Training in progress, step 1200, checkpoint
Browse files- checkpoint-1200/README.md +202 -0
- checkpoint-1200/adapter_config.json +34 -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 +133 -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: microsoft/Phi-3-mini-4k-instruct
|
| 3 |
+
library_name: peft
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Model Card for Model ID
|
| 7 |
+
|
| 8 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
## Model Details
|
| 13 |
+
|
| 14 |
+
### Model Description
|
| 15 |
+
|
| 16 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
- **Developed by:** [More Information Needed]
|
| 21 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 22 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 23 |
+
- **Model type:** [More Information Needed]
|
| 24 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 25 |
+
- **License:** [More Information Needed]
|
| 26 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 27 |
+
|
| 28 |
+
### Model Sources [optional]
|
| 29 |
+
|
| 30 |
+
<!-- Provide the basic links for the model. -->
|
| 31 |
+
|
| 32 |
+
- **Repository:** [More Information Needed]
|
| 33 |
+
- **Paper [optional]:** [More Information Needed]
|
| 34 |
+
- **Demo [optional]:** [More Information Needed]
|
| 35 |
+
|
| 36 |
+
## Uses
|
| 37 |
+
|
| 38 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 39 |
+
|
| 40 |
+
### Direct Use
|
| 41 |
+
|
| 42 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 43 |
+
|
| 44 |
+
[More Information Needed]
|
| 45 |
+
|
| 46 |
+
### Downstream Use [optional]
|
| 47 |
+
|
| 48 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 49 |
+
|
| 50 |
+
[More Information Needed]
|
| 51 |
+
|
| 52 |
+
### Out-of-Scope Use
|
| 53 |
+
|
| 54 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 55 |
+
|
| 56 |
+
[More Information Needed]
|
| 57 |
+
|
| 58 |
+
## Bias, Risks, and Limitations
|
| 59 |
+
|
| 60 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 61 |
+
|
| 62 |
+
[More Information Needed]
|
| 63 |
+
|
| 64 |
+
### Recommendations
|
| 65 |
+
|
| 66 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 67 |
+
|
| 68 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 69 |
+
|
| 70 |
+
## How to Get Started with the Model
|
| 71 |
+
|
| 72 |
+
Use the code below to get started with the model.
|
| 73 |
+
|
| 74 |
+
[More Information Needed]
|
| 75 |
+
|
| 76 |
+
## Training Details
|
| 77 |
+
|
| 78 |
+
### Training Data
|
| 79 |
+
|
| 80 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 81 |
+
|
| 82 |
+
[More Information Needed]
|
| 83 |
+
|
| 84 |
+
### Training Procedure
|
| 85 |
+
|
| 86 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 87 |
+
|
| 88 |
+
#### Preprocessing [optional]
|
| 89 |
+
|
| 90 |
+
[More Information Needed]
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
#### Training Hyperparameters
|
| 94 |
+
|
| 95 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 96 |
+
|
| 97 |
+
#### Speeds, Sizes, Times [optional]
|
| 98 |
+
|
| 99 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 100 |
+
|
| 101 |
+
[More Information Needed]
|
| 102 |
+
|
| 103 |
+
## Evaluation
|
| 104 |
+
|
| 105 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 106 |
+
|
| 107 |
+
### Testing Data, Factors & Metrics
|
| 108 |
+
|
| 109 |
+
#### Testing Data
|
| 110 |
+
|
| 111 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 112 |
+
|
| 113 |
+
[More Information Needed]
|
| 114 |
+
|
| 115 |
+
#### Factors
|
| 116 |
+
|
| 117 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 118 |
+
|
| 119 |
+
[More Information Needed]
|
| 120 |
+
|
| 121 |
+
#### Metrics
|
| 122 |
+
|
| 123 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 124 |
+
|
| 125 |
+
[More Information Needed]
|
| 126 |
+
|
| 127 |
+
### Results
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
#### Summary
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
|
| 135 |
+
## Model Examination [optional]
|
| 136 |
+
|
| 137 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 138 |
+
|
| 139 |
+
[More Information Needed]
|
| 140 |
+
|
| 141 |
+
## Environmental Impact
|
| 142 |
+
|
| 143 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 144 |
+
|
| 145 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 146 |
+
|
| 147 |
+
- **Hardware Type:** [More Information Needed]
|
| 148 |
+
- **Hours used:** [More Information Needed]
|
| 149 |
+
- **Cloud Provider:** [More Information Needed]
|
| 150 |
+
- **Compute Region:** [More Information Needed]
|
| 151 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 152 |
+
|
| 153 |
+
## Technical Specifications [optional]
|
| 154 |
+
|
| 155 |
+
### Model Architecture and Objective
|
| 156 |
+
|
| 157 |
+
[More Information Needed]
|
| 158 |
+
|
| 159 |
+
### Compute Infrastructure
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
#### Hardware
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Software
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
## Citation [optional]
|
| 172 |
+
|
| 173 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 174 |
+
|
| 175 |
+
**BibTeX:**
|
| 176 |
+
|
| 177 |
+
[More Information Needed]
|
| 178 |
+
|
| 179 |
+
**APA:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
## Glossary [optional]
|
| 184 |
+
|
| 185 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 186 |
+
|
| 187 |
+
[More Information Needed]
|
| 188 |
+
|
| 189 |
+
## More Information [optional]
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## Model Card Authors [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Contact
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
### Framework versions
|
| 201 |
+
|
| 202 |
+
- PEFT 0.14.0
|
checkpoint-1200/adapter_config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "microsoft/Phi-3-mini-4k-instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"eva_config": null,
|
| 7 |
+
"exclude_modules": null,
|
| 8 |
+
"fan_in_fan_out": false,
|
| 9 |
+
"inference_mode": true,
|
| 10 |
+
"init_lora_weights": true,
|
| 11 |
+
"layer_replication": null,
|
| 12 |
+
"layers_pattern": null,
|
| 13 |
+
"layers_to_transform": null,
|
| 14 |
+
"loftq_config": {},
|
| 15 |
+
"lora_alpha": 16,
|
| 16 |
+
"lora_bias": false,
|
| 17 |
+
"lora_dropout": 0.0,
|
| 18 |
+
"megatron_config": null,
|
| 19 |
+
"megatron_core": "megatron.core",
|
| 20 |
+
"modules_to_save": null,
|
| 21 |
+
"peft_type": "LORA",
|
| 22 |
+
"r": 8,
|
| 23 |
+
"rank_pattern": {},
|
| 24 |
+
"revision": null,
|
| 25 |
+
"target_modules": [
|
| 26 |
+
"gate_up_proj",
|
| 27 |
+
"qkv_proj",
|
| 28 |
+
"down_proj",
|
| 29 |
+
"o_proj"
|
| 30 |
+
],
|
| 31 |
+
"task_type": "CAUSAL_LM",
|
| 32 |
+
"use_dora": false,
|
| 33 |
+
"use_rslora": false
|
| 34 |
+
}
|
checkpoint-1200/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3868f947ddbd884fa52df246ab09caf38e5bff7707dc9336fb9d29402fef5710
|
| 3 |
+
size 25200088
|
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 sha256:7ca4e6ec6b523530a5129ee684258cc3b1bc22cdad1a1ec0246502bc70a6e8f4
|
| 3 |
+
size 18881328
|
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 sha256:5930c106cf0d4abe66c7ab3b26e08af850cfef524dd6f3c04f4bf3adcfc7de6f
|
| 3 |
+
size 18881328
|
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 sha256:09cd8931cb8e1818c7e2aff35742e74c537bcf2fac36e8c3720c8c90a2e2510c
|
| 3 |
+
size 18881328
|
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:abdcc49a2eba4b55d30ff64cce513887c947d59b877106e1c629572ea17b1342
|
| 3 |
+
size 18881392
|
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 sha256:62c5cf5a4ab9cdb7e2ce528f122513943c7cd86134e14c12f7799f1b7a41769d
|
| 3 |
+
size 18881392
|
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 sha256:4b5f9a615ea024b459b4b0159f1f9c1b53731ffe8e84104dd7e0b4ca79b7524f
|
| 3 |
+
size 18881392
|
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 sha256:1133f3f0a95c45bf00ce44e607a5ca168eda7babdbdcfd6957a8b7902c00ba6a
|
| 3 |
+
size 18881392
|
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:a1503bd23f22d57cd38fd500a89c56022e3f9ee21407722012e9e57c0bc63778
|
| 3 |
+
size 18881392
|
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:a31dfd8198f2a8aa5d1c323bc43f9cc3a7f925c5a7ee0d439df42dc1297326d0
|
| 3 |
+
size 25379244
|
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 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|end|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "<|endoftext|>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
checkpoint-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 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": true,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": false
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "<|endoftext|>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
},
|
| 38 |
+
"32001": {
|
| 39 |
+
"content": "<|assistant|>",
|
| 40 |
+
"lstrip": false,
|
| 41 |
+
"normalized": false,
|
| 42 |
+
"rstrip": true,
|
| 43 |
+
"single_word": false,
|
| 44 |
+
"special": true
|
| 45 |
+
},
|
| 46 |
+
"32002": {
|
| 47 |
+
"content": "<|placeholder1|>",
|
| 48 |
+
"lstrip": false,
|
| 49 |
+
"normalized": false,
|
| 50 |
+
"rstrip": true,
|
| 51 |
+
"single_word": false,
|
| 52 |
+
"special": true
|
| 53 |
+
},
|
| 54 |
+
"32003": {
|
| 55 |
+
"content": "<|placeholder2|>",
|
| 56 |
+
"lstrip": false,
|
| 57 |
+
"normalized": false,
|
| 58 |
+
"rstrip": true,
|
| 59 |
+
"single_word": false,
|
| 60 |
+
"special": true
|
| 61 |
+
},
|
| 62 |
+
"32004": {
|
| 63 |
+
"content": "<|placeholder3|>",
|
| 64 |
+
"lstrip": false,
|
| 65 |
+
"normalized": false,
|
| 66 |
+
"rstrip": true,
|
| 67 |
+
"single_word": false,
|
| 68 |
+
"special": true
|
| 69 |
+
},
|
| 70 |
+
"32005": {
|
| 71 |
+
"content": "<|placeholder4|>",
|
| 72 |
+
"lstrip": false,
|
| 73 |
+
"normalized": false,
|
| 74 |
+
"rstrip": true,
|
| 75 |
+
"single_word": false,
|
| 76 |
+
"special": true
|
| 77 |
+
},
|
| 78 |
+
"32006": {
|
| 79 |
+
"content": "<|system|>",
|
| 80 |
+
"lstrip": false,
|
| 81 |
+
"normalized": false,
|
| 82 |
+
"rstrip": true,
|
| 83 |
+
"single_word": false,
|
| 84 |
+
"special": true
|
| 85 |
+
},
|
| 86 |
+
"32007": {
|
| 87 |
+
"content": "<|end|>",
|
| 88 |
+
"lstrip": false,
|
| 89 |
+
"normalized": false,
|
| 90 |
+
"rstrip": false,
|
| 91 |
+
"single_word": false,
|
| 92 |
+
"special": true
|
| 93 |
+
},
|
| 94 |
+
"32008": {
|
| 95 |
+
"content": "<|placeholder5|>",
|
| 96 |
+
"lstrip": false,
|
| 97 |
+
"normalized": false,
|
| 98 |
+
"rstrip": true,
|
| 99 |
+
"single_word": false,
|
| 100 |
+
"special": true
|
| 101 |
+
},
|
| 102 |
+
"32009": {
|
| 103 |
+
"content": "<|placeholder6|>",
|
| 104 |
+
"lstrip": false,
|
| 105 |
+
"normalized": false,
|
| 106 |
+
"rstrip": true,
|
| 107 |
+
"single_word": false,
|
| 108 |
+
"special": true
|
| 109 |
+
},
|
| 110 |
+
"32010": {
|
| 111 |
+
"content": "<|user|>",
|
| 112 |
+
"lstrip": false,
|
| 113 |
+
"normalized": false,
|
| 114 |
+
"rstrip": true,
|
| 115 |
+
"single_word": false,
|
| 116 |
+
"special": true
|
| 117 |
+
}
|
| 118 |
+
},
|
| 119 |
+
"bos_token": "<s>",
|
| 120 |
+
"chat_template": "{% set system_message = 'You are a helpful AI assistant.' %}{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '<s>' + '<|system|>\n' + system_message + '<|end|>\n' }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|user|>\n' + content + '<|end|>\n<|assistant|>\n' }}{% elif message['role'] == 'assistant' %}{{ content + '<|end|>' + '\n' }}{% endif %}{% endfor %}",
|
| 121 |
+
"clean_up_tokenization_spaces": false,
|
| 122 |
+
"eos_token": "<|end|>",
|
| 123 |
+
"extra_special_tokens": {},
|
| 124 |
+
"legacy": false,
|
| 125 |
+
"model_max_length": 4096,
|
| 126 |
+
"pad_token": "<|endoftext|>",
|
| 127 |
+
"padding_side": "right",
|
| 128 |
+
"sp_model_kwargs": {},
|
| 129 |
+
"split_special_tokens": false,
|
| 130 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 131 |
+
"unk_token": "<unk>",
|
| 132 |
+
"use_default_system_prompt": false
|
| 133 |
+
}
|
checkpoint-1200/trainer_state.json
ADDED
|
@@ -0,0 +1,2217 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.91324200913242,
|
| 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": 0.0076103500761035,
|
| 13 |
+
"grad_norm": 0.058339186012744904,
|
| 14 |
+
"learning_rate": 4.999451708687114e-06,
|
| 15 |
+
"logits/chosen": 14.268467903137207,
|
| 16 |
+
"logits/rejected": 14.600369453430176,
|
| 17 |
+
"logps/chosen": -0.2669850289821625,
|
| 18 |
+
"logps/rejected": -0.3412467837333679,
|
| 19 |
+
"loss": 0.9049,
|
| 20 |
+
"rewards/accuracies": 0.5625,
|
| 21 |
+
"rewards/chosen": -0.4004775583744049,
|
| 22 |
+
"rewards/margins": 0.11139259487390518,
|
| 23 |
+
"rewards/rejected": -0.5118702054023743,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.015220700152207,
|
| 28 |
+
"grad_norm": 0.049545690417289734,
|
| 29 |
+
"learning_rate": 4.997807075247147e-06,
|
| 30 |
+
"logits/chosen": 14.14539623260498,
|
| 31 |
+
"logits/rejected": 15.191584587097168,
|
| 32 |
+
"logps/chosen": -0.25579872727394104,
|
| 33 |
+
