721 |
+
"rewards/rejected": -30.852951049804688,
|
| 1722 |
+
"step": 950
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.6885929147413247,
|
| 1726 |
+
"eval_logits/chosen": -2.371931314468384,
|
| 1727 |
+
"eval_logits/rejected": -2.7571141719818115,
|
| 1728 |
+
"eval_logps/chosen": -8.616922378540039,
|
| 1729 |
+
"eval_logps/rejected": -18.889474868774414,
|
| 1730 |
+
"eval_loss": 0.1089157834649086,
|
| 1731 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 1732 |
+
"eval_rewards/chosen": -12.925383567810059,
|
| 1733 |
+
"eval_rewards/margins": 15.40882682800293,
|
| 1734 |
+
"eval_rewards/rejected": -28.334211349487305,
|
| 1735 |
+
"eval_runtime": 30.3205,
|
| 1736 |
+
"eval_samples_per_second": 29.419,
|
| 1737 |
+
"eval_steps_per_second": 3.694,
|
| 1738 |
+
"step": 950
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.695841261212286,
|
| 1742 |
+
"grad_norm": 2.0531630516052246,
|
| 1743 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 1744 |
+
"logits/chosen": -2.3074121475219727,
|
| 1745 |
+
"logits/rejected": -2.796274423599243,
|
| 1746 |
+
"logps/chosen": -8.068899154663086,
|
| 1747 |
+
"logps/rejected": -18.71575164794922,
|
| 1748 |
+
"loss": 0.8595,
|
| 1749 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1750 |
+
"rewards/chosen": -12.103349685668945,
|
| 1751 |
+
"rewards/margins": 15.9702787399292,
|
| 1752 |
+
"rewards/rejected": -28.07362937927246,
|
| 1753 |
+
"step": 960
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.7030896076832472,
|
| 1757 |
+
"grad_norm": 8.175732612609863,
|
| 1758 |
+
"learning_rate": 1.388412052037682e-06,
|
| 1759 |
+
"logits/chosen": -2.3601155281066895,
|
| 1760 |
+
"logits/rejected": -2.8169026374816895,
|
| 1761 |
+
"logps/chosen": -9.363846778869629,
|
| 1762 |
+
"logps/rejected": -21.162973403930664,
|
| 1763 |
+
"loss": 0.9415,
|
| 1764 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1765 |
+
"rewards/chosen": -14.045768737792969,
|
| 1766 |
+
"rewards/margins": 17.69869041442871,
|
| 1767 |
+
"rewards/rejected": -31.744457244873047,
|
| 1768 |
+
"step": 970
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.7103379541542085,
|
| 1772 |
+
"grad_norm": 21.960830688476562,
|
| 1773 |
+
"learning_rate": 1.3417599122003464e-06,
|
| 1774 |
+
"logits/chosen": -2.2582755088806152,
|
| 1775 |
+
"logits/rejected": -2.8120741844177246,
|
| 1776 |
+
"logps/chosen": -8.260191917419434,
|
| 1777 |
+
"logps/rejected": -20.000268936157227,
|
| 1778 |
+
"loss": 0.8961,
|
| 1779 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1780 |
+
"rewards/chosen": -12.390287399291992,
|
| 1781 |
+
"rewards/margins": 17.610116958618164,
|
| 1782 |
+
"rewards/rejected": -30.000402450561523,
|
| 1783 |
+
"step": 980
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 0.7175863006251699,
|
| 1787 |
+
"grad_norm": 4.283840179443359,
|
| 1788 |
+
"learning_rate": 1.2956158147457116e-06,
|
| 1789 |
+
"logits/chosen": -2.3035526275634766,
|
| 1790 |
+
"logits/rejected": -2.7238144874572754,
|
| 1791 |
+
"logps/chosen": -7.883841037750244,
