Upload folder using huggingface_hub
Browse files- README.md +202 -0
- adapter_config.json +31 -0
- adapter_model.safetensors +3 -0
- additional_config.json +1 -0
- args.json +438 -0
- latest +1 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2734 -0
- training_args.bin +3 -0
- zero_to_fp32.py +760 -0
README.md
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: meta-llama/Llama-3.2-11B-Vision-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 0.15.2
|
adapter_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"corda_config": null,
|
| 7 |
+
"eva_config": null,
|
| 8 |
+
"exclude_modules": null,
|
| 9 |
+
"fan_in_fan_out": false,
|
| 10 |
+
"inference_mode": true,
|
| 11 |
+
"init_lora_weights": true,
|
| 12 |
+
"layer_replication": null,
|
| 13 |
+
"layers_pattern": null,
|
| 14 |
+
"layers_to_transform": null,
|
| 15 |
+
"loftq_config": {},
|
| 16 |
+
"lora_alpha": 32,
|
| 17 |
+
"lora_bias": false,
|
| 18 |
+
"lora_dropout": 0.05,
|
| 19 |
+
"megatron_config": null,
|
| 20 |
+
"megatron_core": "megatron.core",
|
| 21 |
+
"modules_to_save": [],
|
| 22 |
+
"peft_type": "LORA",
|
| 23 |
+
"r": 8,
|
| 24 |
+
"rank_pattern": {},
|
| 25 |
+
"revision": null,
|
| 26 |
+
"target_modules": "^(language_model).*\\.(gate_proj|o_proj|q_proj|up_proj|v_proj|k_proj|down_proj)$",
|
| 27 |
+
"task_type": "CAUSAL_LM",
|
| 28 |
+
"trainable_token_indices": null,
|
| 29 |
+
"use_dora": false,
|
| 30 |
+
"use_rslora": false
|
| 31 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58c9367f703645f1c495252e5787bb7d8a95e3f5bd46091faafc197eae66bb1f
|
| 3 |
+
size 52511776
|
additional_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}
|
args.json
ADDED
|
@@ -0,0 +1,438 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
|
| 3 |
+
"model_type": "llama3_2_vision",
|
| 4 |
+
"model_revision": null,
|
| 5 |
+
"task_type": "causal_lm",
|
| 6 |
+
"torch_dtype": "bfloat16",
|
| 7 |
+
"attn_impl": null,
|
| 8 |
+
"num_labels": null,
|
| 9 |
+
"problem_type": null,
|
| 10 |
+
"rope_scaling": null,
|
| 11 |
+
"device_map": null,
|
| 12 |
+
"max_memory": {},
|
| 13 |
+
"local_repo_path": null,
|
| 14 |
+
"template": "llama3_2_vision",
|
| 15 |
+
"system": "You are a helpful assistant that truthfully answers user questions about the provided image.",
|
| 16 |
+
"max_length": 4096,
|
| 17 |
+
"truncation_strategy": "delete",
|
| 18 |
+
"max_pixels": null,
|
| 19 |
+
"agent_template": null,
|
| 20 |
+
"norm_bbox": null,
|
| 21 |
+
"response_prefix": null,
|
| 22 |
+
"padding_side": "right",
|
| 23 |
+
"loss_scale": "last_round",
|
| 24 |
+
"sequence_parallel_size": 1,
|
| 25 |
+
"use_chat_template": true,
|
| 26 |
+
"template_backend": "swift",
|
| 27 |
+
"dataset": [
|
| 28 |
+
"/data/kdd/crag/cjz/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat.json"
|
| 29 |
+
],
|
| 30 |
+
"val_dataset": [],
|
| 31 |
+
"split_dataset_ratio": 0.01,
|
| 32 |
+
"data_seed": 42,
|
| 33 |
+
"dataset_num_proc": 4,
|
| 34 |
+
"dataset_shuffle": true,
|
| 35 |
+
"val_dataset_shuffle": false,
|
| 36 |
+
"streaming": false,
|
| 37 |
+
"interleave_prob": null,
|
| 38 |
+
"stopping_strategy": "first_exhausted",
|
| 39 |
+
"shuffle_buffer_size": 1000,
|
| 40 |
+
"enable_cache": false,
|
| 41 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 42 |
+
"columns": {},
|
| 43 |
+
"strict": false,
|
| 44 |
+
"remove_unused_columns": true,
|
| 45 |
+
"model_name": [
|
| 46 |
+
null,
|
| 47 |
+
null
|
| 48 |
+
],
|
| 49 |
+
"model_author": [
|
| 50 |
+
null,
|
| 51 |
+
null
|
| 52 |
+
],
|
| 53 |
+
"custom_dataset_info": [],
|
| 54 |
+
"quant_method": null,
|
| 55 |
+
"quant_bits": null,
|
| 56 |
+
"hqq_axis": null,
|
| 57 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 58 |
+
"bnb_4bit_quant_type": "nf4",
|
| 59 |
+
"bnb_4bit_use_double_quant": true,
|
| 60 |
+
"bnb_4bit_quant_storage": null,
|
| 61 |
+
"max_new_tokens": 64,
|
| 62 |
+
"temperature": 0.9,
|
| 63 |
+
"top_k": 50,
|
| 64 |
+
"top_p": 0.9,
|
| 65 |
+
"repetition_penalty": 1.0,
|
| 66 |
+
"num_beams": 1,
|
| 67 |
+
"stream": false,
|
| 68 |
+
"stop_words": [],
|
| 69 |
+
"logprobs": false,
|
| 70 |
+
"top_logprobs": null,
|
| 71 |
+
"ckpt_dir": null,
|
| 72 |
+
"load_dataset_config": null,
|
| 73 |
+
"lora_modules": [],
|
| 74 |
+
"tuner_backend": "peft",
|
| 75 |
+
"train_type": "lora",
|
| 76 |
+
"adapters": [],
|
| 77 |
+
"external_plugins": [],
|
| 78 |
+
"seed": 42,
|
| 79 |
+
"model_kwargs": {},
|
| 80 |
+