"logps/rejected": -0.3931494653224945,
|
| 34 |
+
"loss": 0.8989,
|
| 35 |
+
"rewards/accuracies": 0.625,
|
| 36 |
+
"rewards/chosen": -0.38369807600975037,
|
| 37 |
+
"rewards/margins": 0.2060261219739914,
|
| 38 |
+
"rewards/rejected": -0.5897241830825806,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0228310502283105,
|
| 43 |
+
"grad_norm": 0.061699289828538895,
|
| 44 |
+
"learning_rate": 4.9950668210706795e-06,
|
| 45 |
+
"logits/chosen": 14.284139633178711,
|
| 46 |
+
"logits/rejected": 15.006326675415039,
|
| 47 |
+
"logps/chosen": -0.275672048330307,
|
| 48 |
+
"logps/rejected": -0.3603581488132477,
|
| 49 |
+
"loss": 0.9004,
|
| 50 |
+
"rewards/accuracies": 0.625,
|
| 51 |
+
"rewards/chosen": -0.4135080873966217,
|
| 52 |
+
"rewards/margins": 0.12702910602092743,
|
| 53 |
+
"rewards/rejected": -0.5405372381210327,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.030441400304414,
|
| 58 |
+
"grad_norm": 0.05706426501274109,
|
| 59 |
+
"learning_rate": 4.9912321481237616e-06,
|
| 60 |
+
"logits/chosen": 14.275796890258789,
|
| 61 |
+
"logits/rejected": 14.935521125793457,
|
| 62 |
+
"logps/chosen": -0.2802076041698456,
|
| 63 |
+
"logps/rejected": -0.38278770446777344,
|
| 64 |
+
"loss": 0.9138,
|
| 65 |
+
"rewards/accuracies": 0.512499988079071,
|
| 66 |
+
"rewards/chosen": -0.42031145095825195,
|
| 67 |
+
"rewards/margins": 0.15387018024921417,
|
| 68 |
+
"rewards/rejected": -0.5741815567016602,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0380517503805175,
|
| 73 |
+
"grad_norm": 0.05318514257669449,
|
| 74 |
+
"learning_rate": 4.986304738420684e-06,
|
| 75 |
+
"logits/chosen": 14.433627128601074,
|
| 76 |
+
"logits/rejected": 15.458297729492188,
|
| 77 |
+
"logps/chosen": -0.2581387162208557,
|
| 78 |
+
"logps/rejected": -0.38208404183387756,
|
| 79 |
+
"loss": 0.914,
|
| 80 |
+
"rewards/accuracies": 0.612500011920929,
|
| 81 |
+
"rewards/chosen": -0.38720807433128357,
|
| 82 |
+
"rewards/margins": 0.18591801822185516,
|
| 83 |
+
"rewards/rejected": -0.5731261372566223,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.0380517503805175,
|
| 88 |
+
"eval_logits/chosen": 14.396967887878418,
|
| 89 |
+
"eval_logits/rejected": 15.221076965332031,
|
| 90 |
+
"eval_logps/chosen": -0.27519574761390686,
|
| 91 |
+
"eval_logps/rejected": -0.3709692656993866,
|
| 92 |
+
"eval_loss": 0.9084128141403198,
|
| 93 |
+
"eval_rewards/accuracies": 0.5981308221817017,
|
| 94 |
+
"eval_rewards/chosen": -0.4127936065196991,
|
| 95 |
+
"eval_rewards/margins": 0.14366032183170319,
|
| 96 |
+
"eval_rewards/rejected": -0.5564539432525635,
|
| 97 |
+
"eval_runtime": 30.773,
|
| 98 |
+
"eval_samples_per_second": 27.622,
|
| 99 |
+
"eval_steps_per_second": 3.477,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.045662100456621,
|
| 104 |
+
"grad_norm": 0.06310460716485977,
|
| 105 |
+
"learning_rate": 4.980286753286196e-06,
|
| 106 |
+
"logits/chosen": 14.548416137695312,
|
| 107 |
+
"logits/rejected": 15.526041030883789,
|
| 108 |
+
"logps/chosen": -0.29403647780418396,
|
| 109 |
+
"logps/rejected": -0.40682005882263184,
|
| 110 |
+
"loss": 0.9082,
|
| 111 |
+
"rewards/accuracies": 0.625,
|
| 112 |
+
"rewards/chosen": -0.44105473160743713,
|
| 113 |
+
"rewards/margins": 0.1691754311323166,
|
| 114 |
+
"rewards/rejected": -0.6102300882339478,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0532724505327245,
|
| 119 |
+
"grad_norm": 0.1258806735277176,
|
| 120 |
+
"learning_rate": 4.973180832407471e-06,
|
| 121 |
+
"logits/chosen": 14.390210151672363,
|
| 122 |
+
"logits/rejected": 14.817584037780762,
|
| 123 |
+
"logps/chosen": -0.25258123874664307,
|
| 124 |
+
"logps/rejected": -0.36392712593078613,
|
| 125 |
+
"loss": 0.896,
|
| 126 |
+
"rewards/accuracies": 0.637499988079071,
|
| 127 |
+
"rewards/chosen": -0.3788718581199646,
|
| 128 |
+
"rewards/margins": 0.1670188158750534,
|
| 129 |
+
"rewards/rejected": -0.5458906888961792,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.060882800608828,
|
| 134 |
+
"grad_norm": 0.09006265550851822,
|
| 135 |
+
"learning_rate": 4.964990092676263e-06,
|
| 136 |
+
"logits/chosen": 13.844560623168945,
|
| 137 |
+
"logits/rejected": 14.811120986938477,
|
| 138 |
+
"logps/chosen": -0.2630843222141266,
|
| 139 |
+
"logps/rejected": -0.3794577717781067,
|
| 140 |
+
"loss": 0.8977,
|
| 141 |
+
"rewards/accuracies": 0.6875,
|
| 142 |
+
"rewards/chosen": -0.3946264684200287,
|
| 143 |
+
"rewards/margins": 0.17456015944480896,
|
| 144 |
+
"rewards/rejected": -0.5691865682601929,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.0684931506849315,
|
| 149 |
+
"grad_norm": 0.07123688608407974,
|
| 150 |
+
"learning_rate": 4.9557181268217225e-06,
|
| 151 |
+
"logits/chosen": 13.927327156066895,
|
| 152 |
+
"logits/rejected": 14.746416091918945,
|
| 153 |
+
"logps/chosen": -0.25282323360443115,
|
| 154 |
+
"logps/rejected": -0.3279832601547241,
|
| 155 |
+
"loss": 0.9092,
|
| 156 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 157 |
+
"rewards/chosen": -0.37923485040664673,
|
| 158 |
+
"rewards/margins": 0.11274002492427826,
|
| 159 |
+
"rewards/rejected": -0.4919748902320862,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.076103500761035,
|
| 164 |
+
"grad_norm": 0.08333446085453033,
|
| 165 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 166 |
+
"logits/chosen": 14.406118392944336,
|
| 167 |
+
"logits/rejected": 14.770090103149414,
|
| 168 |
+
"logps/chosen": -0.28569403290748596,
|
| 169 |
+
"logps/rejected": -0.3596845269203186,
|
| 170 |
+
"loss": 0.8932,
|
| 171 |
+
"rewards/accuracies": 0.5625,
|
| 172 |
+
"rewards/chosen": -0.42854103446006775,
|
| 173 |
+
"rewards/margins": 0.11098580062389374,
|
| 174 |
+
"rewards/rejected": -0.5395268201828003,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.076103500761035,
|
| 179 |
+
"eval_logits/chosen": 13.925265312194824,
|
| 180 |
+
"eval_logits/rejected": 14.808513641357422,
|
| 181 |
+
"eval_logps/chosen": -0.2667020559310913,
|
| 182 |
+
"eval_logps/rejected": -0.3739235997200012,
|
| 183 |
+
"eval_loss": 0.8984279036521912,
|
| 184 |
+
"eval_rewards/accuracies": 0.5981308221817017,
|
| 185 |
+
"eval_rewards/chosen": -0.40005311369895935,
|
| 186 |
+
"eval_rewards/margins": 0.16083234548568726,
|
| 187 |
+
"eval_rewards/rejected": -0.5608854293823242,
|
| 188 |
+
"eval_runtime": 30.7791,
|
| 189 |
+
"eval_samples_per_second": 27.616,
|
| 190 |
+
"eval_steps_per_second": 3.476,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.0837138508371385,
|
| 195 |
+
"grad_norm": 0.08474570512771606,
|
| 196 |
+
"learning_rate": 4.933947257182901e-06,
|
| 197 |
+
"logits/chosen": 13.641456604003906,
|
| 198 |
+
"logits/rejected": 14.799921035766602,
|
| 199 |
+
"logps/chosen": -0.2721528708934784,
|
| 200 |
+
"logps/rejected": -0.38378894329071045,
|
| 201 |
+
"loss": 0.8995,
|
| 202 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 203 |
+
"rewards/chosen": -0.40822935104370117,
|
| 204 |
+
"rewards/margins": 0.1674540936946869,
|
| 205 |
+
"rewards/rejected": -0.5756834149360657,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.091324200913242,
|
| 210 |
+
"grad_norm": 0.1004580408334732,
|
| 211 |
+
"learning_rate": 4.921457902821578e-06,
|
| 212 |
+
"logits/chosen": 13.835454940795898,
|
| 213 |
+
"logits/rejected": 14.882522583007812,
|
| 214 |
+
"logps/chosen": -0.28507837653160095,
|
| 215 |
+
"logps/rejected": -0.39737468957901,
|
| 216 |
+
"loss": 0.8795,
|
| 217 |
+
"rewards/accuracies": 0.574999988079071,
|
| 218 |
+
"rewards/chosen": -0.42761754989624023,
|
| 219 |
+
"rewards/margins": 0.16844449937343597,
|
| 220 |
+
"rewards/rejected": -0.5960620641708374,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.0989345509893455,
|
| 225 |
+
"grad_norm": 0.09537151455879211,
|
| 226 |
+
"learning_rate": 4.907906416994146e-06,
|
| 227 |
+
"logits/chosen": 13.607874870300293,
|
| 228 |
+
"logits/rejected": 14.091131210327148,
|
| 229 |
+
"logps/chosen": -0.2739318013191223,
|
| 230 |
+
"logps/rejected": -0.36800479888916016,
|
| 231 |
+
"loss": 0.8912,
|
| 232 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 233 |
+
"rewards/chosen": -0.4108976721763611,
|
| 234 |
+
"rewards/margins": 0.14110951125621796,
|
| 235 |
+
"rewards/rejected": -0.5520071983337402,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.106544901065449,
|
| 240 |
+
"grad_norm": 0.10281535238027573,
|
| 241 |
+
"learning_rate": 4.893298743830168e-06,
|
| 242 |
+
"logits/chosen": 12.017224311828613,
|
| 243 |
+
"logits/rejected": 13.04835319519043,
|
| 244 |
+
"logps/chosen": -0.24072685837745667,
|
| 245 |
+
"logps/rejected": -0.36906492710113525,
|
| 246 |
+
"loss": 0.8908,
|
| 247 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 248 |
+
"rewards/chosen": -0.3610902428627014,
|
| 249 |
+
"rewards/margins": 0.19250717759132385,
|
| 250 |
+
"rewards/rejected": -0.5535974502563477,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.1141552511415525,
|
| 255 |
+
"grad_norm": 0.707987368106842,
|
| 256 |
+
"learning_rate": 4.8776412907378845e-06,
|
| 257 |
+
"logits/chosen": 12.522550582885742,
|
| 258 |
+
"logits/rejected": 13.272679328918457,
|
| 259 |
+
"logps/chosen": -0.2583540081977844,
|
| 260 |
+
"logps/rejected": -0.3796755075454712,
|
| 261 |
+
"loss": 0.8867,
|
| 262 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 263 |
+
"rewards/chosen": -0.38753098249435425,
|
| 264 |
+
"rewards/margins": 0.18198221921920776,
|
| 265 |
+
"rewards/rejected": -0.569513201713562,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.1141552511415525,
|
| 270 |
+
"eval_logits/chosen": 11.989100456237793,
|
| 271 |
+
"eval_logits/rejected": 12.92872142791748,
|
| 272 |
+
"eval_logps/chosen": -0.27158522605895996,
|
| 273 |
+
"eval_logps/rejected": -0.40521273016929626,
|
| 274 |
+
"eval_loss": 0.8765817284584045,
|
| 275 |
+
"eval_rewards/accuracies": 0.5981308221817017,
|
| 276 |
+
"eval_rewards/chosen": -0.40737783908843994,
|
| 277 |
+
"eval_rewards/margins": 0.20044119656085968,
|
| 278 |
+
"eval_rewards/rejected": -0.6078190803527832,
|
| 279 |
+
"eval_runtime": 30.7739,
|
| 280 |
+
"eval_samples_per_second": 27.621,
|
| 281 |
+
"eval_steps_per_second": 3.477,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.121765601217656,
|
| 286 |
+
"grad_norm": 0.19342070817947388,
|
| 287 |
+
"learning_rate": 4.860940925593703e-06,
|
| 288 |
+
"logits/chosen": 11.095940589904785,
|
| 289 |
+
"logits/rejected": 12.351040840148926,
|
| 290 |
+
"logps/chosen": -0.24749942123889923,
|
| 291 |
+
"logps/rejected": -0.43422192335128784,
|
| 292 |
+
"loss": 0.8762,
|
| 293 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 294 |
+
"rewards/chosen": -0.37124913930892944,
|
| 295 |
+
"rewards/margins": 0.2800838053226471,
|
| 296 |
+
"rewards/rejected": -0.6513329744338989,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.1293759512937595,
|
| 301 |
+
"grad_norm": 0.19374576210975647,
|
| 302 |
+
"learning_rate": 4.84320497372973e-06,
|
| 303 |
+
"logits/chosen": 10.510068893432617,
|
| 304 |
+
"logits/rejected": 11.507593154907227,
|
| 305 |
+
"logps/chosen": -0.26223134994506836,
|
| 306 |
+
"logps/rejected": -0.43635931611061096,
|
| 307 |
+
"loss": 0.8581,
|
| 308 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 309 |
+
"rewards/chosen": -0.39334696531295776,
|
| 310 |
+
"rewards/margins": 0.2611919641494751,
|
| 311 |
+
"rewards/rejected": -0.6545389294624329,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.136986301369863,
|
| 316 |
+
"grad_norm": 0.20330430567264557,
|
| 317 |
+
"learning_rate": 4.824441214720629e-06,
|
| 318 |
+
"logits/chosen": 9.89570140838623,
|
| 319 |
+
"logits/rejected": 10.669364929199219,
|
| 320 |
+
"logps/chosen": -0.3143860101699829,
|
| 321 |
+
"logps/rejected": -0.46989941596984863,
|
| 322 |
+
"loss": 0.8558,
|
| 323 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 324 |
+
"rewards/chosen": -0.47157901525497437,
|
| 325 |
+
"rewards/margins": 0.23327013850212097,
|
| 326 |
+
"rewards/rejected": -0.704849123954773,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.1445966514459665,
|
| 331 |
+
"grad_norm": 0.22942212224006653,
|
| 332 |
+
"learning_rate": 4.804657878971252e-06,
|
| 333 |
+
"logits/chosen": 8.887057304382324,
|
| 334 |
+
"logits/rejected": 9.542157173156738,
|
| 335 |
+
"logps/chosen": -0.2906036972999573,
|
| 336 |
+
"logps/rejected": -0.4810206890106201,
|
| 337 |
+
"loss": 0.8554,
|
| 338 |
+
"rewards/accuracies": 0.625,
|
| 339 |
+
"rewards/chosen": -0.4359055459499359,
|
| 340 |
+
"rewards/margins": 0.28562551736831665,
|
| 341 |
+
"rewards/rejected": -0.7215310335159302,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.15220700152207,
|
| 346 |
+
"grad_norm": 0.29071903228759766,
|
| 347 |
+
"learning_rate": 4.783863644106502e-06,
|
| 348 |
+
"logits/chosen": 6.791537284851074,
|
| 349 |
+
"logits/rejected": 7.366445064544678,
|
| 350 |
+
"logps/chosen": -0.31382033228874207,
|
| 351 |
+
"logps/rejected": -0.5417486429214478,
|
| 352 |
+
"loss": 0.838,
|
| 353 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 354 |
+