|
| 1792 |
+
"logps/rejected": -19.16309928894043,
|
| 1793 |
+
"loss": 0.7882,
|
| 1794 |
+
"rewards/accuracies": 1.0,
|
| 1795 |
+
"rewards/chosen": -11.825760841369629,
|
| 1796 |
+
"rewards/margins": 16.91888427734375,
|
| 1797 |
+
"rewards/rejected": -28.744644165039062,
|
| 1798 |
+
"step": 990
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 0.7248346470961312,
|
| 1802 |
+
"grad_norm": 17.20541763305664,
|
| 1803 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1804 |
+
"logits/chosen": -2.3649654388427734,
|
| 1805 |
+
"logits/rejected": -2.760070323944092,
|
| 1806 |
+
"logps/chosen": -9.843507766723633,
|
| 1807 |
+
"logps/rejected": -20.48518943786621,
|
| 1808 |
+
"loss": 1.135,
|
| 1809 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1810 |
+
"rewards/chosen": -14.765260696411133,
|
| 1811 |
+
"rewards/margins": 15.96252155303955,
|
| 1812 |
+
"rewards/rejected": -30.727783203125,
|
| 1813 |
+
"step": 1000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.7248346470961312,
|
| 1817 |
+
"eval_logits/chosen": -2.390676736831665,
|
| 1818 |
+
"eval_logits/rejected": -2.7747442722320557,
|
| 1819 |
+
"eval_logps/chosen": -9.749165534973145,
|
| 1820 |
+
"eval_logps/rejected": -20.03530502319336,
|
| 1821 |
+
"eval_loss": 0.1021483913064003,
|
| 1822 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 1823 |
+
"eval_rewards/chosen": -14.623747825622559,
|
| 1824 |
+
"eval_rewards/margins": 15.42920970916748,
|
| 1825 |
+
"eval_rewards/rejected": -30.05295753479004,
|
| 1826 |
+
"eval_runtime": 30.3161,
|
| 1827 |
+
"eval_samples_per_second": 29.423,
|
| 1828 |
+
"eval_steps_per_second": 3.694,
|
| 1829 |
+
"step": 1000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.7320829935670925,
|
| 1833 |
+
"grad_norm": 5.572669506072998,
|
| 1834 |
+
"learning_rate": 1.204932476567175e-06,
|
| 1835 |
+
"logits/chosen": -2.300497055053711,
|
| 1836 |
+
"logits/rejected": -2.7727770805358887,
|
| 1837 |
+
"logps/chosen": -9.08503532409668,
|
| 1838 |
+
"logps/rejected": -21.310386657714844,
|
| 1839 |
+
"loss": 0.769,
|
| 1840 |
+
"rewards/accuracies": 1.0,
|
| 1841 |
+
"rewards/chosen": -13.62755298614502,
|
| 1842 |
+
"rewards/margins": 18.338027954101562,
|
| 1843 |
+
"rewards/rejected": -31.965579986572266,
|
| 1844 |
+
"step": 1010
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.7393313400380538,
|
| 1848 |
+
"grad_norm": 1.9562855958938599,
|
| 1849 |
+
"learning_rate": 1.160433012552508e-06,
|
| 1850 |
+
"logits/chosen": -2.3115997314453125,
|
| 1851 |
+
"logits/rejected": -2.8770854473114014,
|
| 1852 |
+
"logps/chosen": -8.05511474609375,
|
| 1853 |
+
"logps/rejected": -21.068984985351562,
|
| 1854 |
+
"loss": 0.5362,
|
| 1855 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1856 |
+
"rewards/chosen": -12.082670211791992,
|
| 1857 |
+
"rewards/margins": 19.52080726623535,
|
| 1858 |
+
"rewards/rejected": -31.60347557067871,
|
| 1859 |
+
"step": 1020
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.7465796865090152,
|
| 1863 |
+
"grad_norm": 6.376579761505127,