"load_args": false,
|
| 81 |
+
"load_data_args": false,
|
| 82 |
+
"use_hf": false,
|
| 83 |
+
"hub_token": null,
|
| 84 |
+
"custom_register_path": [],
|
| 85 |
+
"ignore_args_error": false,
|
| 86 |
+
"use_swift_lora": false,
|
| 87 |
+
"output_dir": "/data2/kdd/crag/cjz/output/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat/v1-20250605-104410",
|
| 88 |
+
"overwrite_output_dir": false,
|
| 89 |
+
"do_train": false,
|
| 90 |
+
"do_eval": false,
|
| 91 |
+
"do_predict": false,
|
| 92 |
+
"eval_strategy": "steps",
|
| 93 |
+
"prediction_loss_only": false,
|
| 94 |
+
"per_device_train_batch_size": 1,
|
| 95 |
+
"per_device_eval_batch_size": 1,
|
| 96 |
+
"per_gpu_train_batch_size": null,
|
| 97 |
+
"per_gpu_eval_batch_size": null,
|
| 98 |
+
"gradient_accumulation_steps": 32,
|
| 99 |
+
"eval_accumulation_steps": null,
|
| 100 |
+
"eval_delay": 0,
|
| 101 |
+
"torch_empty_cache_steps": null,
|
| 102 |
+
"learning_rate": 0.0002,
|
| 103 |
+
"weight_decay": 0.1,
|
| 104 |
+
"adam_beta1": 0.9,
|
| 105 |
+
"adam_beta2": 0.95,
|
| 106 |
+
"adam_epsilon": 1e-08,
|
| 107 |
+
"max_grad_norm": 1.0,
|
| 108 |
+
"num_train_epochs": 3.0,
|
| 109 |
+
"max_steps": -1,
|
| 110 |
+
"lr_scheduler_type": "cosine",
|
| 111 |
+
"lr_scheduler_kwargs": null,
|
| 112 |
+
"warmup_ratio": 0.05,
|
| 113 |
+
"warmup_steps": 0,
|
| 114 |
+
"log_level": "passive",
|
| 115 |
+
"log_level_replica": "warning",
|
| 116 |
+
"log_on_each_node": true,
|
| 117 |
+
"logging_dir": "/data2/kdd/crag/cjz/output/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat/v1-20250605-104410/runs",
|
| 118 |
+
"logging_strategy": "steps",
|
| 119 |
+
"logging_first_step": true,
|
| 120 |
+
"logging_steps": 1,
|
| 121 |
+
"logging_nan_inf_filter": true,
|
| 122 |
+
"save_strategy": "steps",
|
| 123 |
+
"save_steps": 50.0,
|
| 124 |
+
"save_total_limit": null,
|
| 125 |
+
"save_safetensors": true,
|
| 126 |
+
"save_on_each_node": false,
|
| 127 |
+
"save_only_model": false,
|
| 128 |
+
"restore_callback_states_from_checkpoint": false,
|
| 129 |
+
"no_cuda": false,
|
| 130 |
+
"use_cpu": false,
|
| 131 |
+
"use_mps_device": false,
|
| 132 |
+
"jit_mode_eval": false,
|
| 133 |
+
"use_ipex": false,
|
| 134 |
+
"bf16": true,
|
| 135 |
+
"fp16": false,
|
| 136 |
+
"fp16_opt_level": "O1",
|
| 137 |
+
"half_precision_backend": "auto",
|
| 138 |
+
"bf16_full_eval": false,
|
| 139 |
+
"fp16_full_eval": false,
|
| 140 |
+
"tf32": null,
|
| 141 |
+
"local_rank": 0,
|
| 142 |
+
"ddp_backend": null,
|
| 143 |
+
"tpu_num_cores": null,
|
| 144 |
+
"tpu_metrics_debug": false,
|
| 145 |
+
"debug": null,
|
| 146 |
+
"dataloader_drop_last": false,
|
| 147 |
+
"eval_steps": 300.0,
|
| 148 |
+
"dataloader_num_workers": 4,
|
| 149 |
+
"dataloader_prefetch_factor": null,
|
| 150 |
+
"past_index": -1,
|
| 151 |
+
"run_name": null,
|
| 152 |
+
"disable_tqdm": null,
|
| 153 |
+
"label_names": null,
|
| 154 |
+
"load_best_model_at_end": false,
|
| 155 |
+
"metric_for_best_model": "loss",
|
| 156 |
+
"greater_is_better": false,
|
| 157 |
+
"ignore_data_skip": false,
|
| 158 |
+
"fsdp": "",
|
| 159 |
+
"fsdp_min_num_params": 0,
|
| 160 |
+
"fsdp_config": null,
|
| 161 |
+
"tp_size": 0,
|
| 162 |
+
"fsdp_transformer_layer_cls_to_wrap": null,
|
| 163 |
+
"accelerator_config": {
|
| 164 |
+
"dispatch_batches": false
|
| 165 |
+
},
|
| 166 |
+
"deepspeed": {
|
| 167 |
+
"fp16": {
|
| 168 |
+
"enabled": "auto",
|
| 169 |
+
"loss_scale": 0,
|
| 170 |
+
"loss_scale_window": 1000,
|
| 171 |
+
"initial_scale_power": 16,
|
| 172 |
+
"hysteresis": 2,
|
| 173 |
+
"min_loss_scale": 1
|
| 174 |
+
},
|
| 175 |
+
"bf16": {
|
| 176 |
+
"enabled": "auto"
|
| 177 |
+
},
|
| 178 |
+
"zero_optimization": {
|
| 179 |
+
"stage": 2,
|
| 180 |
+
"offload_optimizer": {
|
| 181 |
+
"device": "none",
|
| 182 |
+
"pin_memory": true
|
| 183 |
+
},
|
| 184 |
+
"allgather_partitions": true,
|
| 185 |
+
"allgather_bucket_size": 200000000.0,
|
| 186 |
+
"overlap_comm": false,
|
| 187 |
+
"reduce_scatter": true,
|
| 188 |
+
"reduce_bucket_size": 200000000.0,
|
| 189 |
+
"contiguous_gradients": true
|
| 190 |
+
},
|
| 191 |
+
"gradient_accumulation_steps": "auto",
|
| 192 |
+
"gradient_clipping": "auto",
|
| 193 |
+
"steps_per_print": 2000,
|
| 194 |
+
"train_batch_size": "auto",
|
| 195 |
+
"train_micro_batch_size_per_gpu": "auto",
|
| 196 |
+
"wall_clock_breakdown": false
|
| 197 |
+
},
|
| 198 |
+
"label_smoothing_factor": 0.0,
|
| 199 |
+