"rewards/chosen": -0.4707304835319519,
|
| 355 |
+
"rewards/margins": 0.34189245104789734,
|
| 356 |
+
"rewards/rejected": -0.8126228451728821,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.15220700152207,
|
| 361 |
+
"eval_logits/chosen": 7.050150394439697,
|
| 362 |
+
"eval_logits/rejected": 7.516275405883789,
|
| 363 |
+
"eval_logps/chosen": -0.3289315402507782,
|
| 364 |
+
"eval_logps/rejected": -0.5481724143028259,
|
| 365 |
+
"eval_loss": 0.813983678817749,
|
| 366 |
+
"eval_rewards/accuracies": 0.6168224215507507,
|
| 367 |
+
"eval_rewards/chosen": -0.4933973252773285,
|
| 368 |
+
"eval_rewards/margins": 0.3288613557815552,
|
| 369 |
+
"eval_rewards/rejected": -0.8222586512565613,
|
| 370 |
+
"eval_runtime": 30.7734,
|
| 371 |
+
"eval_samples_per_second": 27.621,
|
| 372 |
+
"eval_steps_per_second": 3.477,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.1598173515981735,
|
| 377 |
+
"grad_norm": 0.23101097345352173,
|
| 378 |
+
"learning_rate": 4.762067631165049e-06,
|
| 379 |
+
"logits/chosen": 5.132790565490723,
|
| 380 |
+
"logits/rejected": 5.848537445068359,
|
| 381 |
+
"logps/chosen": -0.33372369408607483,
|
| 382 |
+
"logps/rejected": -0.5993582010269165,
|
| 383 |
+
"loss": 0.8212,
|
| 384 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 385 |
+
"rewards/chosen": -0.5005855560302734,
|
| 386 |
+
"rewards/margins": 0.3984517455101013,
|
| 387 |
+
"rewards/rejected": -0.8990373611450195,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.167427701674277,
|
| 392 |
+
"grad_norm": 0.5136363506317139,
|
| 393 |
+
"learning_rate": 4.7392794005985324e-06,
|
| 394 |
+
"logits/chosen": 3.807554244995117,
|
| 395 |
+
"logits/rejected": 4.600871562957764,
|
| 396 |
+
"logps/chosen": -0.32092416286468506,
|
| 397 |
+
"logps/rejected": -0.651642918586731,
|
| 398 |
+
"loss": 0.7851,
|
| 399 |
+
"rewards/accuracies": 0.699999988079071,
|
| 400 |
+
"rewards/chosen": -0.4813862442970276,
|
| 401 |
+
"rewards/margins": 0.4960783123970032,
|
| 402 |
+
"rewards/rejected": -0.977464497089386,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.1750380517503805,
|
| 407 |
+
"grad_norm": 0.4106898009777069,
|
| 408 |
+
"learning_rate": 4.715508948078037e-06,
|
| 409 |
+
"logits/chosen": 2.760650396347046,
|
| 410 |
+
"logits/rejected": 2.1608071327209473,
|
| 411 |
+
"logps/chosen": -0.43665003776550293,
|
| 412 |
+
"logps/rejected": -0.8352751731872559,
|
| 413 |
+
"loss": 0.7685,
|
| 414 |
+
"rewards/accuracies": 0.6875,
|
| 415 |
+
"rewards/chosen": -0.6549750566482544,
|
| 416 |
+
"rewards/margins": 0.5979377627372742,
|
| 417 |
+
"rewards/rejected": -1.2529128789901733,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.182648401826484,
|
| 422 |
+
"grad_norm": 0.4719419479370117,
|
| 423 |
+
"learning_rate": 4.690766700109659e-06,
|
| 424 |
+
"logits/chosen": 3.1216347217559814,
|
| 425 |
+
"logits/rejected": 2.7202537059783936,
|
| 426 |
+
"logps/chosen": -0.444007933139801,
|
| 427 |
+
"logps/rejected": -0.7697597742080688,
|
| 428 |
+
"loss": 0.7474,
|
| 429 |
+
"rewards/accuracies": 0.6875,
|
| 430 |
+
"rewards/chosen": -0.6660118699073792,
|
| 431 |
+
"rewards/margins": 0.4886276125907898,
|
| 432 |
+
"rewards/rejected": -1.154639482498169,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.1902587519025875,
|
| 437 |
+
"grad_norm": 0.548523485660553,
|
| 438 |
+
"learning_rate": 4.665063509461098e-06,
|
| 439 |
+
"logits/chosen": 1.3678622245788574,
|
| 440 |
+
"logits/rejected": 0.46835970878601074,
|
| 441 |
+
"logps/chosen": -0.48227253556251526,
|
| 442 |
+
"logps/rejected": -0.997289776802063,
|
| 443 |
+
"loss": 0.7017,
|
| 444 |
+
"rewards/accuracies": 0.6875,
|
| 445 |
+
"rewards/chosen": -0.7234088182449341,
|
| 446 |
+
"rewards/margins": 0.7725256681442261,
|
| 447 |
+
"rewards/rejected": -1.4959346055984497,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.1902587519025875,
|
| 452 |
+
"eval_logits/chosen": 2.1362831592559814,
|
| 453 |
+
"eval_logits/rejected": 1.1932121515274048,
|
| 454 |
+
"eval_logps/chosen": -0.500978946685791,
|
| 455 |
+
"eval_logps/rejected": -1.0073517560958862,
|
| 456 |
+
"eval_loss": 0.6914573907852173,
|
| 457 |
+
"eval_rewards/accuracies": 0.6542056202888489,
|
| 458 |
+
"eval_rewards/chosen": -0.7514683604240417,
|
| 459 |
+
"eval_rewards/margins": 0.7595593929290771,
|
| 460 |
+
"eval_rewards/rejected": -1.5110276937484741,
|
| 461 |
+
"eval_runtime": 30.7706,
|
| 462 |
+
"eval_samples_per_second": 27.624,
|
| 463 |
+
"eval_steps_per_second": 3.477,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.197869101978691,
|
| 468 |
+
"grad_norm": 0.700670063495636,
|
| 469 |
+
"learning_rate": 4.638410650401267e-06,
|
| 470 |
+
"logits/chosen": 2.537666082382202,
|
| 471 |
+
"logits/rejected": 1.3070740699768066,
|
| 472 |
+
"logps/chosen": -0.59038907289505,
|
| 473 |
+
"logps/rejected": -1.0600087642669678,
|
| 474 |
+
"loss": 0.6908,
|
| 475 |
+
"rewards/accuracies": 0.637499988079071,
|
| 476 |
+
"rewards/chosen": -0.8855836987495422,
|
| 477 |
+
"rewards/margins": 0.7044296264648438,
|
| 478 |
+
"rewards/rejected": -1.5900132656097412,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2054794520547945,
|
| 483 |
+
"grad_norm": 0.6454456448554993,
|
| 484 |
+
"learning_rate": 4.610819813755038e-06,
|
| 485 |
+
"logits/chosen": 2.312289237976074,
|
| 486 |
+
"logits/rejected": 1.6705052852630615,
|
| 487 |
+
"logps/chosen": -0.601074755191803,
|
| 488 |
+
"logps/rejected": -1.12887442111969,
|
| 489 |
+
"loss": 0.6868,
|
| 490 |
+
"rewards/accuracies": 0.637499988079071,
|
| 491 |
+
"rewards/chosen": -0.9016121029853821,
|
| 492 |
+
"rewards/margins": 0.7916995286941528,
|
| 493 |
+
"rewards/rejected": -1.6933116912841797,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.213089802130898,
|
| 498 |
+
"grad_norm": 0.8001136183738708,
|
| 499 |
+
"learning_rate": 4.582303101775249e-06,
|
| 500 |
+
"logits/chosen": 1.6213299036026,
|
| 501 |
+
"logits/rejected": 0.9048928022384644,
|
| 502 |
+
"logps/chosen": -0.6731385588645935,
|
| 503 |
+
"logps/rejected": -1.3181935548782349,
|
| 504 |
+
"loss": 0.632,
|
| 505 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 506 |
+
"rewards/chosen": -1.0097079277038574,
|
| 507 |
+
"rewards/margins": 0.9675822257995605,
|
| 508 |
+
"rewards/rejected": -1.977290153503418,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.2207001522070015,
|
| 513 |
+
"grad_norm": 0.45858490467071533,
|
| 514 |
+
"learning_rate": 4.55287302283426e-06,
|
| 515 |
+
"logits/chosen": 1.0463030338287354,
|
| 516 |
+
"logits/rejected": 0.05798797681927681,
|
| 517 |
+
"logps/chosen": -0.677167534828186,
|
| 518 |
+
"logps/rejected": -1.4764039516448975,
|
| 519 |
+
"loss": 0.6447,
|
| 520 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 521 |
+
"rewards/chosen": -1.0157512426376343,
|
| 522 |
+
"rewards/margins": 1.1988548040390015,
|
| 523 |
+
"rewards/rejected": -2.2146058082580566,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.228310502283105,
|
| 528 |
+
"grad_norm": 0.5778977870941162,
|
| 529 |
+
"learning_rate": 4.522542485937369e-06,
|
| 530 |
+
"logits/chosen": 2.3259291648864746,
|
| 531 |
+
"logits/rejected": 1.6117414236068726,
|
| 532 |
+
"logps/chosen": -0.7591919302940369,
|
| 533 |
+
"logps/rejected": -1.5995824337005615,
|
| 534 |
+
"loss": 0.5702,
|
| 535 |
+
"rewards/accuracies": 0.637499988079071,
|
| 536 |
+
"rewards/chosen": -1.1387879848480225,
|
| 537 |
+
"rewards/margins": 1.2605856657028198,
|
| 538 |
+
"rewards/rejected": -2.3993735313415527,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.228310502283105,
|
| 543 |
+
"eval_logits/chosen": 1.9625831842422485,
|
| 544 |
+
"eval_logits/rejected": 1.028193473815918,
|
| 545 |
+
"eval_logps/chosen": -0.7516441941261292,
|
| 546 |
+
"eval_logps/rejected": -1.771378517150879,
|
| 547 |
+
"eval_loss": 0.5786539912223816,
|
| 548 |
+
"eval_rewards/accuracies": 0.6915887594223022,
|
| 549 |
+
"eval_rewards/chosen": -1.1274662017822266,
|
| 550 |
+
"eval_rewards/margins": 1.5296014547348022,
|
| 551 |
+
"eval_rewards/rejected": -2.6570677757263184,
|
| 552 |
+
"eval_runtime": 30.7716,
|
| 553 |
+
"eval_samples_per_second": 27.623,
|
| 554 |
+
"eval_steps_per_second": 3.477,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.2359208523592085,
|
| 559 |
+
"grad_norm": 0.5383133292198181,
|
| 560 |
+
"learning_rate": 4.491324795060491e-06,
|
| 561 |
+
"logits/chosen": 1.2824015617370605,
|
| 562 |
+
"logits/rejected": 0.7073851823806763,
|
| 563 |
+
"logps/chosen": -0.8315173387527466,
|
| 564 |
+
"logps/rejected": -1.9733762741088867,
|
| 565 |
+
"loss": 0.587,
|
| 566 |
+
"rewards/accuracies": 0.75,
|
| 567 |
+
"rewards/chosen": -1.2472760677337646,
|
| 568 |
+
"rewards/margins": 1.7127883434295654,
|
| 569 |
+
"rewards/rejected": -2.96006441116333,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.243531202435312,
|
| 574 |
+
"grad_norm": 3.721909284591675,
|
| 575 |
+
"learning_rate": 4.4592336433146e-06,
|
| 576 |
+
"logits/chosen": 1.993947982788086,
|
| 577 |
+
"logits/rejected": 1.192871332168579,
|
| 578 |
+
"logps/chosen": -0.9074883460998535,
|
| 579 |
+
"logps/rejected": -1.9389015436172485,
|
| 580 |
+
"loss": 0.5194,
|
| 581 |
+
"rewards/accuracies": 0.625,
|
| 582 |
+
"rewards/chosen": -1.3612326383590698,
|
| 583 |
+
"rewards/margins": 1.5471194982528687,
|
| 584 |
+
"rewards/rejected": -2.9083518981933594,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.2511415525114155,
|
| 589 |
+
"grad_norm": 0.9611485004425049,
|
| 590 |
+
"learning_rate": 4.426283106939474e-06,
|
| 591 |
+
"logits/chosen": 0.607239305973053,
|
| 592 |
+
"logits/rejected": 0.040740929543972015,
|
| 593 |
+
"logps/chosen": -0.9696615934371948,
|
| 594 |
+
"logps/rejected": -2.3865818977355957,
|
| 595 |
+
"loss": 0.4715,
|
| 596 |
+
"rewards/accuracies": 0.762499988079071,
|
| 597 |
+
"rewards/chosen": -1.4544923305511475,
|
| 598 |
+
"rewards/margins": 2.1253809928894043,
|
| 599 |
+
"rewards/rejected": -3.5798733234405518,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.258751902587519,
|
| 604 |
+
"grad_norm": 3.716665744781494,
|
| 605 |
+
"learning_rate": 4.3924876391293915e-06,
|
| 606 |
+
"logits/chosen": 1.486352801322937,
|
| 607 |
+
"logits/rejected": 0.860406756401062,
|
| 608 |
+
"logps/chosen": -0.9488881826400757,
|
| 609 |
+
"logps/rejected": -2.771193027496338,
|
| 610 |
+
"loss": 0.4584,
|
| 611 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 612 |
+
"rewards/chosen": -1.4233323335647583,
|
| 613 |
+
"rewards/margins": 2.733457088470459,
|
| 614 |
+
"rewards/rejected": -4.156789302825928,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.2663622526636225,
|
| 619 |
+
"grad_norm": 2.496544361114502,
|
| 620 |
+
"learning_rate": 4.357862063693486e-06,
|
| 621 |
+
"logits/chosen": 2.1065332889556885,
|
| 622 |
+
"logits/rejected": 1.4116215705871582,
|
| 623 |
+
"logps/chosen": -0.9290377497673035,
|
| 624 |
+
"logps/rejected": -2.717181444168091,
|
| 625 |
+
"loss": 0.4126,
|
| 626 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 627 |
+
"rewards/chosen": -1.393556833267212,
|
| 628 |
+
"rewards/margins": 2.682215452194214,
|
| 629 |
+
"rewards/rejected": -4.075772285461426,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.2663622526636225,
|
| 634 |
+
"eval_logits/chosen": 2.3063719272613525,
|
| 635 |
+
"eval_logits/rejected": 1.7392665147781372,
|
| 636 |
+
"eval_logps/chosen": -0.9553582072257996,
|
| 637 |
+
"eval_logps/rejected": -2.8578038215637207,
|
| 638 |
+
"eval_loss": 0.43925610184669495,
|
| 639 |
+
"eval_rewards/accuracies": 0.7196261882781982,
|
| 640 |
+
"eval_rewards/chosen": -1.433037281036377,
|
| 641 |
+
"eval_rewards/margins": 2.853668212890625,
|
| 642 |
+
"eval_rewards/rejected": -4.286705493927002,
|
| 643 |
+
"eval_runtime": 30.7732,
|
| 644 |
+
"eval_samples_per_second": 27.621,
|
| 645 |
+
"eval_steps_per_second": 3.477,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.273972602739726,
|
| 650 |
+
"grad_norm": 1.0364434719085693,
|
| 651 |
+
"learning_rate": 4.322421568553529e-06,
|
| 652 |
+
"logits/chosen": 3.5145366191864014,
|
| 653 |
+
"logits/rejected": 2.562318801879883,
|
| 654 |
+
"logps/chosen": -0.9316509366035461,
|
| 655 |
+
"logps/rejected": -2.7451562881469727,
|
| 656 |
+
"loss": 0.4566,
|
| 657 |
+
"rewards/accuracies": 0.675000011920929,
|
| 658 |
+
"rewards/chosen": -1.3974764347076416,
|
| 659 |
+
"rewards/margins": 2.7202582359313965,
|
| 660 |
+
"rewards/rejected": -4.117734432220459,
|
| 661 |
+
"step": 360
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.2815829528158295,
|
| 665 |
+
"grad_norm": 0.7246320843696594,
|
| 666 |
+
"learning_rate": 4.286181699082008e-06,
|
| 667 |
+
"logits/chosen": 1.6608537435531616,
|
| 668 |
+
"logits/rejected": 1.27449631690979,
|
| 669 |
+
"logps/chosen": -1.0797128677368164,
|
| 670 |
+
"logps/rejected": -3.467390537261963,
|
| 671 |
+
"loss": 0.4299,
|
| 672 |
+
"rewards/accuracies": 0.762499988079071,
|
| 673 |
+
"rewards/chosen": -1.6195694208145142,
|
| 674 |
+
"rewards/margins": 3.5815162658691406,