|
| 1864 |
+
"learning_rate": 1.11652112689164e-06,
|
| 1865 |
+
"logits/chosen": -2.349559783935547,
|
| 1866 |
+
"logits/rejected": -2.8144681453704834,
|
| 1867 |
+
"logps/chosen": -8.122370719909668,
|
| 1868 |
+
"logps/rejected": -19.65790557861328,
|
| 1869 |
+
"loss": 0.8026,
|
| 1870 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1871 |
+
"rewards/chosen": -12.183557510375977,
|
| 1872 |
+
"rewards/margins": 17.303298950195312,
|
| 1873 |
+
"rewards/rejected": -29.48685646057129,
|
| 1874 |
+
"step": 1030
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.7538280329799765,
|
| 1878 |
+
"grad_norm": 9.438390731811523,
|
| 1879 |
+
"learning_rate": 1.073216080788921e-06,
|
| 1880 |
+
"logits/chosen": -2.3181960582733154,
|
| 1881 |
+
"logits/rejected": -2.7197091579437256,
|
| 1882 |
+
"logps/chosen": -7.880677700042725,
|
| 1883 |
+
"logps/rejected": -19.085203170776367,
|
| 1884 |
+
"loss": 0.7519,
|
| 1885 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1886 |
+
"rewards/chosen": -11.821017265319824,
|
| 1887 |
+
"rewards/margins": 16.806785583496094,
|
| 1888 |
+
"rewards/rejected": -28.6278018951416,
|
| 1889 |
+
"step": 1040
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.7610763794509378,
|
| 1893 |
+
"grad_norm": 5.646973609924316,
|
| 1894 |
+
"learning_rate": 1.0305368692688175e-06,
|
| 1895 |
+
"logits/chosen": -2.271108865737915,
|
| 1896 |
+
"logits/rejected": -2.7761292457580566,
|
| 1897 |
+
"logps/chosen": -7.734799861907959,
|
| 1898 |
+
"logps/rejected": -19.717693328857422,
|
| 1899 |
+
"loss": 0.5528,
|
| 1900 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1901 |
+
"rewards/chosen": -11.602200508117676,
|
| 1902 |
+
"rewards/margins": 17.97433853149414,
|
| 1903 |
+
"rewards/rejected": -29.5765380859375,
|
| 1904 |
+
"step": 1050
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.7610763794509378,
|
| 1908 |
+
"eval_logits/chosen": -2.3761844635009766,
|
| 1909 |
+
"eval_logits/rejected": -2.758427381515503,
|
| 1910 |
+
"eval_logps/chosen": -9.39671802520752,
|
| 1911 |
+
"eval_logps/rejected": -19.70557975769043,
|
| 1912 |
+
"eval_loss": 0.1007753536105156,
|
| 1913 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 1914 |
+
"eval_rewards/chosen": -14.095077514648438,
|
| 1915 |
+
"eval_rewards/margins": 15.463290214538574,
|
| 1916 |
+
"eval_rewards/rejected": -29.55837059020996,
|
| 1917 |
+
"eval_runtime": 30.3239,
|
| 1918 |
+
"eval_samples_per_second": 29.416,
|
| 1919 |
+
"eval_steps_per_second": 3.693,
|
| 1920 |
+
"step": 1050
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.768324725921899,
|
| 1924 |
+
"grad_norm": 3.293710947036743,
|
| 1925 |
+
"learning_rate": 9.88502212844063e-07,
|
| 1926 |
+
"logits/chosen": -2.3491718769073486,
|
| 1927 |
+
"logits/rejected": -2.7819125652313232,
|
| 1928 |
+
"logps/chosen": -8.037856101989746,
|
| 1929 |
+
"logps/rejected": -19.4016170501709,
|
| 1930 |
+
"loss": 0.792,
|
| 1931 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1932 |
+
"rewards/chosen": -12.056783676147461,
|
| 1933 |