"optim": "adamw_torch",
|
| 200 |
+
"optim_args": null,
|
| 201 |
+
"adafactor": false,
|
| 202 |
+
"group_by_length": false,
|
| 203 |
+
"length_column_name": "length",
|
| 204 |
+
"report_to": [
|
| 205 |
+
"tensorboard"
|
| 206 |
+
],
|
| 207 |
+
"ddp_find_unused_parameters": null,
|
| 208 |
+
"ddp_bucket_cap_mb": null,
|
| 209 |
+
"ddp_broadcast_buffers": null,
|
| 210 |
+
"dataloader_pin_memory": true,
|
| 211 |
+
"dataloader_persistent_workers": false,
|
| 212 |
+
"skip_memory_metrics": true,
|
| 213 |
+
"use_legacy_prediction_loop": false,
|
| 214 |
+
"push_to_hub": false,
|
| 215 |
+
"resume_from_checkpoint": null,
|
| 216 |
+
"hub_model_id": null,
|
| 217 |
+
"hub_strategy": "every_save",
|
| 218 |
+
"hub_private_repo": null,
|
| 219 |
+
"hub_always_push": false,
|
| 220 |
+
"gradient_checkpointing": true,
|
| 221 |
+
"gradient_checkpointing_kwargs": null,
|
| 222 |
+
"include_inputs_for_metrics": false,
|
| 223 |
+
"include_for_metrics": [],
|
| 224 |
+
"eval_do_concat_batches": true,
|
| 225 |
+
"fp16_backend": "auto",
|
| 226 |
+
"push_to_hub_model_id": null,
|
| 227 |
+
"push_to_hub_organization": null,
|
| 228 |
+
"push_to_hub_token": null,
|
| 229 |
+
"mp_parameters": "",
|
| 230 |
+
"auto_find_batch_size": false,
|
| 231 |
+
"full_determinism": false,
|
| 232 |
+
"torchdynamo": null,
|
| 233 |
+
"ray_scope": "last",
|
| 234 |
+
"ddp_timeout": 1800,
|
| 235 |
+
"torch_compile": false,
|
| 236 |
+
"torch_compile_backend": null,
|
| 237 |
+
"torch_compile_mode": null,
|
| 238 |
+
"include_tokens_per_second": false,
|
| 239 |
+
"include_num_input_tokens_seen": false,
|
| 240 |
+
"neftune_noise_alpha": null,
|
| 241 |
+
"optim_target_modules": null,
|
| 242 |
+
"batch_eval_metrics": false,
|
| 243 |
+
"eval_on_start": false,
|
| 244 |
+
"use_liger_kernel": false,
|
| 245 |
+
"eval_use_gather_object": false,
|
| 246 |
+
"average_tokens_across_devices": false,
|
| 247 |
+
"sortish_sampler": false,
|
| 248 |
+
"predict_with_generate": false,
|
| 249 |
+
"generation_max_length": null,
|
| 250 |
+
"generation_num_beams": null,
|
| 251 |
+
"generation_config": null,
|
| 252 |
+
"check_model": true,
|
| 253 |
+
"acc_strategy": "token",
|
| 254 |
+
"train_dataloader_shuffle": true,
|
| 255 |
+
"metric_warmup_step": 0,
|
| 256 |
+
"fsdp_num": 1,
|
| 257 |
+
"acc_steps": 1,
|
| 258 |
+
"eval_use_evalscope": false,
|
| 259 |
+
"eval_datasets": [],
|
| 260 |
+
"eval_limit": null,
|
| 261 |
+
"eval_datasets_args": null,
|
| 262 |
+
"eval_generation_config": null,
|
| 263 |
+
"freeze_parameters": [
|
| 264 |
+
"vision_model",
|
| 265 |
+
"multi_modal_projector"
|
| 266 |
+
],
|
| 267 |
+
"freeze_parameters_ratio": 0.0,
|
| 268 |
+
"trainable_parameters": [],
|
| 269 |
+
"freeze_llm": false,
|
| 270 |
+
"freeze_vit": true,
|
| 271 |
+
"freeze_aligner": true,
|
| 272 |
+
"target_modules": [
|
| 273 |
+
"all-linear"
|
| 274 |
+
],
|
| 275 |
+
"target_regex": null,
|
| 276 |
+
"modules_to_save": [],
|
| 277 |
+
"lora_rank": 8,
|
| 278 |
+
"lora_alpha": 32,
|
| 279 |
+
"lora_dropout": 0.05,
|
| 280 |
+
"lora_bias": "none",
|
| 281 |
+
"lora_dtype": null,
|
| 282 |
+
"lorap_lr_ratio": null,
|
| 283 |
+
"use_rslora": false,
|
| 284 |
+
"use_dora": false,
|
| 285 |
+
"lora_ga_batch_size": 2,
|
| 286 |
+
"lora_ga_iters": 2,
|
| 287 |
+
"lora_ga_max_length": 1024,
|
| 288 |
+
"lora_ga_direction": "ArB2r",
|
| 289 |
+
"lora_ga_scale": "stable",
|
| 290 |
+
"lora_ga_stable_gamma": 16,
|
| 291 |
+
"init_weights": true,
|
| 292 |
+
"fourier_n_frequency": 2000,
|
| 293 |
+
"fourier_scaling": 300.0,
|
| 294 |
+
"boft_block_size": 4,
|
| 295 |
+
"boft_block_num": 0,
|
| 296 |
+
"boft_n_butterfly_factor": 1,
|
| 297 |
+
"boft_dropout": 0.0,
|
| 298 |
+
"vera_rank": 256,
|
| 299 |
+
"vera_projection_prng_key": 0,
|
| 300 |
+
"vera_dropout": 0.0,
|
| 301 |
+
"vera_d_initial": 0.1,
|
| 302 |
+
"adapter_act": "gelu",
|
| 303 |
+
"adapter_length": 128,
|
| 304 |
+
"use_galore": false,
|
| 305 |
+
"galore_target_modules": null,
|
| 306 |
+
"galore_rank": 128,
|
| 307 |
+
"galore_update_proj_gap": 50,
|
| 308 |
+
"galore_scale": 1.0,
|
| 309 |
+
"galore_proj_type": "std",
|
| 310 |
+
"galore_optim_per_parameter": false,
|
| 311 |
+
"galore_with_embedding": false,
|
| 312 |
+
"galore_quantization": false,
|
| 313 |
+
"galore_proj_quant": false,
|
| 314 |
+
"galore_proj_bits": 4,
|
| 315 |
+
"galore_proj_group_size": 256,
|
| 316 |
+
"galore_cos_threshold": 0.4,
|
| 317 |