|
| 675 |
+
"rewards/rejected": -5.201085567474365,
|
| 676 |
+
"step": 370
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.289193302891933,
|
| 680 |
+
"grad_norm": 0.942298173904419,
|
| 681 |
+
"learning_rate": 4.249158351283414e-06,
|
| 682 |
+
"logits/chosen": 2.1106579303741455,
|
| 683 |
+
"logits/rejected": 1.5492799282073975,
|
| 684 |
+
"logps/chosen": -1.2671682834625244,
|
| 685 |
+
"logps/rejected": -3.201054811477661,
|
| 686 |
+
"loss": 0.4114,
|
| 687 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 688 |
+
"rewards/chosen": -1.900752305984497,
|
| 689 |
+
"rewards/margins": 2.900829792022705,
|
| 690 |
+
"rewards/rejected": -4.801582336425781,
|
| 691 |
+
"step": 380
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.2968036529680365,
|
| 695 |
+
"grad_norm": 0.4278697371482849,
|
| 696 |
+
"learning_rate": 4.211367764821722e-06,
|
| 697 |
+
"logits/chosen": 3.2620933055877686,
|
| 698 |
+
"logits/rejected": 2.7777600288391113,
|
| 699 |
+
"logps/chosen": -1.0661684274673462,
|
| 700 |
+
"logps/rejected": -3.025578022003174,
|
| 701 |
+
"loss": 0.4259,
|
| 702 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 703 |
+
"rewards/chosen": -1.599252462387085,
|
| 704 |
+
"rewards/margins": 2.9391140937805176,
|
| 705 |
+
"rewards/rejected": -4.53836727142334,
|
| 706 |
+
"step": 390
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.30441400304414,
|
| 710 |
+
"grad_norm": 0.6019588708877563,
|
| 711 |
+
"learning_rate": 4.172826515897146e-06,
|
| 712 |
+
"logits/chosen": 3.057295560836792,
|
| 713 |
+
"logits/rejected": 2.397916078567505,
|
| 714 |
+
"logps/chosen": -1.0584070682525635,
|
| 715 |
+
"logps/rejected": -3.479670286178589,
|
| 716 |
+
"loss": 0.4167,
|
| 717 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 718 |
+
"rewards/chosen": -1.5876106023788452,
|
| 719 |
+
"rewards/margins": 3.631894588470459,
|
| 720 |
+
"rewards/rejected": -5.219505310058594,
|
| 721 |
+
"step": 400
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.30441400304414,
|
| 725 |
+
"eval_logits/chosen": 3.2906479835510254,
|
| 726 |
+
"eval_logits/rejected": 2.9191884994506836,
|
| 727 |
+
"eval_logps/chosen": -1.1546303033828735,
|
| 728 |
+
"eval_logps/rejected": -3.499722957611084,
|
| 729 |
+
"eval_loss": 0.4080003499984741,
|
| 730 |
+
"eval_rewards/accuracies": 0.7102803587913513,
|
| 731 |
+
"eval_rewards/chosen": -1.731945514678955,
|
| 732 |
+
"eval_rewards/margins": 3.517639398574829,
|
| 733 |
+
"eval_rewards/rejected": -5.249584674835205,
|
| 734 |
+
"eval_runtime": 30.7819,
|
| 735 |
+
"eval_samples_per_second": 27.614,
|
| 736 |
+
"eval_steps_per_second": 3.476,
|
| 737 |
+
"step": 400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.3120243531202435,
|
| 741 |
+
"grad_norm": 0.610105037689209,
|
| 742 |
+
"learning_rate": 4.133551509975264e-06,
|
| 743 |
+
"logits/chosen": 2.7366433143615723,
|
| 744 |
+
"logits/rejected": 2.350151538848877,
|
| 745 |
+
"logps/chosen": -1.3425816297531128,
|
| 746 |
+
"logps/rejected": -4.451743125915527,
|
| 747 |
+
"loss": 0.3834,
|
| 748 |
+
"rewards/accuracies": 0.8125,
|
| 749 |
+
"rewards/chosen": -2.0138726234436035,
|
| 750 |
+
"rewards/margins": 4.663742542266846,
|
| 751 |
+
"rewards/rejected": -6.677615165710449,
|
| 752 |
+
"step": 410
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.319634703196347,
|
| 756 |
+
"grad_norm": 0.9136129021644592,
|
| 757 |
+
"learning_rate": 4.093559974371725e-06,
|
| 758 |
+
"logits/chosen": 3.8271331787109375,
|
| 759 |
+
"logits/rejected": 3.666091203689575,
|
| 760 |
+
"logps/chosen": -1.3493579626083374,
|
| 761 |
+
"logps/rejected": -3.8569908142089844,
|
| 762 |
+
"loss": 0.397,
|
| 763 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 764 |
+
"rewards/chosen": -2.0240368843078613,
|
| 765 |
+
"rewards/margins": 3.7614493370056152,
|
| 766 |
+
"rewards/rejected": -5.785486221313477,
|
| 767 |
+
"step": 420
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.3272450532724505,
|
| 771 |
+
"grad_norm": 0.6076493859291077,
|
| 772 |
+
"learning_rate": 4.052869450695776e-06,
|
| 773 |
+
"logits/chosen": 3.027143955230713,
|
| 774 |
+
"logits/rejected": 2.2761549949645996,
|
| 775 |
+
"logps/chosen": -1.2890465259552002,
|
| 776 |
+
"logps/rejected": -4.363173961639404,
|
| 777 |
+
"loss": 0.3623,
|
| 778 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 779 |
+
"rewards/chosen": -1.9335696697235107,
|
| 780 |
+
"rewards/margins": 4.6111907958984375,
|
| 781 |
+
"rewards/rejected": -6.544760704040527,
|
| 782 |
+
"step": 430
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.334855403348554,
|
| 786 |
+
"grad_norm": 1.0694931745529175,
|
| 787 |
+
"learning_rate": 4.011497787155938e-06,
|
| 788 |
+
"logits/chosen": 3.989302158355713,
|
| 789 |
+
"logits/rejected": 3.3767571449279785,
|
| 790 |
+
"logps/chosen": -1.3799726963043213,
|
| 791 |
+
"logps/rejected": -4.611227512359619,
|
| 792 |
+
"loss": 0.3786,
|
| 793 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 794 |
+
"rewards/chosen": -2.0699591636657715,
|
| 795 |
+
"rewards/margins": 4.846882343292236,
|
| 796 |
+
"rewards/rejected": -6.916840553283691,
|
| 797 |
+
"step": 440
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.3424657534246575,
|
| 801 |
+
"grad_norm": 2.3523929119110107,
|
| 802 |
+
"learning_rate": 3.969463130731183e-06,
|
| 803 |
+
"logits/chosen": 3.046278953552246,
|
| 804 |
+
"logits/rejected": 2.7509286403656006,
|
| 805 |
+
"logps/chosen": -1.577859878540039,
|
| 806 |
+
"logps/rejected": -4.554004669189453,
|
| 807 |
+
"loss": 0.3948,
|
| 808 |
+
"rewards/accuracies": 0.762499988079071,
|
| 809 |
+
"rewards/chosen": -2.3667898178100586,
|
| 810 |
+
"rewards/margins": 4.464217185974121,
|
| 811 |
+
"rewards/rejected": -6.831006050109863,
|
| 812 |
+
"step": 450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.3424657534246575,
|
| 816 |
+
"eval_logits/chosen": 3.555213451385498,
|
| 817 |
+
"eval_logits/rejected": 3.359722375869751,
|
| 818 |
+
"eval_logps/chosen": -1.6125141382217407,
|
| 819 |
+
"eval_logps/rejected": -4.374329566955566,
|
| 820 |
+
"eval_loss": 0.3748260736465454,
|
| 821 |
+
"eval_rewards/accuracies": 0.7943925261497498,
|
| 822 |
+
"eval_rewards/chosen": -2.418771266937256,
|
| 823 |
+
"eval_rewards/margins": 4.142723560333252,
|
| 824 |
+
"eval_rewards/rejected": -6.56149435043335,
|
| 825 |
+
"eval_runtime": 30.7795,
|
| 826 |
+
"eval_samples_per_second": 27.616,
|
| 827 |
+
"eval_steps_per_second": 3.476,
|
| 828 |
+
"step": 450
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.350076103500761,
|
| 832 |
+
"grad_norm": 1.113964557647705,
|
| 833 |
+
"learning_rate": 3.92678391921108e-06,
|
| 834 |
+
"logits/chosen": 3.61175274848938,
|
| 835 |
+
"logits/rejected": 3.547903537750244,
|
| 836 |
+
"logps/chosen": -1.7464786767959595,
|
| 837 |
+
"logps/rejected": -5.045803070068359,
|
| 838 |
+
"loss": 0.3717,
|
| 839 |
+
"rewards/accuracies": 0.762499988079071,
|
| 840 |
+
"rewards/chosen": -2.619718074798584,
|
| 841 |
+
"rewards/margins": 4.948986053466797,
|
| 842 |
+
"rewards/rejected": -7.568705081939697,
|
| 843 |
+
"step": 460
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.3576864535768645,
|
| 847 |
+
"grad_norm": 1.5195355415344238,
|
| 848 |
+
"learning_rate": 3.88347887310836e-06,
|
| 849 |
+
"logits/chosen": 3.0807926654815674,
|
| 850 |
+
"logits/rejected": 3.012016773223877,
|
| 851 |
+
"logps/chosen": -2.164515733718872,
|
| 852 |
+
"logps/rejected": -5.039651393890381,
|
| 853 |
+
"loss": 0.3507,
|
| 854 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 855 |
+
"rewards/chosen": -3.2467732429504395,
|
| 856 |
+
"rewards/margins": 4.312704086303711,
|
| 857 |
+
"rewards/rejected": -7.55947732925415,
|
| 858 |
+
"step": 470
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.365296803652968,
|
| 862 |
+
"grad_norm": 2.3880045413970947,
|
| 863 |
+
"learning_rate": 3.839566987447492e-06,
|
| 864 |
+
"logits/chosen": 2.4990105628967285,
|
| 865 |
+
"logits/rejected": 2.5192058086395264,
|
| 866 |
+
"logps/chosen": -2.5131685733795166,
|
| 867 |
+
"logps/rejected": -5.811826705932617,
|
| 868 |
+
"loss": 0.3326,
|
| 869 |
+
"rewards/accuracies": 0.949999988079071,
|
| 870 |
+
"rewards/chosen": -3.7697532176971436,
|
| 871 |
+
"rewards/margins": 4.947987079620361,
|
| 872 |
+
"rewards/rejected": -8.717740058898926,
|
| 873 |
+
"step": 480
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.3729071537290715,
|
| 877 |
+
"grad_norm": 4.61068868637085,
|
| 878 |
+
"learning_rate": 3.795067523432826e-06,
|
| 879 |
+
"logits/chosen": 2.1001622676849365,
|
| 880 |
+
"logits/rejected": 2.0562539100646973,
|
| 881 |
+
"logps/chosen": -2.7572569847106934,
|
| 882 |
+
"logps/rejected": -6.228929042816162,
|
| 883 |
+
"loss": 0.3227,
|
| 884 |
+
"rewards/accuracies": 0.9375,
|
| 885 |
+
"rewards/chosen": -4.135885715484619,
|
| 886 |
+
"rewards/margins": 5.207508563995361,
|
| 887 |
+
"rewards/rejected": -9.343393325805664,
|
| 888 |
+
"step": 490
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.380517503805175,
|
| 892 |
+
"grad_norm": 8.403047561645508,
|
| 893 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 894 |
+
"logits/chosen": 3.1137287616729736,
|
| 895 |
+
"logits/rejected": 2.6646764278411865,
|
| 896 |
+
"logps/chosen": -2.8061861991882324,
|
| 897 |
+
"logps/rejected": -6.236757755279541,
|
| 898 |
+
"loss": 0.3422,
|
| 899 |
+
"rewards/accuracies": 0.862500011920929,
|
| 900 |
+
"rewards/chosen": -4.2092790603637695,
|
| 901 |
+
"rewards/margins": 5.1458563804626465,
|
| 902 |
+
"rewards/rejected": -9.355135917663574,
|
| 903 |
+
"step": 500
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.380517503805175,
|
| 907 |
+
"eval_logits/chosen": 3.3083701133728027,
|
| 908 |
+
"eval_logits/rejected": 3.13222336769104,
|
| 909 |
+
"eval_logps/chosen": -2.6677865982055664,
|
| 910 |
+
"eval_logps/rejected": -5.843282222747803,
|
| 911 |
+
"eval_loss": 0.30595287680625916,
|
| 912 |
+
"eval_rewards/accuracies": 0.8878504633903503,
|
| 913 |
+
"eval_rewards/chosen": -4.001679420471191,
|
| 914 |
+
"eval_rewards/margins": 4.763244152069092,
|
| 915 |
+
"eval_rewards/rejected": -8.764924049377441,
|
| 916 |
+
"eval_runtime": 30.7793,
|
| 917 |
+
"eval_samples_per_second": 27.616,
|
| 918 |
+
"eval_steps_per_second": 3.476,
|
| 919 |
+
"step": 500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.3881278538812785,
|
| 923 |
+
"grad_norm": 2.3582851886749268,
|
| 924 |
+
"learning_rate": 3.7043841852542884e-06,
|
| 925 |
+
"logits/chosen": 2.7116522789001465,
|
| 926 |
+
"logits/rejected": 2.5776076316833496,
|
| 927 |
+
"logps/chosen": -2.7367191314697266,
|
| 928 |
+
"logps/rejected": -6.1324052810668945,
|
| 929 |
+
"loss": 0.2649,
|
| 930 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 931 |
+
"rewards/chosen": -4.10507869720459,
|
| 932 |
+
"rewards/margins": 5.093530178070068,
|
| 933 |
+
"rewards/rejected": -9.1986083984375,
|
| 934 |
+
"step": 510
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.395738203957382,
|
| 938 |
+
"grad_norm": 2.8183226585388184,
|
| 939 |
+
"learning_rate": 3.658240087799655e-06,
|
| 940 |
+
"logits/chosen": 2.327544689178467,
|
| 941 |
+
"logits/rejected": 2.3745343685150146,
|
| 942 |
+
"logps/chosen": -2.6957223415374756,
|
| 943 |
+
"logps/rejected": -6.291537284851074,
|
| 944 |
+
"loss": 0.295,
|
| 945 |
+
"rewards/accuracies": 0.925000011920929,
|
| 946 |
+
"rewards/chosen": -4.043583393096924,
|
| 947 |
+
"rewards/margins": 5.393722057342529,
|
| 948 |
+
"rewards/rejected": -9.437305450439453,
|
| 949 |
+
"step": 520
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.4033485540334855,
|
| 953 |
+
"grad_norm": 1.8313360214233398,
|
| 954 |
+
"learning_rate": 3.611587947962319e-06,
|
| 955 |
+
"logits/chosen": 2.4468109607696533,
|
| 956 |
+
"logits/rejected": 2.4551472663879395,
|
| 957 |
+
"logps/chosen": -2.7839953899383545,
|
| 958 |
+
"logps/rejected": -6.5379180908203125,
|
| 959 |
+
"loss": 0.2618,
|
| 960 |
+
"rewards/accuracies": 0.875,
|
| 961 |
+
"rewards/chosen": -4.175992965698242,
|
| 962 |
+
"rewards/margins": 5.63088321685791,
|
| 963 |
+
"rewards/rejected": -9.806875228881836,
|
| 964 |
+
"step": 530
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.410958904109589,
|
| 968 |
+
"grad_norm": 2.2132411003112793,
|
| 969 |
+
"learning_rate": 3.564448228912682e-06,
|
| 970 |
+
"logits/chosen": 3.125279664993286,
|
| 971 |
+
"logits/rejected": 2.7795650959014893,
|
| 972 |
+
"logps/chosen": -3.349208116531372,
|
| 973 |
+
"logps/rejected": -6.923414707183838,
|
| 974 |
+
"loss": 0.2592,
|
| 975 |
+
"rewards/accuracies": 0.925000011920929,
|
| 976 |
+
"rewards/chosen": -5.023811340332031,
|
| 977 |
+
"rewards/margins": 5.361310005187988,
|
| 978 |
+
"rewards/rejected": -10.385122299194336,
|
| 979 |
+
"step": 540
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 0.4185692541856925,
|
| 983 |
+
"grad_norm": 6.05848503112793,
|
| 984 |
+
"learning_rate": 3.516841607689501e-06,
|
| 985 |
+
"logits/chosen": 2.841399669647217,
|
| 986 |