+
"rewards/margins": 17.04564094543457,
|
| 1934 |
+
"rewards/rejected": -29.1024227142334,
|
| 1935 |
+
"step": 1060
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.7755730723928603,
|
| 1939 |
+
"grad_norm": 3.9602127075195312,
|
| 1940 |
+
"learning_rate": 9.471305493042243e-07,
|
| 1941 |
+
"logits/chosen": -2.236292600631714,
|
| 1942 |
+
"logits/rejected": -2.6696109771728516,
|
| 1943 |
+
"logps/chosen": -7.1002068519592285,
|
| 1944 |
+
"logps/rejected": -19.54210090637207,
|
| 1945 |
+
"loss": 0.5324,
|
| 1946 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1947 |
+
"rewards/chosen": -10.650310516357422,
|
| 1948 |
+
"rewards/margins": 18.662839889526367,
|
| 1949 |
+
"rewards/rejected": -29.31315040588379,
|
| 1950 |
+
"step": 1070
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 0.7828214188638217,
|
| 1954 |
+
"grad_norm": 2.300848960876465,
|
| 1955 |
+
"learning_rate": 9.064400256282757e-07,
|
| 1956 |
+
"logits/chosen": -2.3093066215515137,
|
| 1957 |
+
"logits/rejected": -2.6977267265319824,
|
| 1958 |
+
"logps/chosen": -8.959060668945312,
|
| 1959 |
+
"logps/rejected": -18.99527359008789,
|
| 1960 |
+
"loss": 0.5667,
|
| 1961 |
+
"rewards/accuracies": 1.0,
|
| 1962 |
+
"rewards/chosen": -13.438589096069336,
|
| 1963 |
+
"rewards/margins": 15.05431842803955,
|
| 1964 |
+
"rewards/rejected": -28.492910385131836,
|
| 1965 |
+
"step": 1080
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.790069765334783,
|
| 1969 |
+
"grad_norm": 1.5451587438583374,
|
| 1970 |
+
"learning_rate": 8.664484900247363e-07,
|
| 1971 |
+
"logits/chosen": -2.414778232574463,
|
| 1972 |
+
"logits/rejected": -2.750126838684082,
|
| 1973 |
+
"logps/chosen": -9.486712455749512,
|
| 1974 |
+
"logps/rejected": -20.353275299072266,
|
| 1975 |
+
"loss": 0.6379,
|
| 1976 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1977 |
+
"rewards/chosen": -14.230069160461426,
|
| 1978 |
+
"rewards/margins": 16.299842834472656,
|
| 1979 |
+
"rewards/rejected": -30.529911041259766,
|
| 1980 |
+
"step": 1090
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 0.7973181118057443,
|
| 1984 |
+
"grad_norm": 7.686610221862793,
|
| 1985 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1986 |
+
"logits/chosen": -2.383478879928589,
|
| 1987 |
+
"logits/rejected": -2.6949422359466553,
|
| 1988 |
+
"logps/chosen": -9.218855857849121,
|
| 1989 |
+
"logps/rejected": -20.50551414489746,
|
| 1990 |
+
"loss": 0.6618,
|
| 1991 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1992 |
+
"rewards/chosen": -13.828283309936523,
|
| 1993 |
+
"rewards/margins": 16.929988861083984,
|
| 1994 |
+
"rewards/rejected": -30.758275985717773,
|
| 1995 |
+
"step": 1100
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 0.7973181118057443,
|
| 1999 |
+
"eval_logits/chosen": -2.357304811477661,
|
| 2000 |
+
"eval_logits/rejected": -2.741511821746826,
|
| 2001 |
+
"eval_logps/chosen": -8.644713401794434,
|
| 2002 |
+
"eval_logps/rejected": -19.651016235351562,
|
| 2003 |
+
"eval_loss": 0.0923290029168129,
|