+
"galore_gamma_proj": 2,
|
| 318 |
+
"galore_queue_size": 5,
|
| 319 |
+
"adalora_target_r": 8,
|
| 320 |
+
"adalora_init_r": 12,
|
| 321 |
+
"adalora_tinit": 0,
|
| 322 |
+
"adalora_tfinal": 0,
|
| 323 |
+
"adalora_deltaT": 1,
|
| 324 |
+
"adalora_beta1": 0.85,
|
| 325 |
+
"adalora_beta2": 0.85,
|
| 326 |
+
"adalora_orth_reg_weight": 0.5,
|
| 327 |
+
"llamapro_num_new_blocks": 4,
|
| 328 |
+
"llamapro_num_groups": null,
|
| 329 |
+
"lisa_activated_layers": 0,
|
| 330 |
+
"lisa_step_interval": 20,
|
| 331 |
+
"reft_layer_key": null,
|
| 332 |
+
"reft_layers": null,
|
| 333 |
+
"reft_rank": 4,
|
| 334 |
+
"reft_intervention_type": "LoreftIntervention",
|
| 335 |
+
"reft_args": null,
|
| 336 |
+
"swanlab_token": null,
|
| 337 |
+
"swanlab_project": null,
|
| 338 |
+
"swanlab_workspace": null,
|
| 339 |
+
"swanlab_exp_name": null,
|
| 340 |
+
"swanlab_mode": "cloud",
|
| 341 |
+
"add_version": true,
|
| 342 |
+
"resume_only_model": false,
|
| 343 |
+
"create_checkpoint_symlink": false,
|
| 344 |
+
"packing": false,
|
| 345 |
+
"lazy_tokenize": true,
|
| 346 |
+
"loss_type": "sigmoid",
|
| 347 |
+
"optimizer": null,
|
| 348 |
+
"metric": null,
|
| 349 |
+
"zero_hpz_partition_size": null,
|
| 350 |
+
"reward_model": null,
|
| 351 |
+
"reward_adapters": [],
|
| 352 |
+
"reward_model_type": null,
|
| 353 |
+
"reward_model_revision": null,
|
| 354 |
+
"num_ppo_epochs": 4,
|
| 355 |
+
"whiten_rewards": false,
|
| 356 |
+
"kl_coef": 0.05,
|
| 357 |
+
"cliprange": 0.2,
|
| 358 |
+
"vf_coef": 0.1,
|
| 359 |
+
"cliprange_value": 0.2,
|
| 360 |
+
"gamma": 1.0,
|
| 361 |
+
"lam": 0.95,
|
| 362 |
+
"num_mini_batches": 1,
|
| 363 |
+
"local_rollout_forward_batch_size": 64,
|
| 364 |
+
"num_sample_generations": 10,
|
| 365 |
+
"response_length": 512,
|
| 366 |
+
"missing_eos_penalty": null,
|
| 367 |
+
"epsilon": 0.2,
|
| 368 |
+
"epsilon_high": null,
|
| 369 |
+
"num_infer_workers": 1,
|
| 370 |
+
"vllm_device": [
|
| 371 |
+
"auto"
|
| 372 |
+
],
|
| 373 |
+
"vllm_gpu_memory_utilization": 0.9,
|
| 374 |
+
"vllm_max_model_len": null,
|
| 375 |
+
"vllm_max_num_seqs": 256,
|
| 376 |
+
"vllm_enforce_eager": false,
|
| 377 |
+
"vllm_limit_mm_per_prompt": null,
|
| 378 |
+
"vllm_enable_prefix_caching": true,
|
| 379 |
+
"cosine_min_len_value_wrong": -0.5,
|
| 380 |
+
"cosine_max_len_value_wrong": 0.0,
|
| 381 |
+
"cosine_min_len_value_correct": 1.0,
|
| 382 |
+
"cosine_max_len_value_correct": 0.5,
|
| 383 |
+
"cosine_max_len": null,
|
| 384 |
+
"repetition_n_grams": 3,
|
| 385 |
+
"repetition_max_penalty": -1.0,
|
| 386 |
+
"use_lmdeploy": false,
|
| 387 |
+
"lmdeploy_device": "auto",
|
| 388 |
+
"lmdeploy_session_len": null,
|
| 389 |
+
"lmdeploy_cache_max_entry_count": 0.8,
|
| 390 |
+
"async_generate": false,
|
| 391 |
+
"tensor_parallel_size": 1,
|
| 392 |
+
"sleep_level": 0,
|
| 393 |
+
"move_model_batches": null,
|
| 394 |
+
"offload_optimizer": false,
|
| 395 |
+
"offload_model": false,
|
| 396 |
+
"gc_collect_after_offload": false,
|
| 397 |
+
"multi_turn_func": null,
|
| 398 |
+
"dynamic_sample": false,
|
| 399 |
+
"max_resample_times": 3,
|
| 400 |
+
"overlong_filter": false,
|
| 401 |
+
"soft_max_length": null,
|
| 402 |
+
"soft_cache_length": null,
|
| 403 |
+
"scale_rewards": true,
|
| 404 |
+
"wandb_log_unique_prompts": null,
|
| 405 |
+
"vllm_server_host": null,
|
| 406 |
+
"vllm_server_port": 8000,
|
| 407 |
+
"vllm_server_timeout": 240.0,
|
| 408 |
+
"num_generations": 8,
|
| 409 |
+
"max_completion_length": 512,
|
| 410 |
+
"ds3_gather_for_generation": true,
|
| 411 |
+
"reward_funcs": [],
|
| 412 |
+
"reward_weights": null,
|
| 413 |
+
"log_completions": false,
|
| 414 |
+
"use_vllm": false,
|
| 415 |
+
"num_iterations": 1,
|
| 416 |
+
"rlhf_type": "dpo",
|
| 417 |
+
"ref_model": null,
|
| 418 |
+
"ref_model_type": null,
|
| 419 |
+
"ref_model_revision": null,
|
| 420 |
+
"beta": 0.1,
|
| 421 |
+
"label_smoothing": 0,
|
| 422 |
+
"rpo_alpha": 1.0,
|
| 423 |
+
"cpo_alpha": 1.0,
|
| 424 |
+
"simpo_gamma": 1,
|
| 425 |
+
"desirable_weight": 1.0,
|
| 426 |
+
"undesirable_weight": 1.0,
|
| 427 |
+
"center_rewards_coefficient": null,
|
| 428 |
+
"rank": 0,
|
| 429 |
+
"global_world_size": 1,
|
| 430 |
+
"local_world_size": 1,
|
| 431 |
+
"model_suffix": "Llama-3.2-11B-Vision-Instruct",
|
| 432 |
+