+
"logits/rejected": 2.997351884841919,
|
| 987 |
+
"logps/chosen": -3.256176710128784,
|
| 988 |
+
"logps/rejected": -7.098822593688965,
|
| 989 |
+
"loss": 0.2603,
|
| 990 |
+
"rewards/accuracies": 0.9375,
|
| 991 |
+
"rewards/chosen": -4.884264945983887,
|
| 992 |
+
"rewards/margins": 5.763968467712402,
|
| 993 |
+
"rewards/rejected": -10.648235321044922,
|
| 994 |
+
"step": 550
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.4185692541856925,
|
| 998 |
+
"eval_logits/chosen": 3.109469413757324,
|
| 999 |
+
"eval_logits/rejected": 3.010756492614746,
|
| 1000 |
+
"eval_logps/chosen": -3.216036558151245,
|
| 1001 |
+
"eval_logps/rejected": -6.825747013092041,
|
| 1002 |
+
"eval_loss": 0.27887609601020813,
|
| 1003 |
+
"eval_rewards/accuracies": 0.8878504633903503,
|
| 1004 |
+
"eval_rewards/chosen": -4.824055194854736,
|
| 1005 |
+
"eval_rewards/margins": 5.414565563201904,
|
| 1006 |
+
"eval_rewards/rejected": -10.23862075805664,
|
| 1007 |
+
"eval_runtime": 30.773,
|
| 1008 |
+
"eval_samples_per_second": 27.622,
|
| 1009 |
+
"eval_steps_per_second": 3.477,
|
| 1010 |
+
"step": 550
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.426179604261796,
|
| 1014 |
+
"grad_norm": 1.8734403848648071,
|
| 1015 |
+
"learning_rate": 3.4687889661302577e-06,
|
| 1016 |
+
"logits/chosen": 1.8899682760238647,
|
| 1017 |
+
"logits/rejected": 1.7766664028167725,
|
| 1018 |
+
"logps/chosen": -3.1907763481140137,
|
| 1019 |
+
"logps/rejected": -7.273028373718262,
|
| 1020 |
+
"loss": 0.2669,
|
| 1021 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1022 |
+
"rewards/chosen": -4.7861647605896,
|
| 1023 |
+
"rewards/margins": 6.123377323150635,
|
| 1024 |
+
"rewards/rejected": -10.909541130065918,
|
| 1025 |
+
"step": 560
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.4337899543378995,
|
| 1029 |
+
"grad_norm": 3.2115261554718018,
|
| 1030 |
+
"learning_rate": 3.4203113817116955e-06,
|
| 1031 |
+
"logits/chosen": 2.563091278076172,
|
| 1032 |
+
"logits/rejected": 2.530696392059326,
|
| 1033 |
+
"logps/chosen": -3.620448589324951,
|
| 1034 |
+
"logps/rejected": -7.546849250793457,
|
| 1035 |
+
"loss": 0.2683,
|
| 1036 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1037 |
+
"rewards/chosen": -5.430673599243164,
|
| 1038 |
+
"rewards/margins": 5.8896002769470215,
|
| 1039 |
+
"rewards/rejected": -11.320273399353027,
|
| 1040 |
+
"step": 570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.441400304414003,
|
| 1044 |
+
"grad_norm": 3.684910297393799,
|
| 1045 |
+
"learning_rate": 3.3714301183045382e-06,
|
| 1046 |
+
"logits/chosen": 2.873882293701172,
|
| 1047 |
+
"logits/rejected": 3.193092107772827,
|
| 1048 |
+
"logps/chosen": -3.386859178543091,
|
| 1049 |
+
"logps/rejected": -7.514338493347168,
|
| 1050 |
+
"loss": 0.2715,
|
| 1051 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1052 |
+
"rewards/chosen": -5.080288887023926,
|
| 1053 |
+
"rewards/margins": 6.191219329833984,
|
| 1054 |
+
"rewards/rejected": -11.271509170532227,
|
| 1055 |
+
"step": 580
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.4490106544901065,
|
| 1059 |
+
"grad_norm": 2.661367416381836,
|
| 1060 |
+
"learning_rate": 3.3221666168464584e-06,
|
| 1061 |
+
"logits/chosen": 2.5157277584075928,
|
| 1062 |
+
"logits/rejected": 2.5739080905914307,
|
| 1063 |
+
"logps/chosen": -3.658534526824951,
|
| 1064 |
+
"logps/rejected": -7.9988884925842285,
|
| 1065 |
+
"loss": 0.2454,
|
| 1066 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1067 |
+
"rewards/chosen": -5.487801551818848,
|
| 1068 |
+
"rewards/margins": 6.510530948638916,
|
| 1069 |
+
"rewards/rejected": -11.998331069946289,
|
| 1070 |
+
"step": 590
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.45662100456621,
|
| 1074 |
+
"grad_norm": 1.8180292844772339,
|
| 1075 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1076 |
+
"logits/chosen": 2.6391870975494385,
|
| 1077 |
+
"logits/rejected": 2.72003173828125,
|
| 1078 |
+
"logps/chosen": -3.382587432861328,
|
| 1079 |
+
"logps/rejected": -8.08546257019043,
|
| 1080 |
+
"loss": 0.2706,
|
| 1081 |
+
"rewards/accuracies": 0.875,
|
| 1082 |
+
"rewards/chosen": -5.073880672454834,
|
| 1083 |
+
"rewards/margins": 7.054312229156494,
|
| 1084 |
+
"rewards/rejected": -12.128194808959961,
|
| 1085 |
+
"step": 600
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.45662100456621,
|
| 1089 |
+
"eval_logits/chosen": 3.235776901245117,
|
| 1090 |
+
"eval_logits/rejected": 3.2310192584991455,
|
| 1091 |
+
"eval_logps/chosen": -3.201641082763672,
|
| 1092 |
+
"eval_logps/rejected": -7.113856315612793,
|
| 1093 |
+
"eval_loss": 0.26078400015830994,
|
| 1094 |
+
"eval_rewards/accuracies": 0.8878504633903503,
|
| 1095 |
+
"eval_rewards/chosen": -4.802461624145508,
|
| 1096 |
+
"eval_rewards/margins": 5.8683247566223145,
|
| 1097 |
+
"eval_rewards/rejected": -10.67078685760498,
|
| 1098 |
+
"eval_runtime": 30.7777,
|
| 1099 |
+
"eval_samples_per_second": 27.617,
|
| 1100 |
+
"eval_steps_per_second": 3.477,
|
| 1101 |
+
"step": 600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.4642313546423135,
|
| 1105 |
+
"grad_norm": 3.1237454414367676,
|
| 1106 |
+
"learning_rate": 3.222579492361179e-06,
|
| 1107 |
+
"logits/chosen": 3.097729444503784,
|
| 1108 |
+
"logits/rejected": 2.8835232257843018,
|
| 1109 |
+
"logps/chosen": -3.2986862659454346,
|
| 1110 |
+
"logps/rejected": -7.824951171875,
|
| 1111 |
+
"loss": 0.2455,
|
| 1112 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1113 |
+
"rewards/chosen": -4.948029518127441,
|
| 1114 |
+
"rewards/margins": 6.789399147033691,
|
| 1115 |
+
"rewards/rejected": -11.7374267578125,
|
| 1116 |
+
"step": 610
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.471841704718417,
|
| 1120 |
+
"grad_norm": 2.250023365020752,
|
| 1121 |
+
"learning_rate": 3.1722995515381644e-06,
|
| 1122 |
+
"logits/chosen": 3.303495407104492,
|
| 1123 |
+
"logits/rejected": 3.124060869216919,
|
| 1124 |
+
"logps/chosen": -2.9880855083465576,
|
| 1125 |
+
"logps/rejected": -6.581275939941406,
|
| 1126 |
+
"loss": 0.2754,
|
| 1127 |
+
"rewards/accuracies": 0.875,
|
| 1128 |
+
"rewards/chosen": -4.482128143310547,
|
| 1129 |
+
"rewards/margins": 5.389786243438721,
|
| 1130 |
+
"rewards/rejected": -9.87191390991211,
|
| 1131 |
+
"step": 620
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.4794520547945205,
|
| 1135 |
+
"grad_norm": 4.364448547363281,
|
| 1136 |
+
"learning_rate": 3.121724717912138e-06,
|
| 1137 |
+
"logits/chosen": 2.8994319438934326,
|
| 1138 |
+
"logits/rejected": 2.593780755996704,
|
| 1139 |
+
"logps/chosen": -3.4450290203094482,
|
| 1140 |
+
"logps/rejected": -7.1797990798950195,
|
| 1141 |
+
"loss": 0.2316,
|
| 1142 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1143 |
+
"rewards/chosen": -5.167544364929199,
|
| 1144 |
+
"rewards/margins": 5.602154731750488,
|
| 1145 |
+
"rewards/rejected": -10.769698143005371,
|
| 1146 |
+
"step": 630
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.487062404870624,
|
| 1150 |
+
"grad_norm": 3.65561580657959,
|
| 1151 |
+
"learning_rate": 3.0708771752766397e-06,
|
| 1152 |
+
"logits/chosen": 3.1075518131256104,
|
| 1153 |
+
"logits/rejected": 2.8703231811523438,
|
| 1154 |
+
"logps/chosen": -3.26599383354187,
|
| 1155 |
+
"logps/rejected": -7.536534786224365,
|
| 1156 |
+
"loss": 0.2549,
|
| 1157 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1158 |
+
"rewards/chosen": -4.898990631103516,
|
| 1159 |
+
"rewards/margins": 6.4058122634887695,
|
| 1160 |
+
"rewards/rejected": -11.304803848266602,
|
| 1161 |
+
"step": 640
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.4946727549467275,
|
| 1165 |
+
"grad_norm": 3.1211891174316406,
|
| 1166 |
+
"learning_rate": 3.019779227044398e-06,
|
| 1167 |
+
"logits/chosen": 3.2364888191223145,
|
| 1168 |
+
"logits/rejected": 3.3938751220703125,
|
| 1169 |
+
"logps/chosen": -3.538849353790283,
|
| 1170 |
+
"logps/rejected": -7.827691555023193,
|
| 1171 |
+
"loss": 0.2571,
|
| 1172 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1173 |
+
"rewards/chosen": -5.308274269104004,
|
| 1174 |
+
"rewards/margins": 6.433261871337891,
|
| 1175 |
+
"rewards/rejected": -11.741537094116211,
|
| 1176 |
+
"step": 650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.4946727549467275,
|
| 1180 |
+
"eval_logits/chosen": 3.091616153717041,
|
| 1181 |
+
"eval_logits/rejected": 3.0459396839141846,
|
| 1182 |
+
"eval_logps/chosen": -3.361125946044922,
|
| 1183 |
+
"eval_logps/rejected": -7.390212535858154,
|
| 1184 |
+
"eval_loss": 0.2536354660987854,
|
| 1185 |
+
"eval_rewards/accuracies": 0.8971962332725525,
|
| 1186 |
+
"eval_rewards/chosen": -5.041689395904541,
|
| 1187 |
+
"eval_rewards/margins": 6.043630599975586,
|
| 1188 |
+
"eval_rewards/rejected": -11.085319519042969,
|
| 1189 |
+
"eval_runtime": 30.7751,
|
| 1190 |
+
"eval_samples_per_second": 27.62,
|
| 1191 |
+
"eval_steps_per_second": 3.477,
|
| 1192 |
+
"step": 650
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.502283105022831,
|
| 1196 |
+
"grad_norm": 4.375415802001953,
|
| 1197 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 1198 |
+
"logits/chosen": 1.8274192810058594,
|
| 1199 |
+
"logits/rejected": 1.9628839492797852,
|
| 1200 |
+
"logps/chosen": -3.083608627319336,
|
| 1201 |
+
"logps/rejected": -8.370513916015625,
|
| 1202 |
+
"loss": 0.2166,
|
| 1203 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1204 |
+
"rewards/chosen": -4.625412940979004,
|
| 1205 |
+
"rewards/margins": 7.930357456207275,
|
| 1206 |
+
"rewards/rejected": -12.555770874023438,
|
| 1207 |
+
"step": 660
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.5098934550989346,
|
| 1211 |
+
"grad_norm": 3.6583638191223145,
|
| 1212 |
+
"learning_rate": 2.9169218667902562e-06,
|
| 1213 |
+
"logits/chosen": 2.5409281253814697,
|
| 1214 |
+
"logits/rejected": 2.46968150138855,
|
| 1215 |
+
"logps/chosen": -3.252990245819092,
|
| 1216 |
+
"logps/rejected": -7.6487884521484375,
|
| 1217 |
+
"loss": 0.2103,
|
| 1218 |
+
"rewards/accuracies": 0.9375,
|
| 1219 |
+
"rewards/chosen": -4.879485607147217,
|
| 1220 |
+
"rewards/margins": 6.593697547912598,
|
| 1221 |
+
"rewards/rejected": -11.473182678222656,
|
| 1222 |
+
"step": 670
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.517503805175038,
|
| 1226 |
+
"grad_norm": 2.009876251220703,
|
| 1227 |
+
"learning_rate": 2.8652075714060296e-06,
|
| 1228 |
+
"logits/chosen": 3.553900957107544,
|
| 1229 |
+
"logits/rejected": 3.4104526042938232,
|
| 1230 |
+
"logps/chosen": -2.9901890754699707,
|
| 1231 |
+
"logps/rejected": -7.472288608551025,
|
| 1232 |
+
"loss": 0.2405,
|
| 1233 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1234 |
+
"rewards/chosen": -4.485283374786377,
|
| 1235 |
+
"rewards/margins": 6.723149299621582,
|
| 1236 |
+
"rewards/rejected": -11.208433151245117,
|
| 1237 |
+
"step": 680
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.5251141552511416,
|
| 1241 |
+
"grad_norm": 2.9065611362457275,
|
| 1242 |
+
"learning_rate": 2.813333083910761e-06,
|
| 1243 |
+
"logits/chosen": 2.323111057281494,
|
| 1244 |
+
"logits/rejected": 2.0086140632629395,
|
| 1245 |
+
"logps/chosen": -3.430807590484619,
|
| 1246 |
+
"logps/rejected": -8.105338096618652,
|
| 1247 |
+
"loss": 0.2182,
|
| 1248 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1249 |
+
"rewards/chosen": -5.146210670471191,
|
| 1250 |
+
"rewards/margins": 7.011796474456787,
|
| 1251 |
+
"rewards/rejected": -12.15800666809082,
|
| 1252 |
+
"step": 690
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.532724505327245,
|
| 1256 |
+
"grad_norm": 4.097139358520508,
|
| 1257 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1258 |
+
"logits/chosen": 1.9292926788330078,
|
| 1259 |
+
"logits/rejected": 2.0105385780334473,
|
| 1260 |
+
"logps/chosen": -3.337139129638672,
|
| 1261 |
+
"logps/rejected": -7.7645721435546875,
|
| 1262 |
+
"loss": 0.245,
|
| 1263 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1264 |
+
"rewards/chosen": -5.005709171295166,
|
| 1265 |
+
"rewards/margins": 6.641148567199707,
|
| 1266 |
+
"rewards/rejected": -11.646858215332031,
|
| 1267 |
+
"step": 700
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 0.532724505327245,
|
| 1271 |
+
"eval_logits/chosen": 2.946713924407959,
|
| 1272 |
+
"eval_logits/rejected": 2.912729501724243,
|
| 1273 |
+
"eval_logps/chosen": -3.3646414279937744,
|
| 1274 |
+
"eval_logps/rejected": -7.633481979370117,
|
| 1275 |
+
"eval_loss": 0.2412233203649521,
|
| 1276 |
+
"eval_rewards/accuracies": 0.9065420627593994,
|
| 1277 |
+
"eval_rewards/chosen": -5.046962261199951,
|
| 1278 |
+
"eval_rewards/margins": 6.403261184692383,
|
| 1279 |
+
"eval_rewards/rejected": -11.450223922729492,
|
| 1280 |
+
"eval_runtime": 30.7742,
|
| 1281 |
+
"eval_samples_per_second": 27.621,
|
| 1282 |
+
"eval_steps_per_second": 3.477,
|
| 1283 |
+
"step": 700
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.5403348554033486,
|
| 1287 |
+
"grad_norm": 4.534071922302246,
|
| 1288 |
+
"learning_rate": 2.70919460833079e-06,
|
| 1289 |
+
"logits/chosen": 2.4406304359436035,