| 2004 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 2005 |
+
"eval_rewards/chosen": -12.967069625854492,
|
| 2006 |
+
"eval_rewards/margins": 16.50945472717285,
|
| 2007 |
+
"eval_rewards/rejected": -29.476526260375977,
|
| 2008 |
+
"eval_runtime": 30.3348,
|
| 2009 |
+
"eval_samples_per_second": 29.405,
|
| 2010 |
+
"eval_steps_per_second": 3.692,
|
| 2011 |
+
"step": 1100
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.8045664582767056,
|
| 2015 |
+
"grad_norm": 2.6959259510040283,
|
| 2016 |
+
"learning_rate": 7.886322351782782e-07,
|
| 2017 |
+
"logits/chosen": -2.300200939178467,
|
| 2018 |
+
"logits/rejected": -2.759124994277954,
|
| 2019 |
+
"logps/chosen": -8.86776351928711,
|
| 2020 |
+
"logps/rejected": -21.157636642456055,
|
| 2021 |
+
"loss": 0.5303,
|
| 2022 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2023 |
+
"rewards/chosen": -13.301645278930664,
|
| 2024 |
+
"rewards/margins": 18.434810638427734,
|
| 2025 |
+
"rewards/rejected": -31.736459732055664,
|
| 2026 |
+
"step": 1110
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.811814804747667,
|
| 2030 |
+
"grad_norm": 1.7385252714157104,
|
| 2031 |
+
"learning_rate": 7.508416487165862e-07,
|
| 2032 |
+
"logits/chosen": -2.2611403465270996,
|
| 2033 |
+
"logits/rejected": -2.7048158645629883,
|
| 2034 |
+
"logps/chosen": -8.681452751159668,
|
| 2035 |
+
"logps/rejected": -20.63593101501465,
|
| 2036 |
+
"loss": 0.9835,
|
| 2037 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2038 |
+
"rewards/chosen": -13.022178649902344,
|
| 2039 |
+
"rewards/margins": 17.931718826293945,
|
| 2040 |
+
"rewards/rejected": -30.95389747619629,
|
| 2041 |
+
"step": 1120
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 0.8190631512186283,
|
| 2045 |
+
"grad_norm": 31.765613555908203,
|
| 2046 |
+
"learning_rate": 7.138183009179922e-07,
|
| 2047 |
+
"logits/chosen": -2.347527027130127,
|
| 2048 |
+
"logits/rejected": -2.8267199993133545,
|
| 2049 |
+
"logps/chosen": -7.8143181800842285,
|
| 2050 |
+
"logps/rejected": -21.345165252685547,
|
| 2051 |
+
"loss": 0.9257,
|
| 2052 |
+
"rewards/accuracies": 1.0,
|
| 2053 |
+
"rewards/chosen": -11.721475601196289,
|
| 2054 |
+
"rewards/margins": 20.296268463134766,
|
| 2055 |
+
"rewards/rejected": -32.01774597167969,
|
| 2056 |
+
"step": 1130
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 0.8263114976895896,
|
| 2060 |
+
"grad_norm": 2.3969268798828125,
|
| 2061 |
+
"learning_rate": 6.775784314464717e-07,
|
| 2062 |
+
"logits/chosen": -2.298384428024292,
|
| 2063 |
+
"logits/rejected": -2.7226641178131104,
|
| 2064 |
+
"logps/chosen": -8.508520126342773,
|
| 2065 |
+
"logps/rejected": -21.121414184570312,
|
| 2066 |
+
"loss": 0.5027,
|
| 2067 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2068 |
+
"rewards/chosen": -12.762779235839844,
|
| 2069 |
+
"rewards/margins": 18.919343948364258,
|
| 2070 |
+
"rewards/rejected": -31.6821231842041,
|
| 2071 |
+
"step": 1140
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 0.8335598441605508,
|
| 2075 |
+
"grad_norm": 1.3444470167160034,
|
| 2076 |
+