"model_info": "ModelInfo(model_type='llama3_2_vision', model_dir='/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct', torch_dtype=torch.bfloat16, max_model_len=131072, quant_method=None, quant_bits=None, rope_scaling={'factor': 8.0, 'high_freq_factor': 4.0, 'low_freq_factor': 1.0, 'original_max_position_embeddings': 8192, 'rope_type': 'llama3'}, config=None, task_type='causal_lm', num_labels=None)",
|
| 433 |
+
"model_meta": "ModelMeta(model_type='llama3_2_vision', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-11B-Vision-Instruct', hf_model_id='meta-llama/Llama-3.2-11B-Vision-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-90B-Vision-Instruct', hf_model_id='meta-llama/Llama-3.2-90B-Vision-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-11B-Vision', hf_model_id='meta-llama/Llama-3.2-11B-Vision', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-90B-Vision', hf_model_id='meta-llama/Llama-3.2-90B-Vision', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2_vision', get_function=<function get_model_tokenizer_llama3_2_vision at 0x73c800526d40>, model_arch='llama3_2_vision', architectures=['MllamaForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.45'], tags=[])",
|
| 434 |
+
"model_dir": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
|
| 435 |
+
"hub": "<class 'swift.hub.hub.MSHub'>",
|
| 436 |
+
"evaluation_strategy": "steps",
|
| 437 |
+
"training_args": "DPOConfig(output_dir='/data2/kdd/crag/cjz/output/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat/v1-20250605-104410', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=32, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0002, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/data2/kdd/crag/cjz/output/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat/v1-20250605-104410/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=None, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=300, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/data2/kdd/crag/cjz/output/task2_dpo_absgt10_taskgt4__pari0.5_0.2cheat/v1-20250605-104410', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, tp_size=0, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 2, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'allgather_partitions': True, 'allgather_bucket_size': 200000000.0, 'overlap_comm': False, 'reduce_scatter': True, 'reduce_bucket_size': 200000000.0, 'contiguous_gradients': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['tensorboard'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=None, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, model_init_kwargs=None, ref_model_init_kwargs=None, model_adapter_name=None, ref_adapter_name=None, force_use_ref_model=False, disable_dropout=True, use_logits_to_keep=False, dataset_num_proc=4, padding_value=None, label_pad_token_id=None, max_prompt_length=512, max_completion_length=512, max_length=4096, truncation_mode='keep_end', padding_free=False, precompute_ref_log_probs=False, precompute_ref_batch_size=None, tools=None, loss_type='sigmoid', beta=0.1, f_divergence_type=<FDivergenceType.REVERSE_KL: 'reverse_kl'>, f_alpha_divergence_coef=1.0, reference_free=False, label_smoothing=0, use_weighting=False, rpo_alpha=1.0, discopop_tau=0.05, sync_ref_model=False, ref_model_mixup_alpha=0.6, ref_model_sync_steps=512, generate_during_eval=False, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, metric_warmup_step=0, fsdp_num=1, acc_steps=1, eval_use_evalscope=False, eval_datasets=[], eval_limit=None, eval_datasets_args=None, eval_generation_config=None, train_type='lora', optimizer=None, local_repo_path=None, galore_config=None)"
|
| 438 |
+
}
|
latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step150
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:835f869ea325fd6edf27b48b589309fb66641cb92b45f2fc13d1bb6e8814106c
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:46c844d8813cd01ce0e8acd5ae87fe478319c1b2582465b49d4c3f463b2e17ce
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2734 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.35882484862076697,
|
| 6 |
+
"eval_steps": 300,
|
| 7 |
+
"global_step": 150,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00239216565747178,
|
| 14 |
+
"grad_norm": 13.569995880126953,
|
| 15 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 16 |
+
"logits/chosen": -0.6432992815971375,
|
| 17 |
+
"logits/rejected": -0.6486618518829346,
|
| 18 |
+
"logps/chosen": -10.907301902770996,
|
| 19 |
+
"logps/rejected": -27.245113372802734,
|
| 20 |
+
"loss": 1.6750074625015259,
|
| 21 |
+