|
| 1290 |
+
"logits/rejected": 2.713369607925415,
|
| 1291 |
+
"logps/chosen": -3.4392433166503906,
|
| 1292 |
+
"logps/rejected": -8.228872299194336,
|
| 1293 |
+
"loss": 0.2245,
|
| 1294 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1295 |
+
"rewards/chosen": -5.158864498138428,
|
| 1296 |
+
"rewards/margins": 7.184444427490234,
|
| 1297 |
+
"rewards/rejected": -12.34330940246582,
|
| 1298 |
+
"step": 710
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.547945205479452,
|
| 1302 |
+
"grad_norm": 1.5941667556762695,
|
| 1303 |
+
"learning_rate": 2.6569762988232838e-06,
|
| 1304 |
+
"logits/chosen": 3.3259758949279785,
|
| 1305 |
+
"logits/rejected": 3.3186354637145996,
|
| 1306 |
+
"logps/chosen": -3.7198212146759033,
|
| 1307 |
+
"logps/rejected": -7.6400909423828125,
|
| 1308 |
+
"loss": 0.2339,
|
| 1309 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1310 |
+
"rewards/chosen": -5.5797319412231445,
|
| 1311 |
+
"rewards/margins": 5.8804030418396,
|
| 1312 |
+
"rewards/rejected": -11.460134506225586,
|
| 1313 |
+
"step": 720
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.5555555555555556,
|
| 1317 |
+
"grad_norm": 5.972750186920166,
|
| 1318 |
+
"learning_rate": 2.604689134322999e-06,
|
| 1319 |
+
"logits/chosen": 2.455244302749634,
|
| 1320 |
+
"logits/rejected": 2.5398240089416504,
|
| 1321 |
+
"logps/chosen": -3.892965316772461,
|
| 1322 |
+
"logps/rejected": -8.59666633605957,
|
| 1323 |
+
"loss": 0.2041,
|
| 1324 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1325 |
+
"rewards/chosen": -5.83944845199585,
|
| 1326 |
+
"rewards/margins": 7.055548667907715,
|
| 1327 |
+
"rewards/rejected": -12.894998550415039,
|
| 1328 |
+
"step": 730
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.563165905631659,
|
| 1332 |
+
"grad_norm": 3.1441280841827393,
|
| 1333 |
+
"learning_rate": 2.5523560497083927e-06,
|
| 1334 |
+
"logits/chosen": 2.3029608726501465,
|
| 1335 |
+
"logits/rejected": 2.3662524223327637,
|
| 1336 |
+
"logps/chosen": -3.682513475418091,
|
| 1337 |
+
"logps/rejected": -8.340951919555664,
|
| 1338 |
+
"loss": 0.2723,
|
| 1339 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1340 |
+
"rewards/chosen": -5.523770332336426,
|
| 1341 |
+
"rewards/margins": 6.9876580238342285,
|
| 1342 |
+
"rewards/rejected": -12.511428833007812,
|
| 1343 |
+
"step": 740
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.5707762557077626,
|
| 1347 |
+
"grad_norm": 2.30711030960083,
|
| 1348 |
+
"learning_rate": 2.5e-06,
|
| 1349 |
+
"logits/chosen": 1.8446356058120728,
|
| 1350 |
+
"logits/rejected": 1.9468234777450562,
|
| 1351 |
+
"logps/chosen": -3.808454990386963,
|
| 1352 |
+
"logps/rejected": -8.047523498535156,
|
| 1353 |
+
"loss": 0.1936,
|
| 1354 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1355 |
+
"rewards/chosen": -5.712681770324707,
|
| 1356 |
+
"rewards/margins": 6.3586015701293945,
|
| 1357 |
+
"rewards/rejected": -12.071284294128418,
|
| 1358 |
+
"step": 750
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.5707762557077626,
|
| 1362 |
+
"eval_logits/chosen": 2.7250914573669434,
|
| 1363 |
+
"eval_logits/rejected": 2.7465131282806396,
|
| 1364 |
+
"eval_logps/chosen": -3.5594334602355957,
|
| 1365 |
+
"eval_logps/rejected": -7.899675369262695,
|
| 1366 |
+
"eval_loss": 0.2329329252243042,
|
| 1367 |
+
"eval_rewards/accuracies": 0.9345794320106506,
|
| 1368 |
+
"eval_rewards/chosen": -5.339150905609131,
|
| 1369 |
+
"eval_rewards/margins": 6.510361671447754,
|
| 1370 |
+
"eval_rewards/rejected": -11.84951114654541,
|
| 1371 |
+
"eval_runtime": 30.7783,
|
| 1372 |
+
"eval_samples_per_second": 27.617,
|
| 1373 |
+
"eval_steps_per_second": 3.476,
|
| 1374 |
+
"step": 750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.578386605783866,
|
| 1378 |
+
"grad_norm": 3.0719997882843018,
|
| 1379 |
+
"learning_rate": 2.447643950291608e-06,
|
| 1380 |
+
"logits/chosen": 2.3534064292907715,
|
| 1381 |
+
"logits/rejected": 2.401563882827759,
|
| 1382 |
+
"logps/chosen": -3.610807418823242,
|
| 1383 |
+
"logps/rejected": -7.985041618347168,
|
| 1384 |
+
"loss": 0.1847,
|
| 1385 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1386 |
+
"rewards/chosen": -5.416211128234863,
|
| 1387 |
+
"rewards/margins": 6.561351776123047,
|
| 1388 |
+
"rewards/rejected": -11.97756290435791,
|
| 1389 |
+
"step": 760
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.5859969558599696,
|
| 1393 |
+
"grad_norm": 3.6076536178588867,
|
| 1394 |
+
"learning_rate": 2.3953108656770018e-06,
|
| 1395 |
+
"logits/chosen": 1.969351053237915,
|
| 1396 |
+
"logits/rejected": 2.2207655906677246,
|
| 1397 |
+
"logps/chosen": -2.9666709899902344,
|
| 1398 |
+
"logps/rejected": -8.2462797164917,
|
| 1399 |
+
"loss": 0.2211,
|
| 1400 |
+
"rewards/accuracies": 0.9375,
|
| 1401 |
+
"rewards/chosen": -4.450006484985352,
|
| 1402 |
+
"rewards/margins": 7.919413089752197,
|
| 1403 |
+
"rewards/rejected": -12.369420051574707,
|
| 1404 |
+
"step": 770
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.593607305936073,
|
| 1408 |
+
"grad_norm": 3.9795055389404297,
|
| 1409 |
+
"learning_rate": 2.3430237011767166e-06,
|
| 1410 |
+
"logits/chosen": 2.6336114406585693,
|
| 1411 |
+
"logits/rejected": 2.438262462615967,
|
| 1412 |
+
"logps/chosen": -3.4108245372772217,
|
| 1413 |
+
"logps/rejected": -8.591341018676758,
|
| 1414 |
+
"loss": 0.2306,
|
| 1415 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1416 |
+
"rewards/chosen": -5.116236209869385,
|
| 1417 |
+
"rewards/margins": 7.770774841308594,
|
| 1418 |
+
"rewards/rejected": -12.887011528015137,
|
| 1419 |
+
"step": 780
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.6012176560121766,
|
| 1423 |
+
"grad_norm": 3.288729667663574,
|
| 1424 |
+
"learning_rate": 2.290805391669212e-06,
|
| 1425 |
+
"logits/chosen": 1.6185804605484009,
|
| 1426 |
+
"logits/rejected": 1.6822645664215088,
|
| 1427 |
+
"logps/chosen": -3.2728798389434814,
|
| 1428 |
+
"logps/rejected": -7.791805267333984,
|
| 1429 |
+
"loss": 0.2329,
|
| 1430 |
+
"rewards/accuracies": 0.9375,
|
| 1431 |
+
"rewards/chosen": -4.909319877624512,
|
| 1432 |
+
"rewards/margins": 6.778387546539307,
|
| 1433 |
+
"rewards/rejected": -11.687707901000977,
|
| 1434 |
+
"step": 790
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.60882800608828,
|
| 1438 |
+
"grad_norm": 2.315206289291382,
|
| 1439 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1440 |
+
"logits/chosen": 2.018578052520752,
|
| 1441 |
+
"logits/rejected": 1.8465204238891602,
|
| 1442 |
+
"logps/chosen": -3.2794997692108154,
|
| 1443 |
+
"logps/rejected": -8.164708137512207,
|
| 1444 |
+
"loss": 0.2082,
|
| 1445 |
+
"rewards/accuracies": 0.9375,
|
| 1446 |
+
"rewards/chosen": -4.919250011444092,
|
| 1447 |
+
"rewards/margins": 7.327812194824219,
|
| 1448 |
+
"rewards/rejected": -12.247061729431152,
|
| 1449 |
+
"step": 800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.60882800608828,
|
| 1453 |
+
"eval_logits/chosen": 3.08111572265625,
|
| 1454 |
+
"eval_logits/rejected": 3.0503101348876953,
|
| 1455 |
+
"eval_logps/chosen": -3.4727284908294678,
|
| 1456 |
+
"eval_logps/rejected": -7.895880699157715,
|
| 1457 |
+
"eval_loss": 0.22325536608695984,
|
| 1458 |
+
"eval_rewards/accuracies": 0.9252336621284485,
|
| 1459 |
+
"eval_rewards/chosen": -5.209092617034912,
|
| 1460 |
+
"eval_rewards/margins": 6.634730339050293,
|
| 1461 |
+
"eval_rewards/rejected": -11.84382152557373,
|
| 1462 |
+
"eval_runtime": 30.7753,
|
| 1463 |
+
"eval_samples_per_second": 27.62,
|
| 1464 |
+
"eval_steps_per_second": 3.477,
|
| 1465 |
+
"step": 800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.6164383561643836,
|
| 1469 |
+
"grad_norm": 2.4823191165924072,
|
| 1470 |
+
"learning_rate": 2.186666916089239e-06,
|
| 1471 |
+
"logits/chosen": 2.6345906257629395,
|
| 1472 |
+
"logits/rejected": 2.5206990242004395,
|
| 1473 |
+
"logps/chosen": -3.123882532119751,
|
| 1474 |
+
"logps/rejected": -7.772116184234619,
|
| 1475 |
+
"loss": 0.1931,
|
| 1476 |
+
"rewards/accuracies": 0.9375,
|
| 1477 |
+
"rewards/chosen": -4.685823440551758,
|
| 1478 |
+
"rewards/margins": 6.972352027893066,
|
| 1479 |
+
"rewards/rejected": -11.658174514770508,
|
| 1480 |
+
"step": 810
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.624048706240487,
|
| 1484 |
+
"grad_norm": 1.3935630321502686,
|
| 1485 |
+
"learning_rate": 2.134792428593971e-06,
|
| 1486 |
+
"logits/chosen": 2.4640917778015137,
|
| 1487 |
+
"logits/rejected": 2.531430721282959,
|
| 1488 |
+
"logps/chosen": -3.5584664344787598,
|
| 1489 |
+
"logps/rejected": -8.64016342163086,
|
| 1490 |
+
"loss": 0.205,
|
| 1491 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1492 |
+
"rewards/chosen": -5.337700366973877,
|
| 1493 |
+
"rewards/margins": 7.622546195983887,
|
| 1494 |
+
"rewards/rejected": -12.960246086120605,
|
| 1495 |
+
"step": 820
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.6316590563165906,
|
| 1499 |
+
"grad_norm": 2.7727415561676025,
|
| 1500 |
+
"learning_rate": 2.0830781332097446e-06,
|
| 1501 |
+
"logits/chosen": 3.5152816772460938,
|
| 1502 |
+
"logits/rejected": 3.3117728233337402,
|
| 1503 |
+
"logps/chosen": -3.820496082305908,
|
| 1504 |
+
"logps/rejected": -8.241179466247559,
|
| 1505 |
+
"loss": 0.2315,
|
| 1506 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1507 |
+
"rewards/chosen": -5.730743408203125,
|
| 1508 |
+
"rewards/margins": 6.631025791168213,
|
| 1509 |
+
"rewards/rejected": -12.361770629882812,
|
| 1510 |
+
"step": 830
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 0.639269406392694,
|
| 1514 |
+
"grad_norm": 3.5436055660247803,
|
| 1515 |
+
"learning_rate": 2.031546713535688e-06,
|
| 1516 |
+
"logits/chosen": 2.813575506210327,
|
| 1517 |
+
"logits/rejected": 2.650791645050049,
|
| 1518 |
+
"logps/chosen": -3.552276134490967,
|
| 1519 |
+
"logps/rejected": -8.587823867797852,
|
| 1520 |
+
"loss": 0.1964,
|
| 1521 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1522 |
+
"rewards/chosen": -5.328413963317871,
|
| 1523 |
+
"rewards/margins": 7.553321838378906,
|
| 1524 |
+
"rewards/rejected": -12.881736755371094,
|
| 1525 |
+
"step": 840
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.6468797564687976,
|
| 1529 |
+
"grad_norm": 4.140048027038574,
|
| 1530 |
+
"learning_rate": 1.9802207729556023e-06,
|
| 1531 |
+
"logits/chosen": 2.695815324783325,
|
| 1532 |
+
"logits/rejected": 2.509413242340088,
|
| 1533 |
+
"logps/chosen": -3.706005573272705,
|
| 1534 |
+
"logps/rejected": -8.673744201660156,
|
| 1535 |
+
"loss": 0.1882,
|
| 1536 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1537 |
+
"rewards/chosen": -5.55900764465332,
|
| 1538 |
+
"rewards/margins": 7.451608180999756,
|
| 1539 |
+
"rewards/rejected": -13.01061725616455,
|
| 1540 |
+
"step": 850
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.6468797564687976,
|
| 1544 |
+
"eval_logits/chosen": 3.0777766704559326,
|
| 1545 |
+
"eval_logits/rejected": 3.120222806930542,
|
| 1546 |
+
"eval_logps/chosen": -3.801321268081665,
|
| 1547 |
+
"eval_logps/rejected": -8.376675605773926,
|
| 1548 |
+
"eval_loss": 0.21778903901576996,
|
| 1549 |
+
"eval_rewards/accuracies": 0.9345794320106506,
|
| 1550 |
+
"eval_rewards/chosen": -5.701981067657471,
|
| 1551 |
+
"eval_rewards/margins": 6.863031387329102,
|
| 1552 |
+
"eval_rewards/rejected": -12.565014839172363,
|
| 1553 |
+
"eval_runtime": 30.7706,
|
| 1554 |
+
"eval_samples_per_second": 27.624,
|
| 1555 |
+
"eval_steps_per_second": 3.477,
|
| 1556 |
+
"step": 850
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.654490106544901,
|
| 1560 |
+
"grad_norm": 3.0274274349212646,
|
| 1561 |
+
"learning_rate": 1.9291228247233607e-06,
|
| 1562 |
+
"logits/chosen": 2.446882486343384,
|
| 1563 |
+
"logits/rejected": 2.352177143096924,
|
| 1564 |
+
"logps/chosen": -3.8477814197540283,
|
| 1565 |
+
"logps/rejected": -9.340738296508789,
|
| 1566 |
+
"loss": 0.1916,
|
| 1567 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1568 |
+
"rewards/chosen": -5.771672248840332,
|
| 1569 |
+
"rewards/margins": 8.239435195922852,
|
| 1570 |
+
"rewards/rejected": -14.011106491088867,
|
| 1571 |
+
"step": 860
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.6621004566210046,
|
| 1575 |
+
"grad_norm": 4.4860520362854,
|
| 1576 |
+
"learning_rate": 1.8782752820878636e-06,
|
| 1577 |
+
"logits/chosen": 2.77601957321167,
|
| 1578 |
+
"logits/rejected": 2.853281021118164,
|
| 1579 |
+
"logps/chosen": -4.01420259475708,
|
| 1580 |
+
"logps/rejected": -9.463164329528809,
|
| 1581 |
+
"loss": 0.2013,
|
| 1582 |
+
"rewards/accuracies": 0.9375,
|
| 1583 |
+
"rewards/chosen": -6.021303653717041,
|
| 1584 |
+
"rewards/margins": 8.173443794250488,
|
| 1585 |
+
"rewards/rejected": -14.194747924804688,
|
| 1586 |
+
"step": 870
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.669710806697108,
|
| 1590 |
+
"grad_norm": 6.593935012817383,
|
| 1591 |
+
"learning_rate": 1.827700448461836e-06,
|
| 1592 |
+
"logits/chosen": 2.5073022842407227,
|
| 1593 |
+
"logits/rejected": 2.9579524993896484,
|
| 1594 |
+
"logps/chosen": -3.9602768421173096,