"learning_rate": 6.421379363065142e-07,
|
| 2077 |
+
"logits/chosen": -2.274933338165283,
|
| 2078 |
+
"logits/rejected": -2.871828317642212,
|
| 2079 |
+
"logps/chosen": -8.347254753112793,
|
| 2080 |
+
"logps/rejected": -22.654220581054688,
|
| 2081 |
+
"loss": 0.7009,
|
| 2082 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2083 |
+
"rewards/chosen": -12.520882606506348,
|
| 2084 |
+
"rewards/margins": 21.46044921875,
|
| 2085 |
+
"rewards/rejected": -33.98133087158203,
|
| 2086 |
+
"step": 1150
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 0.8335598441605508,
|
| 2090 |
+
"eval_logits/chosen": -2.3756191730499268,
|
| 2091 |
+
"eval_logits/rejected": -2.7599036693573,
|
| 2092 |
+
"eval_logps/chosen": -9.69633960723877,
|
| 2093 |
+
"eval_logps/rejected": -20.757549285888672,
|
| 2094 |
+
"eval_loss": 0.09489800035953522,
|
| 2095 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 2096 |
+
"eval_rewards/chosen": -14.544509887695312,
|
| 2097 |
+
"eval_rewards/margins": 16.591815948486328,
|
| 2098 |
+
"eval_rewards/rejected": -31.136323928833008,
|
| 2099 |
+
"eval_runtime": 30.3194,
|
| 2100 |
+
"eval_samples_per_second": 29.42,
|
| 2101 |
+
"eval_steps_per_second": 3.694,
|
| 2102 |
+
"step": 1150
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.8408081906315121,
|
| 2106 |
+
"grad_norm": 3.957143783569336,
|
| 2107 |
+
"learning_rate": 6.075123608706093e-07,
|
| 2108 |
+
"logits/chosen": -2.2645936012268066,
|
| 2109 |
+
"logits/rejected": -2.7766642570495605,
|
| 2110 |
+
"logps/chosen": -9.488225936889648,
|
| 2111 |
+
"logps/rejected": -20.858362197875977,
|
| 2112 |
+
"loss": 0.9432,
|
| 2113 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2114 |
+
"rewards/chosen": -14.232337951660156,
|
| 2115 |
+
"rewards/margins": 17.05520248413086,
|
| 2116 |
+
"rewards/rejected": -31.287546157836914,
|
| 2117 |
+
"step": 1160
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.8480565371024735,
|
| 2121 |
+
"grad_norm": 0.2772391140460968,
|
| 2122 |
+
"learning_rate": 5.737168930605272e-07,
|
| 2123 |
+
"logits/chosen": -2.322751760482788,
|
| 2124 |
+
"logits/rejected": -2.762363910675049,
|
| 2125 |
+
"logps/chosen": -8.993582725524902,
|
| 2126 |
+
"logps/rejected": -22.878589630126953,
|
| 2127 |
+
"loss": 0.5761,
|
| 2128 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2129 |
+
"rewards/chosen": -13.490374565124512,
|
| 2130 |
+
"rewards/margins": 20.8275089263916,
|
| 2131 |
+
"rewards/rejected": -34.3178825378418,
|
| 2132 |
+
"step": 1170
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.8553048835734348,
|
| 2136 |
+
"grad_norm": 1.3643966913223267,
|
| 2137 |
+
"learning_rate": 5.407663566854008e-07,
|
| 2138 |
+
"logits/chosen": -2.365164279937744,
|
| 2139 |
+
"logits/rejected": -2.7776386737823486,
|
| 2140 |
+
"logps/chosen": -8.55947208404541,
|
| 2141 |
+
"logps/rejected": -21.43111801147461,
|
| 2142 |
+
"loss": 0.8883,
|
| 2143 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2144 |
+
"rewards/chosen": -12.839208602905273,
|
| 2145 |
+
"rewards/margins": 19.30746841430664,