"memory(GiB)": 45.5,
|
| 22 |
+
"nll_loss": 0.9818602204322815,
|
| 23 |
+
"rewards/accuracies": 0.0,
|
| 24 |
+
"rewards/chosen": 0.0,
|
| 25 |
+
"rewards/margins": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1,
|
| 28 |
+
"train_speed(iter/s)": 0.005543
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.00478433131494356,
|
| 32 |
+
"grad_norm": 6.833891868591309,
|
| 33 |
+
"learning_rate": 6.349206349206349e-06,
|
| 34 |
+
"logits/chosen": -0.6526631116867065,
|
| 35 |
+
"logits/rejected": -0.6589193344116211,
|
| 36 |
+
"logps/chosen": -13.911490440368652,
|
| 37 |
+
"logps/rejected": -26.267818450927734,
|
| 38 |
+
"loss": 1.5071839094161987,
|
| 39 |
+
"memory(GiB)": 45.7,
|
| 40 |
+
"nll_loss": 0.8140366077423096,
|
| 41 |
+
"rewards/accuracies": 0.0,
|
| 42 |
+
"rewards/chosen": 0.0,
|
| 43 |
+
"rewards/margins": 0.0,
|
| 44 |
+
"rewards/rejected": 0.0,
|
| 45 |
+
"step": 2,
|
| 46 |
+
"train_speed(iter/s)": 0.0056
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.00717649697241534,
|
| 50 |
+
"grad_norm": 4.8643999099731445,
|
| 51 |
+
"learning_rate": 9.523809523809523e-06,
|
| 52 |
+
"logits/chosen": -0.6420565247535706,
|
| 53 |
+
"logits/rejected": -0.6463347673416138,
|
| 54 |
+
"logps/chosen": -10.297296524047852,
|
| 55 |
+
"logps/rejected": -21.898191452026367,
|
| 56 |
+
"loss": 1.3684215545654297,
|
| 57 |
+
"memory(GiB)": 45.7,
|
| 58 |
+
"nll_loss": 0.6745598316192627,
|
| 59 |
+
"rewards/accuracies": 0.40625,
|
| 60 |
+
"rewards/chosen": 0.000917938828933984,
|
| 61 |
+
"rewards/margins": -0.0013222515117377043,
|
| 62 |
+
"rewards/rejected": 0.0022401907481253147,
|
| 63 |
+
"step": 3,
|
| 64 |
+
"train_speed(iter/s)": 0.005679
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.00956866262988712,
|
| 68 |
+
"grad_norm": 13.630061149597168,
|
| 69 |
+
"learning_rate": 1.2698412698412699e-05,
|
| 70 |
+
"logits/chosen": -0.6085749864578247,
|
| 71 |
+
"logits/rejected": -0.6145466566085815,
|
| 72 |
+
"logps/chosen": -9.835967063903809,
|
| 73 |
+
"logps/rejected": -20.03702163696289,
|
| 74 |
+
"loss": 1.6158230304718018,
|
| 75 |
+
"memory(GiB)": 45.7,
|
| 76 |
+
"nll_loss": 0.9227687120437622,
|
| 77 |
+
"rewards/accuracies": 0.5625,
|
| 78 |
+
"rewards/chosen": 0.002711558947339654,
|
| 79 |
+
"rewards/margins": 0.0003595724410843104,
|
| 80 |
+
"rewards/rejected": 0.0023519867099821568,
|
| 81 |
+
"step": 4,
|
| 82 |
+
"train_speed(iter/s)": 0.005709
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.0119608282873589,
|
| 86 |
+
"grad_norm": 9.080723762512207,
|
| 87 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 88 |
+
"logits/chosen": -0.6302319765090942,
|
| 89 |
+
"logits/rejected": -0.6327166557312012,
|
| 90 |
+
"logps/chosen": -12.458501815795898,
|
| 91 |
+
"logps/rejected": -20.92184066772461,
|
| 92 |
+
"loss": 1.7525994777679443,
|
| 93 |
+
"memory(GiB)": 45.7,
|
| 94 |
+
"nll_loss": 1.0644183158874512,
|
| 95 |
+
"rewards/accuracies": 0.71875,
|
| 96 |
+
"rewards/chosen": 0.015323596075177193,
|
| 97 |
+
"rewards/margins": 0.010122949257493019,
|
| 98 |
+
"rewards/rejected": 0.005200647283345461,
|
| 99 |
+
"step": 5,
|
| 100 |
+
"train_speed(iter/s)": 0.005683
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01435299394483068,
|
| 104 |
+
"grad_norm": 10.974760055541992,
|
| 105 |
+
"learning_rate": 1.9047619047619046e-05,
|
| 106 |
+
"logits/chosen": -0.6460072994232178,
|
| 107 |
+
"logits/rejected": -0.6503366231918335,
|
| 108 |
+
"logps/chosen": -8.966805458068848,
|
| 109 |
+
"logps/rejected": -29.11065101623535,
|
| 110 |
+
"loss": 1.2482141256332397,
|
| 111 |
+
"memory(GiB)": 45.7,
|
| 112 |
+
"nll_loss": 0.5670613050460815,
|
| 113 |
+
"rewards/accuracies": 0.59375,
|
| 114 |
+
"rewards/chosen": 0.03081551007926464,
|
| 115 |
+
"rewards/margins": 0.024802545085549355,
|
| 116 |
+
"rewards/rejected": 0.006012964062392712,
|
| 117 |
+
"step": 6,
|
| 118 |
+
"train_speed(iter/s)": 0.005657
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.01674515960230246,
|
| 122 |
+
"grad_norm": 9.0870943069458,
|
| 123 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 124 |
+
"logits/chosen": -0.5985114574432373,
|
| 125 |
+
"logits/rejected": -0.6029279232025146,
|
| 126 |
+
"logps/chosen": -12.831599235534668,
|
| 127 |
+
"logps/rejected": -22.661216735839844,
|
| 128 |
+
"loss": 1.4283000230789185,
|
| 129 |
+
"memory(GiB)": 45.7,
|
| 130 |
+
"nll_loss": 0.7556005120277405,