|
| 1595 |
+
"logps/rejected": -9.482460021972656,
|
| 1596 |
+
"loss": 0.2163,
|
| 1597 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1598 |
+
"rewards/chosen": -5.940415382385254,
|
| 1599 |
+
"rewards/margins": 8.283275604248047,
|
| 1600 |
+
"rewards/rejected": -14.2236909866333,
|
| 1601 |
+
"step": 880
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.6773211567732116,
|
| 1605 |
+
"grad_norm": 3.101680040359497,
|
| 1606 |
+
"learning_rate": 1.7774205076388207e-06,
|
| 1607 |
+
"logits/chosen": 2.27485990524292,
|
| 1608 |
+
"logits/rejected": 2.5005619525909424,
|
| 1609 |
+
"logps/chosen": -3.828552722930908,
|
| 1610 |
+
"logps/rejected": -8.851189613342285,
|
| 1611 |
+
"loss": 0.1935,
|
| 1612 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1613 |
+
"rewards/chosen": -5.742828845977783,
|
| 1614 |
+
"rewards/margins": 7.5339555740356445,
|
| 1615 |
+
"rewards/rejected": -13.27678394317627,
|
| 1616 |
+
"step": 890
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.684931506849315,
|
| 1620 |
+
"grad_norm": 3.6742115020751953,
|
| 1621 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1622 |
+
"logits/chosen": 2.381319761276245,
|
| 1623 |
+
"logits/rejected": 2.543203353881836,
|
| 1624 |
+
"logps/chosen": -3.4646923542022705,
|
| 1625 |
+
"logps/rejected": -9.015104293823242,
|
| 1626 |
+
"loss": 0.1903,
|
| 1627 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1628 |
+
"rewards/chosen": -5.197038173675537,
|
| 1629 |
+
"rewards/margins": 8.325616836547852,
|
| 1630 |
+
"rewards/rejected": -13.522656440734863,
|
| 1631 |
+
"step": 900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 0.684931506849315,
|
| 1635 |
+
"eval_logits/chosen": 3.069916009902954,
|
| 1636 |
+
"eval_logits/rejected": 3.104632616043091,
|
| 1637 |
+
"eval_logps/chosen": -3.6102888584136963,
|
| 1638 |
+
"eval_logps/rejected": -8.349987983703613,
|
| 1639 |
+
"eval_loss": 0.2161051481962204,
|
| 1640 |
+
"eval_rewards/accuracies": 0.9345794320106506,
|
| 1641 |
+
"eval_rewards/chosen": -5.415433406829834,
|
| 1642 |
+
"eval_rewards/margins": 7.1095476150512695,
|
| 1643 |
+
"eval_rewards/rejected": -12.524979591369629,
|
| 1644 |
+
"eval_runtime": 30.7844,
|
| 1645 |
+
"eval_samples_per_second": 27.611,
|
| 1646 |
+
"eval_steps_per_second": 3.476,
|
| 1647 |
+
"step": 900
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.6925418569254186,
|
| 1651 |
+
"grad_norm": 2.9264719486236572,
|
| 1652 |
+
"learning_rate": 1.677833383153542e-06,
|
| 1653 |
+
"logits/chosen": 2.0955498218536377,
|
| 1654 |
+
"logits/rejected": 2.1619057655334473,
|
| 1655 |
+
"logps/chosen": -3.460951328277588,
|
| 1656 |
+
"logps/rejected": -8.898179054260254,
|
| 1657 |
+
"loss": 0.1687,
|
| 1658 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1659 |
+
"rewards/chosen": -5.191426753997803,
|
| 1660 |
+
"rewards/margins": 8.155839920043945,
|
| 1661 |
+
"rewards/rejected": -13.347267150878906,
|
| 1662 |
+
"step": 910
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.700152207001522,
|
| 1666 |
+
"grad_norm": 2.0860531330108643,
|
| 1667 |
+
"learning_rate": 1.6285698816954626e-06,
|
| 1668 |
+
"logits/chosen": 3.346489429473877,
|
| 1669 |
+
"logits/rejected": 3.2421345710754395,
|
| 1670 |
+
"logps/chosen": -3.63940691947937,
|
| 1671 |
+
"logps/rejected": -8.418205261230469,
|
| 1672 |
+
"loss": 0.2161,
|
| 1673 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1674 |
+
"rewards/chosen": -5.459110260009766,
|
| 1675 |
+
"rewards/margins": 7.168198585510254,
|
| 1676 |
+
"rewards/rejected": -12.627306938171387,
|
| 1677 |
+
"step": 920
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.7077625570776256,
|
| 1681 |
+
"grad_norm": 2.778262138366699,
|
| 1682 |
+
"learning_rate": 1.5796886182883053e-06,
|
| 1683 |
+
"logits/chosen": 1.7459516525268555,
|
| 1684 |
+
"logits/rejected": 2.1579155921936035,
|
| 1685 |
+
"logps/chosen": -3.3034520149230957,
|
| 1686 |
+
"logps/rejected": -9.191883087158203,
|
| 1687 |
+
"loss": 0.2007,
|
| 1688 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1689 |
+
"rewards/chosen": -4.955178260803223,
|
| 1690 |
+
"rewards/margins": 8.832646369934082,
|
| 1691 |
+
"rewards/rejected": -13.787823677062988,
|
| 1692 |
+
"step": 930
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 0.715372907153729,
|
| 1696 |
+
"grad_norm": 3.695908546447754,
|
| 1697 |
+
"learning_rate": 1.5312110338697427e-06,
|
| 1698 |
+
"logits/chosen": 2.7586328983306885,
|
| 1699 |
+
"logits/rejected": 2.920232057571411,
|
| 1700 |
+
"logps/chosen": -3.6659629344940186,
|
| 1701 |
+
"logps/rejected": -8.657777786254883,
|
| 1702 |
+
"loss": 0.1907,
|
| 1703 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1704 |
+
"rewards/chosen": -5.498944282531738,
|
| 1705 |
+
"rewards/margins": 7.487723350524902,
|
| 1706 |
+
"rewards/rejected": -12.986666679382324,
|
| 1707 |
+
"step": 940
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.7229832572298326,
|
| 1711 |
+
"grad_norm": 3.2104835510253906,
|
| 1712 |
+
"learning_rate": 1.4831583923105e-06,
|
| 1713 |
+
"logits/chosen": 2.8813366889953613,
|
| 1714 |
+
"logits/rejected": 3.21830677986145,
|
| 1715 |
+
"logps/chosen": -3.41332745552063,
|
| 1716 |
+
"logps/rejected": -8.479570388793945,
|
| 1717 |
+
"loss": 0.2074,
|
| 1718 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1719 |
+
"rewards/chosen": -5.119990825653076,
|
| 1720 |
+
"rewards/margins": 7.599363803863525,
|
| 1721 |
+
"rewards/rejected": -12.719354629516602,
|
| 1722 |
+
"step": 950
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.7229832572298326,
|
| 1726 |
+
"eval_logits/chosen": 3.281101703643799,
|
| 1727 |
+
"eval_logits/rejected": 3.3282196521759033,
|
| 1728 |
+
"eval_logps/chosen": -3.6136667728424072,
|
| 1729 |
+
"eval_logps/rejected": -8.419679641723633,
|
| 1730 |
+
"eval_loss": 0.21113382279872894,
|
| 1731 |
+
"eval_rewards/accuracies": 0.9252336621284485,
|
| 1732 |
+
"eval_rewards/chosen": -5.4205002784729,
|
| 1733 |
+
"eval_rewards/margins": 7.209019660949707,
|
| 1734 |
+
"eval_rewards/rejected": -12.629520416259766,
|
| 1735 |
+
"eval_runtime": 30.777,
|
| 1736 |
+
"eval_samples_per_second": 27.618,
|
| 1737 |
+
"eval_steps_per_second": 3.477,
|
| 1738 |
+
"step": 950
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.730593607305936,
|
| 1742 |
+
"grad_norm": 6.422901153564453,
|
| 1743 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 1744 |
+
"logits/chosen": 2.0251851081848145,
|
| 1745 |
+
"logits/rejected": 2.1296639442443848,
|
| 1746 |
+
"logps/chosen": -3.15910005569458,
|
| 1747 |
+
"logps/rejected": -8.686817169189453,
|
| 1748 |
+
"loss": 0.187,
|
| 1749 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1750 |
+
"rewards/chosen": -4.738650321960449,
|
| 1751 |
+
"rewards/margins": 8.29157543182373,
|
| 1752 |
+
"rewards/rejected": -13.03022575378418,
|
| 1753 |
+
"step": 960
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.7382039573820396,
|
| 1757 |
+
"grad_norm": 1.537139654159546,
|
| 1758 |
+
"learning_rate": 1.388412052037682e-06,
|
| 1759 |
+
"logits/chosen": 3.1659352779388428,
|
| 1760 |
+
"logits/rejected": 3.161760091781616,
|
| 1761 |
+
"logps/chosen": -3.755237102508545,
|
| 1762 |
+
"logps/rejected": -8.484071731567383,
|
| 1763 |
+
"loss": 0.1823,
|
| 1764 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1765 |
+
"rewards/chosen": -5.6328558921813965,
|
| 1766 |
+
"rewards/margins": 7.093251705169678,
|
| 1767 |
+
"rewards/rejected": -12.726107597351074,
|
| 1768 |
+
"step": 970
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.745814307458143,
|
| 1772 |
+
"grad_norm": 3.024386405944824,
|
| 1773 |
+
"learning_rate": 1.3417599122003464e-06,
|
| 1774 |
+
"logits/chosen": 2.7146401405334473,
|
| 1775 |
+
"logits/rejected": 2.9919748306274414,
|
| 1776 |
+
"logps/chosen": -3.507610321044922,
|
| 1777 |
+
"logps/rejected": -9.035795211791992,
|
| 1778 |
+
"loss": 0.2122,
|
| 1779 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1780 |
+
"rewards/chosen": -5.261415004730225,
|
| 1781 |
+
"rewards/margins": 8.292278289794922,
|
| 1782 |
+
"rewards/rejected": -13.553693771362305,
|
| 1783 |
+
"step": 980
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 0.7534246575342466,
|
| 1787 |
+
"grad_norm": 3.5992636680603027,
|
| 1788 |
+
"learning_rate": 1.2956158147457116e-06,
|
| 1789 |
+
"logits/chosen": 2.9386534690856934,
|
| 1790 |
+
"logits/rejected": 2.974626064300537,
|
| 1791 |
+
"logps/chosen": -3.5809288024902344,
|
| 1792 |
+
"logps/rejected": -9.47168254852295,
|
| 1793 |
+
"loss": 0.2058,
|
| 1794 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1795 |
+
"rewards/chosen": -5.371392726898193,
|
| 1796 |
+
"rewards/margins": 8.836130142211914,
|
| 1797 |
+
"rewards/rejected": -14.20752239227295,
|
| 1798 |
+
"step": 990
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 0.76103500761035,
|
| 1802 |
+
"grad_norm": 3.0734636783599854,
|
| 1803 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1804 |
+
"logits/chosen": 2.984003782272339,
|
| 1805 |
+
"logits/rejected": 3.055147647857666,
|
| 1806 |
+
"logps/chosen": -4.086684226989746,
|
| 1807 |
+
"logps/rejected": -9.243562698364258,
|
| 1808 |
+
"loss": 0.2135,
|
| 1809 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1810 |
+
"rewards/chosen": -6.130026817321777,
|
| 1811 |
+
"rewards/margins": 7.735315799713135,
|
| 1812 |
+
"rewards/rejected": -13.865341186523438,
|
| 1813 |
+
"step": 1000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.76103500761035,
|
| 1817 |
+
"eval_logits/chosen": 3.18290638923645,
|
| 1818 |
+
"eval_logits/rejected": 3.2487096786499023,
|
| 1819 |
+
"eval_logps/chosen": -3.576455593109131,
|
| 1820 |
+
"eval_logps/rejected": -8.458187103271484,
|
| 1821 |
+
"eval_loss": 0.20949731767177582,
|
| 1822 |
+
"eval_rewards/accuracies": 0.9439252614974976,
|
| 1823 |
+
"eval_rewards/chosen": -5.364683151245117,
|
| 1824 |
+
"eval_rewards/margins": 7.322596549987793,
|
| 1825 |
+
"eval_rewards/rejected": -12.687278747558594,
|
| 1826 |
+
"eval_runtime": 30.7759,
|
| 1827 |
+
"eval_samples_per_second": 27.619,
|
| 1828 |
+
"eval_steps_per_second": 3.477,
|
| 1829 |
+
"step": 1000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.7686453576864536,
|
| 1833 |
+
"grad_norm": 4.717663288116455,
|
| 1834 |
+
"learning_rate": 1.204932476567175e-06,
|
| 1835 |
+
"logits/chosen": 2.96572208404541,
|
| 1836 |
+
"logits/rejected": 2.8679356575012207,
|
| 1837 |
+
"logps/chosen": -3.6380767822265625,
|
| 1838 |
+
"logps/rejected": -8.982633590698242,
|
| 1839 |
+
"loss": 0.1686,
|
| 1840 |
+
"rewards/accuracies": 0.9375,
|
| 1841 |
+
"rewards/chosen": -5.457114219665527,
|
| 1842 |
+
"rewards/margins": 8.01683521270752,
|
| 1843 |
+
"rewards/rejected": -13.473950386047363,
|
| 1844 |
+
"step": 1010
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.776255707762557,
|
| 1848 |
+
"grad_norm": 2.1752161979675293,
|
| 1849 |
+
"learning_rate": 1.160433012552508e-06,
|
| 1850 |
+
"logits/chosen": 2.227342128753662,
|
| 1851 |
+
"logits/rejected": 2.576072931289673,
|
| 1852 |
+
"logps/chosen": -3.2803902626037598,
|
| 1853 |
+
"logps/rejected": -8.933613777160645,
|
| 1854 |
+
"loss": 0.1802,
|
| 1855 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1856 |
+
"rewards/chosen": -4.920585632324219,
|
| 1857 |
+
"rewards/margins": 8.479835510253906,
|
| 1858 |
+
"rewards/rejected": -13.400421142578125,
|
| 1859 |
+
"step": 1020
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.7838660578386606,
|
| 1863 |
+
"grad_norm": 3.252668857574463,
|
| 1864 |
+
"learning_rate": 1.11652112689164e-06,
|
| 1865 |
+
"logits/chosen": 2.558826446533203,
|
| 1866 |
+
"logits/rejected": 2.4980645179748535,
|
| 1867 |
+
"logps/chosen": -3.639244794845581,
|
| 1868 |
+
"logps/rejected": -9.068904876708984,
|
| 1869 |
+
"loss": 0.2142,
|
| 1870 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1871 |
+
"rewards/chosen": -5.458867073059082,
|
| 1872 |
+
"rewards/margins": 8.144488334655762,
|
| 1873 |
+
"rewards/rejected": -13.603355407714844,
|
| 1874 |
+
"step": 1030
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.791476407914764,
|
| 1878 |
+
"grad_norm": 4.632419109344482,
|
| 1879 |
+
"learning_rate": 1.073216080788921e-06,
|
| 1880 |
+
"logits/chosen": 1.8568460941314697,
|
| 1881 |
+
"logits/rejected": 2.2069149017333984,
|
| 1882 |
+
"logps/chosen": -3.4306697845458984,
|
| 1883 |
+
"logps/rejected": -8.670662879943848,
|
| 1884 |
+
"loss": 0.169,
|
| 1885 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1886 |
+
"rewards/chosen": -5.146004676818848,
|
| 1887 |
+
"rewards/margins": 7.859990119934082,
|
| 1888 |
+
"rewards/rejected": -13.00599479675293,
|
| 1889 |
+
"step": 1040
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.7990867579908676,
|
| 1893 |
+
"grad_norm": 4.621112823486328,
|
| 1894 |
+
"learning_rate": 1.0305368692688175e-06,
|
| 1895 |
+
"logits/chosen": 2.480006694793701,
|
| 1896 |
+
"logits/rejected": 2.7665817737579346,
|
| 1897 |
+
"logps/chosen": -3.486351728439331,