|
| 2146 |
+
"rewards/rejected": -32.14667892456055,
|
| 2147 |
+
"step": 1180
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 0.8625532300443961,
|
| 2151 |
+
"grad_norm": 5.047004222869873,
|
| 2152 |
+
"learning_rate": 5.086752049395094e-07,
|
| 2153 |
+
"logits/chosen": -2.245209217071533,
|
| 2154 |
+
"logits/rejected": -2.725189447402954,
|
| 2155 |
+
"logps/chosen": -8.62891674041748,
|
| 2156 |
+
"logps/rejected": -22.020822525024414,
|
| 2157 |
+
"loss": 1.1177,
|
| 2158 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2159 |
+
"rewards/chosen": -12.943374633789062,
|
| 2160 |
+
"rewards/margins": 20.087858200073242,
|
| 2161 |
+
"rewards/rejected": -33.0312385559082,
|
| 2162 |
+
"step": 1190
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.8698015765153574,
|
| 2166 |
+
"grad_norm": 5.6105451583862305,
|
| 2167 |
+
"learning_rate": 4.774575140626317e-07,
|
| 2168 |
+
"logits/chosen": -2.2240893840789795,
|
| 2169 |
+
"logits/rejected": -2.685760498046875,
|
| 2170 |
+
"logps/chosen": -7.7927656173706055,
|
| 2171 |
+
"logps/rejected": -21.007144927978516,
|
| 2172 |
+
"loss": 0.4369,
|
| 2173 |
+
"rewards/accuracies": 1.0,
|
| 2174 |
+
"rewards/chosen": -11.68914794921875,
|
| 2175 |
+
"rewards/margins": 19.821565628051758,
|
| 2176 |
+
"rewards/rejected": -31.51071548461914,
|
| 2177 |
+
"step": 1200
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 0.8698015765153574,
|
| 2181 |
+
"eval_logits/chosen": -2.3455567359924316,
|
| 2182 |
+
"eval_logits/rejected": -2.725982189178467,
|
| 2183 |
+
"eval_logps/chosen": -8.64130973815918,
|
| 2184 |
+
"eval_logps/rejected": -19.878324508666992,
|
| 2185 |
+
"eval_loss": 0.08837977796792984,
|
| 2186 |
+
"eval_rewards/accuracies": 0.9553571343421936,
|
| 2187 |
+
"eval_rewards/chosen": -12.96196460723877,
|
| 2188 |
+
"eval_rewards/margins": 16.855520248413086,
|
| 2189 |
+
"eval_rewards/rejected": -29.817485809326172,
|
| 2190 |
+
"eval_runtime": 30.3327,
|
| 2191 |
+
"eval_samples_per_second": 29.407,
|
| 2192 |
+
"eval_steps_per_second": 3.692,
|
| 2193 |
+
"step": 1200
|
| 2194 |
+
}
|
| 2195 |
+
],
|
| 2196 |
+
"logging_steps": 10,
|
| 2197 |
+
"max_steps": 1500,
|
| 2198 |
+
"num_input_tokens_seen": 0,
|
| 2199 |
+
"num_train_epochs": 2,
|
| 2200 |
+
"save_steps": 50,
|
| 2201 |
+
"stateful_callbacks": {
|
| 2202 |
+
"TrainerControl": {
|
| 2203 |
+
"args": {
|
| 2204 |
+
"should_epoch_stop": false,
|
| 2205 |
+
"should_evaluate": false,
|
| 2206 |
+
"should_log": false,
|
| 2207 |
+
"should_save": true,
|
| 2208 |
+
"should_training_stop": false
|
| 2209 |
+
},
|
| 2210 |
+
"attributes": {}
|
| 2211 |
+
}
|
| 2212 |
+
},
|
| 2213 |
+
"total_flos": 4.462855621980455e+18,
|
| 2214 |
+
"train_batch_size": 1,
|
| 2215 |
+
"trial_name": null,
|
| 2216 |
+
"trial_params": null
|
| 2217 |
+
}
|
checkpoint-1200/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91b9d0e6e8c92ae4a2c435b81fedf969ba2fc02907285b004770a8256166d71d
|
| 3 |
+
size 7224
|
checkpoint-1200/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)
|