|
| 131 |
+
"rewards/accuracies": 0.625,
|
| 132 |
+
"rewards/chosen": 0.05840035900473595,
|
| 133 |
+
"rewards/margins": 0.044253990054130554,
|
| 134 |
+
"rewards/rejected": 0.01414636429399252,
|
| 135 |
+
"step": 7,
|
| 136 |
+
"train_speed(iter/s)": 0.005641
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.01913732525977424,
|
| 140 |
+
"grad_norm": 7.071105003356934,
|
| 141 |
+
"learning_rate": 2.5396825396825397e-05,
|
| 142 |
+
"logits/chosen": -0.6822581887245178,
|
| 143 |
+
"logits/rejected": -0.6836903095245361,
|
| 144 |
+
"logps/chosen": -14.09546184539795,
|
| 145 |
+
"logps/rejected": -17.29296112060547,
|
| 146 |
+
"loss": 1.3682955503463745,
|
| 147 |
+
"memory(GiB)": 45.7,
|
| 148 |
+
"nll_loss": 0.6906765699386597,
|
| 149 |
+
"rewards/accuracies": 0.625,
|
| 150 |
+
"rewards/chosen": 0.11469551920890808,
|
| 151 |
+
"rewards/margins": 0.042920827865600586,
|
| 152 |
+
"rewards/rejected": 0.07177469879388809,
|
| 153 |
+
"step": 8,
|
| 154 |
+
"train_speed(iter/s)": 0.00562
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.02152949091724602,
|
| 158 |
+
"grad_norm": 6.927362442016602,
|
| 159 |
+
"learning_rate": 2.857142857142857e-05,
|
| 160 |
+
"logits/chosen": -0.6293038725852966,
|
| 161 |
+
"logits/rejected": -0.632675051689148,
|
| 162 |
+
"logps/chosen": -13.061110496520996,
|
| 163 |
+
"logps/rejected": -22.598844528198242,
|
| 164 |
+
"loss": 1.382551670074463,
|
| 165 |
+
"memory(GiB)": 45.7,
|
| 166 |
+
"nll_loss": 0.7284406423568726,
|
| 167 |
+
"rewards/accuracies": 0.65625,
|
| 168 |
+
"rewards/chosen": 0.11742030084133148,
|
| 169 |
+
"rewards/margins": 0.10634221881628036,
|
| 170 |
+
"rewards/rejected": 0.011078080162405968,
|
| 171 |
+
"step": 9,
|
| 172 |
+
"train_speed(iter/s)": 0.005614
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.0239216565747178,
|
| 176 |
+
"grad_norm": 3.728199005126953,
|
| 177 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 178 |
+
"logits/chosen": -0.6618058681488037,
|
| 179 |
+
"logits/rejected": -0.6639875173568726,
|
| 180 |
+
"logps/chosen": -13.987442970275879,
|
| 181 |
+
"logps/rejected": -25.95743179321289,
|
| 182 |
+
"loss": 1.0269697904586792,
|
| 183 |
+
"memory(GiB)": 45.7,
|
| 184 |
+
"nll_loss": 0.4300592243671417,
|
| 185 |
+
"rewards/accuracies": 0.5625,
|
| 186 |
+
"rewards/chosen": 0.24149325489997864,
|
| 187 |
+
"rewards/margins": 0.27823469042778015,
|
| 188 |
+
"rewards/rejected": -0.03674142807722092,
|
| 189 |
+
"step": 10,
|
| 190 |
+
"train_speed(iter/s)": 0.005615
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.02631382223218958,
|
| 194 |
+
"grad_norm": 7.327150344848633,
|
| 195 |
+
"learning_rate": 3.492063492063492e-05,
|
| 196 |
+
"logits/chosen": -0.6170505285263062,
|
| 197 |
+
"logits/rejected": -0.6200059056282043,
|
| 198 |
+
"logps/chosen": -12.316822052001953,
|
| 199 |
+
"logps/rejected": -22.407135009765625,
|
| 200 |
+
"loss": 1.2084441184997559,
|
| 201 |
+
"memory(GiB)": 45.7,
|
| 202 |
+
"nll_loss": 0.525244951248169,
|
| 203 |
+
"rewards/accuracies": 0.59375,
|
| 204 |
+
"rewards/chosen": 0.011396417394280434,
|
| 205 |
+
"rewards/margins": 0.15045319497585297,
|
| 206 |
+
"rewards/rejected": -0.13905677199363708,
|
| 207 |
+
"step": 11,
|
| 208 |
+
"train_speed(iter/s)": 0.00562
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.02870598788966136,
|
| 212 |
+
"grad_norm": 5.15468168258667,
|
| 213 |
+
"learning_rate": 3.809523809523809e-05,
|
| 214 |
+
"logits/chosen": -0.6354009509086609,
|
| 215 |
+
"logits/rejected": -0.6367737054824829,
|
| 216 |
+
"logps/chosen": -16.074636459350586,
|
| 217 |
+
"logps/rejected": -23.24802017211914,
|
| 218 |
+
"loss": 1.352920413017273,
|
| 219 |
+
"memory(GiB)": 45.7,
|
| 220 |
+
"nll_loss": 0.6531318426132202,
|
| 221 |
+
"rewards/accuracies": 0.53125,
|
| 222 |
+
"rewards/chosen": -0.10709743946790695,
|
| 223 |
+
"rewards/margins": 0.1539226770401001,
|
| 224 |
+
"rewards/rejected": -0.26102012395858765,
|
| 225 |
+
"step": 12,
|
| 226 |
+
"train_speed(iter/s)": 0.005618
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.031098153547133137,
|
| 230 |
+
"grad_norm": 9.377359390258789,
|
| 231 |
+
"learning_rate": 4.126984126984127e-05,
|
| 232 |
+
"logits/chosen": -0.5982232093811035,
|
| 233 |
+
"logits/rejected": -0.5973206758499146,
|
| 234 |
+
"logps/chosen": -17.6663761138916,
|
| 235 |
+
"logps/rejected": -19.997753143310547,
|
| 236 |