|
| 1898 |
+
"logps/rejected": -8.768314361572266,
|
| 1899 |
+
"loss": 0.1901,
|
| 1900 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1901 |
+
"rewards/chosen": -5.229527473449707,
|
| 1902 |
+
"rewards/margins": 7.922944068908691,
|
| 1903 |
+
"rewards/rejected": -13.152471542358398,
|
| 1904 |
+
"step": 1050
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.7990867579908676,
|
| 1908 |
+
"eval_logits/chosen": 3.231480598449707,
|
| 1909 |
+
"eval_logits/rejected": 3.311223030090332,
|
| 1910 |
+
"eval_logps/chosen": -3.6841485500335693,
|
| 1911 |
+
"eval_logps/rejected": -8.640739440917969,
|
| 1912 |
+
"eval_loss": 0.20705747604370117,
|
| 1913 |
+
"eval_rewards/accuracies": 0.9252336621284485,
|
| 1914 |
+
"eval_rewards/chosen": -5.526222229003906,
|
| 1915 |
+
"eval_rewards/margins": 7.434886455535889,
|
| 1916 |
+
"eval_rewards/rejected": -12.961106300354004,
|
| 1917 |
+
"eval_runtime": 30.7718,
|
| 1918 |
+
"eval_samples_per_second": 27.623,
|
| 1919 |
+
"eval_steps_per_second": 3.477,
|
| 1920 |
+
"step": 1050
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.806697108066971,
|
| 1924 |
+
"grad_norm": 3.463608503341675,
|
| 1925 |
+
"learning_rate": 9.88502212844063e-07,
|
| 1926 |
+
"logits/chosen": 2.379103183746338,
|
| 1927 |
+
"logits/rejected": 2.5155222415924072,
|
| 1928 |
+
"logps/chosen": -3.9454503059387207,
|
| 1929 |
+
"logps/rejected": -9.293844223022461,
|
| 1930 |
+
"loss": 0.1937,
|
| 1931 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1932 |
+
"rewards/chosen": -5.918175220489502,
|
| 1933 |
+
"rewards/margins": 8.022592544555664,
|
| 1934 |
+
"rewards/rejected": -13.940768241882324,
|
| 1935 |
+
"step": 1060
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.8143074581430746,
|
| 1939 |
+
"grad_norm": 5.665669918060303,
|
| 1940 |
+
"learning_rate": 9.471305493042243e-07,
|
| 1941 |
+
"logits/chosen": 2.449711322784424,
|
| 1942 |
+
"logits/rejected": 2.758481502532959,
|
| 1943 |
+
"logps/chosen": -3.280363082885742,
|
| 1944 |
+
"logps/rejected": -8.576845169067383,
|
| 1945 |
+
"loss": 0.2166,
|
| 1946 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1947 |
+
"rewards/chosen": -4.920544624328613,
|
| 1948 |
+
"rewards/margins": 7.9447221755981445,
|
| 1949 |
+
"rewards/rejected": -12.865266799926758,
|
| 1950 |
+
"step": 1070
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 0.821917808219178,
|
| 1954 |
+
"grad_norm": 5.21759033203125,
|
| 1955 |
+
"learning_rate": 9.064400256282757e-07,
|
| 1956 |
+
"logits/chosen": 2.5535693168640137,
|
| 1957 |
+
"logits/rejected": 2.780562162399292,
|
| 1958 |
+
"logps/chosen": -3.6120028495788574,
|
| 1959 |
+
"logps/rejected": -9.69387435913086,
|
| 1960 |
+
"loss": 0.1694,
|
| 1961 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1962 |
+
"rewards/chosen": -5.418004035949707,
|
| 1963 |
+
"rewards/margins": 9.122807502746582,
|
| 1964 |
+
"rewards/rejected": -14.540811538696289,
|
| 1965 |
+
"step": 1080
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.8295281582952816,
|
| 1969 |
+
"grad_norm": 3.6049561500549316,
|
| 1970 |
+
"learning_rate": 8.664484900247363e-07,
|
| 1971 |
+
"logits/chosen": 3.09112811088562,
|
| 1972 |
+
"logits/rejected": 3.2834911346435547,
|
| 1973 |
+
"logps/chosen": -3.5656745433807373,
|
| 1974 |
+
"logps/rejected": -8.692608833312988,
|
| 1975 |
+
"loss": 0.1771,
|
| 1976 |
+
"rewards/accuracies": 0.9375,
|
| 1977 |
+
"rewards/chosen": -5.348511695861816,
|
| 1978 |
+
"rewards/margins": 7.69040060043335,
|
| 1979 |
+
"rewards/rejected": -13.038912773132324,
|
| 1980 |
+
"step": 1090
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 0.837138508371385,
|
| 1984 |
+
"grad_norm": 6.876165390014648,
|
| 1985 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1986 |
+
"logits/chosen": 1.9753879308700562,
|
| 1987 |
+
"logits/rejected": 2.3351285457611084,
|
| 1988 |
+
"logps/chosen": -3.95485258102417,
|
| 1989 |
+
"logps/rejected": -9.459092140197754,
|
| 1990 |
+
"loss": 0.2078,
|
| 1991 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1992 |
+
"rewards/chosen": -5.932278633117676,
|
| 1993 |
+
"rewards/margins": 8.256359100341797,
|
| 1994 |
+
"rewards/rejected": -14.188638687133789,
|
| 1995 |
+
"step": 1100
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 0.837138508371385,
|
| 1999 |
+
"eval_logits/chosen": 3.2720582485198975,
|
| 2000 |
+
"eval_logits/rejected": 3.3373920917510986,
|
| 2001 |
+
"eval_logps/chosen": -3.704756498336792,
|
| 2002 |
+
"eval_logps/rejected": -8.690947532653809,
|
| 2003 |
+
"eval_loss": 0.20036298036575317,
|
| 2004 |
+
"eval_rewards/accuracies": 0.9439252614974976,
|
| 2005 |
+
"eval_rewards/chosen": -5.557135105133057,
|
| 2006 |
+
"eval_rewards/margins": 7.479285717010498,
|
| 2007 |
+
"eval_rewards/rejected": -13.036419868469238,
|
| 2008 |
+
"eval_runtime": 30.7733,
|
| 2009 |
+
"eval_samples_per_second": 27.621,
|
| 2010 |
+
"eval_steps_per_second": 3.477,
|
| 2011 |
+
"step": 1100
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.8447488584474886,
|
| 2015 |
+
"grad_norm": 5.0880126953125,
|
| 2016 |
+
"learning_rate": 7.886322351782782e-07,
|
| 2017 |
+
"logits/chosen": 2.924659013748169,
|
| 2018 |
+
"logits/rejected": 3.131314277648926,
|
| 2019 |
+
"logps/chosen": -3.665433883666992,
|
| 2020 |
+
"logps/rejected": -9.120814323425293,
|
| 2021 |
+
"loss": 0.1727,
|
| 2022 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 2023 |
+
"rewards/chosen": -5.4981513023376465,
|
| 2024 |
+
"rewards/margins": 8.183070182800293,
|
| 2025 |
+
"rewards/rejected": -13.681221008300781,
|
| 2026 |
+
"step": 1110
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.852359208523592,
|
| 2030 |
+
"grad_norm": 2.2846970558166504,
|
| 2031 |
+
"learning_rate": 7.508416487165862e-07,
|
| 2032 |
+
"logits/chosen": 1.7218126058578491,
|
| 2033 |
+
"logits/rejected": 2.0311954021453857,
|
| 2034 |
+
"logps/chosen": -3.7894043922424316,
|
| 2035 |
+
"logps/rejected": -10.045234680175781,
|
| 2036 |
+
"loss": 0.1494,
|
| 2037 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2038 |
+
"rewards/chosen": -5.684107303619385,
|
| 2039 |
+
"rewards/margins": 9.383744239807129,
|
| 2040 |
+
"rewards/rejected": -15.067851066589355,
|
| 2041 |
+
"step": 1120
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 0.8599695585996956,
|
| 2045 |
+
"grad_norm": 2.9382874965667725,
|
| 2046 |
+
"learning_rate": 7.138183009179922e-07,
|
| 2047 |
+
"logits/chosen": 4.044937610626221,
|
| 2048 |
+
"logits/rejected": 3.507354736328125,
|
| 2049 |
+
"logps/chosen": -3.7679309844970703,
|
| 2050 |
+
"logps/rejected": -8.613094329833984,
|
| 2051 |
+
"loss": 0.2204,
|
| 2052 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2053 |
+
"rewards/chosen": -5.6518964767456055,
|
| 2054 |
+
"rewards/margins": 7.267745018005371,
|
| 2055 |
+
"rewards/rejected": -12.919641494750977,
|
| 2056 |
+
"step": 1130
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 0.867579908675799,
|
| 2060 |
+
"grad_norm": 3.7730531692504883,
|
| 2061 |
+
"learning_rate": 6.775784314464717e-07,
|
| 2062 |
+
"logits/chosen": 2.7572951316833496,
|
| 2063 |
+
"logits/rejected": 2.948090076446533,
|
| 2064 |
+
"logps/chosen": -3.9021010398864746,
|
| 2065 |
+
"logps/rejected": -8.59765625,
|
| 2066 |
+
"loss": 0.177,
|
| 2067 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2068 |
+
"rewards/chosen": -5.853151798248291,
|
| 2069 |
+
"rewards/margins": 7.043332576751709,
|
| 2070 |
+
"rewards/rejected": -12.896484375,
|
| 2071 |
+
"step": 1140
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 0.8751902587519026,
|
| 2075 |
+
"grad_norm": 6.001643657684326,
|
| 2076 |
+
"learning_rate": 6.421379363065142e-07,
|
| 2077 |
+
"logits/chosen": 2.880664825439453,
|
| 2078 |
+
"logits/rejected": 2.9854893684387207,
|
| 2079 |
+
"logps/chosen": -3.7101027965545654,
|
| 2080 |
+
"logps/rejected": -9.663796424865723,
|
| 2081 |
+
"loss": 0.1909,
|
| 2082 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 2083 |
+
"rewards/chosen": -5.565154075622559,
|
| 2084 |
+
"rewards/margins": 8.93053913116455,
|
| 2085 |
+
"rewards/rejected": -14.495694160461426,
|
| 2086 |
+
"step": 1150
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 0.8751902587519026,
|
| 2090 |
+
"eval_logits/chosen": 3.2326529026031494,
|
| 2091 |
+
"eval_logits/rejected": 3.3147149085998535,
|
| 2092 |
+
"eval_logps/chosen": -3.758188486099243,
|
| 2093 |
+
"eval_logps/rejected": -8.758670806884766,
|
| 2094 |
+
"eval_loss": 0.19985900819301605,
|
| 2095 |
+
"eval_rewards/accuracies": 0.9439252614974976,
|
| 2096 |
+
"eval_rewards/chosen": -5.637282371520996,
|
| 2097 |
+
"eval_rewards/margins": 7.500723361968994,
|
| 2098 |
+
"eval_rewards/rejected": -13.138005256652832,
|
| 2099 |
+
"eval_runtime": 30.7775,
|
| 2100 |
+
"eval_samples_per_second": 27.618,
|
| 2101 |
+
"eval_steps_per_second": 3.477,
|
| 2102 |
+
"step": 1150
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.882800608828006,
|
| 2106 |
+
"grad_norm": 2.555748224258423,
|
| 2107 |
+
"learning_rate": 6.075123608706093e-07,
|
| 2108 |
+
"logits/chosen": 2.5560011863708496,
|
| 2109 |
+
"logits/rejected": 2.8420631885528564,
|
| 2110 |
+
"logps/chosen": -3.545081377029419,
|
| 2111 |
+
"logps/rejected": -9.308977127075195,
|
| 2112 |
+
"loss": 0.1999,
|
| 2113 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2114 |
+
"rewards/chosen": -5.317622184753418,
|
| 2115 |
+
"rewards/margins": 8.645845413208008,
|
| 2116 |
+
"rewards/rejected": -13.963467597961426,
|
| 2117 |
+
"step": 1160
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.8904109589041096,
|
| 2121 |
+
"grad_norm": 2.0732498168945312,
|
| 2122 |
+
"learning_rate": 5.737168930605272e-07,
|
| 2123 |
+
"logits/chosen": 3.55595326423645,
|
| 2124 |
+
"logits/rejected": 3.61430025100708,
|
| 2125 |
+
"logps/chosen": -3.8471646308898926,
|
| 2126 |
+
"logps/rejected": -9.333372116088867,
|
| 2127 |
+
"loss": 0.2348,
|
| 2128 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2129 |
+
"rewards/chosen": -5.770747661590576,
|
| 2130 |
+
"rewards/margins": 8.229310989379883,
|
| 2131 |
+
"rewards/rejected": -14.0000581741333,
|
| 2132 |
+
"step": 1170
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.898021308980213,
|
| 2136 |
+
"grad_norm": 3.7742130756378174,
|
| 2137 |
+
"learning_rate": 5.407663566854008e-07,
|
| 2138 |
+
"logits/chosen": 3.0770957469940186,
|
| 2139 |
+
"logits/rejected": 3.1060760021209717,
|
| 2140 |
+
"logps/chosen": -3.524754285812378,
|
| 2141 |
+
"logps/rejected": -9.05833625793457,
|
| 2142 |
+
"loss": 0.1691,
|
| 2143 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 2144 |
+
"rewards/chosen": -5.2871317863464355,
|
| 2145 |
+
"rewards/margins": 8.300371170043945,
|
| 2146 |
+
"rewards/rejected": -13.587503433227539,
|
| 2147 |
+
"step": 1180
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 0.9056316590563166,
|
| 2151 |
+
"grad_norm": 3.0568952560424805,
|
| 2152 |
+
"learning_rate": 5.086752049395094e-07,
|
| 2153 |
+
"logits/chosen": 2.4866814613342285,
|
| 2154 |
+
"logits/rejected": 2.541372537612915,
|
| 2155 |
+
"logps/chosen": -3.6101276874542236,
|
| 2156 |
+
"logps/rejected": -8.999945640563965,
|
| 2157 |
+
"loss": 0.1596,
|
| 2158 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2159 |
+
"rewards/chosen": -5.415192127227783,
|
| 2160 |
+
"rewards/margins": 8.08472728729248,
|
| 2161 |
+
"rewards/rejected": -13.499918937683105,
|
| 2162 |
+
"step": 1190
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.91324200913242,
|
| 2166 |
+
"grad_norm": 2.707932949066162,
|
| 2167 |
+
"learning_rate": 4.774575140626317e-07,
|
| 2168 |
+
"logits/chosen": 3.445946216583252,
|
| 2169 |
+
"logits/rejected": 3.263925075531006,
|
| 2170 |
+
"logps/chosen": -4.120098114013672,
|
| 2171 |
+
"logps/rejected": -9.42192554473877,
|
| 2172 |
+
"loss": 0.1671,
|
| 2173 |
+
"rewards/accuracies": 0.9375,
|
| 2174 |
+
"rewards/chosen": -6.18014669418335,
|
| 2175 |
+
"rewards/margins": 7.952740669250488,
|
| 2176 |
+
"rewards/rejected": -14.13288688659668,
|
| 2177 |
+
"step": 1200
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 0.91324200913242,
|
| 2181 |
+
"eval_logits/chosen": 3.2238974571228027,
|
| 2182 |
+
"eval_logits/rejected": 3.331603527069092,
|
| 2183 |
+
"eval_logps/chosen": -3.8190717697143555,
|
| 2184 |
+
"eval_logps/rejected": -8.885968208312988,
|
| 2185 |
+
"eval_loss": 0.1992855668067932,
|
| 2186 |
+
"eval_rewards/accuracies": 0.9532710313796997,
|
| 2187 |
+
"eval_rewards/chosen": -5.728607654571533,
|
| 2188 |
+
"eval_rewards/margins": 7.600344181060791,
|
| 2189 |
+
"eval_rewards/rejected": -13.328951835632324,
|
| 2190 |
+
"eval_runtime": 30.7793,
|
| 2191 |
+
"eval_samples_per_second": 27.616,
|
| 2192 |
+
"eval_steps_per_second": 3.476,
|
| 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": 2.7939277248533627e+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 sha256:41f4cd2f804ea1e74c05047097e9244ba2f0dfdda24bfccd07e006f56a3fae84
|
| 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)
|