+
"loss": 1.681210994720459,
|
| 237 |
+
"memory(GiB)": 45.7,
|
| 238 |
+
"nll_loss": 0.8636733889579773,
|
| 239 |
+
"rewards/accuracies": 0.40625,
|
| 240 |
+
"rewards/chosen": -0.24621205031871796,
|
| 241 |
+
"rewards/margins": -0.07156651467084885,
|
| 242 |
+
"rewards/rejected": -0.17464551329612732,
|
| 243 |
+
"step": 13,
|
| 244 |
+
"train_speed(iter/s)": 0.005624
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.03349031920460492,
|
| 248 |
+
"grad_norm": 3.9681458473205566,
|
| 249 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 250 |
+
"logits/chosen": -0.6217225193977356,
|
| 251 |
+
"logits/rejected": -0.6225230097770691,
|
| 252 |
+
"logps/chosen": -12.139883995056152,
|
| 253 |
+
"logps/rejected": -25.268104553222656,
|
| 254 |
+
"loss": 1.2384405136108398,
|
| 255 |
+
"memory(GiB)": 46.29,
|
| 256 |
+
"nll_loss": 0.5618223547935486,
|
| 257 |
+
"rewards/accuracies": 0.625,
|
| 258 |
+
"rewards/chosen": -0.0858367532491684,
|
| 259 |
+
"rewards/margins": 0.20134860277175903,
|
| 260 |
+
"rewards/rejected": -0.28718534111976624,
|
| 261 |
+
"step": 14,
|
| 262 |
+
"train_speed(iter/s)": 0.005621
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0358824848620767,
|
| 266 |
+
"grad_norm": 3.7261455059051514,
|
| 267 |
+
"learning_rate": 4.761904761904762e-05,
|
| 268 |
+
"logits/chosen": -0.5990071892738342,
|
| 269 |
+
"logits/rejected": -0.603517472743988,
|
| 270 |
+
"logps/chosen": -9.707769393920898,
|
| 271 |
+
"logps/rejected": -30.03302574157715,
|
| 272 |
+
"loss": 1.114662766456604,
|
| 273 |
+
"memory(GiB)": 46.29,
|
| 274 |
+
"nll_loss": 0.5713129639625549,
|
| 275 |
+
"rewards/accuracies": 0.75,
|
| 276 |
+
"rewards/chosen": 0.08963251113891602,
|
| 277 |
+
"rewards/margins": 0.5134348273277283,
|
| 278 |
+
"rewards/rejected": -0.42380231618881226,
|
| 279 |
+
"step": 15,
|
| 280 |
+
"train_speed(iter/s)": 0.005614
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.03827465051954848,
|
| 284 |
+
"grad_norm": 4.653024673461914,
|
| 285 |
+
"learning_rate": 5.0793650793650794e-05,
|
| 286 |
+
"logits/chosen": -0.6132736206054688,
|
| 287 |
+
"logits/rejected": -0.6161543726921082,
|
| 288 |
+
"logps/chosen": -15.514596939086914,
|
| 289 |
+
"logps/rejected": -26.068016052246094,
|
| 290 |
+
"loss": 1.281750202178955,
|
| 291 |
+
"memory(GiB)": 46.29,
|
| 292 |
+
"nll_loss": 0.5498376488685608,
|
| 293 |
+
"rewards/accuracies": 0.4375,
|
| 294 |
+
"rewards/chosen": -0.09057736396789551,
|
| 295 |
+
"rewards/margins": 0.08230985701084137,
|
| 296 |
+
"rewards/rejected": -0.17288722097873688,
|
| 297 |
+
"step": 16,
|
| 298 |
+
"train_speed(iter/s)": 0.005619
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.04066681617702026,
|
| 302 |
+
"grad_norm": 3.9547042846679688,
|
| 303 |
+
"learning_rate": 5.396825396825397e-05,
|
| 304 |
+
"logits/chosen": -0.5502676367759705,
|
| 305 |
+
"logits/rejected": -0.5536152124404907,
|
| 306 |
+
"logps/chosen": -13.769225120544434,
|
| 307 |
+
"logps/rejected": -27.007709503173828,
|
| 308 |
+
"loss": 1.413275122642517,
|
| 309 |
+
"memory(GiB)": 46.29,
|
| 310 |
+
"nll_loss": 0.7366797924041748,
|
| 311 |
+
"rewards/accuracies": 0.59375,
|
| 312 |
+
"rewards/chosen": -0.05718833580613136,
|
| 313 |
+
"rewards/margins": 0.1863359659910202,
|
| 314 |
+
"rewards/rejected": -0.24352429807186127,
|
| 315 |
+
"step": 17,
|
| 316 |
+
"train_speed(iter/s)": 0.005631
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.04305898183449204,
|
| 320 |
+
"grad_norm": 3.8449220657348633,
|
| 321 |
+
"learning_rate": 5.714285714285714e-05,
|
| 322 |
+
"logits/chosen": -0.5983889102935791,
|
| 323 |
+
"logits/rejected": -0.601371705532074,
|
| 324 |
+
"logps/chosen": -11.734331130981445,
|
| 325 |
+
"logps/rejected": -22.929622650146484,
|
| 326 |
+
"loss": 1.3582308292388916,
|
| 327 |
+
"memory(GiB)": 46.29,
|
| 328 |
+
"nll_loss": 0.639457106590271,
|
| 329 |
+
"rewards/accuracies": 0.4375,
|
| 330 |
+
"rewards/chosen": 0.03578497841954231,
|
| 331 |
+
"rewards/margins": 0.019625335931777954,
|
| 332 |
+
"rewards/rejected": 0.016159646213054657,
|
| 333 |
+
"step": 18,
|
| 334 |
+
"train_speed(iter/s)": 0.005646
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.04545114749196382,
|
| 338 |
+
"grad_norm": 3.037893295288086,
|
| 339 |
+
"learning_rate": 6.0317460317460316e-05,
|
| 340 |
+
"logits/chosen": -0.6029090881347656,
|
| 341 |
+
"logits/rejected": -0.6060330867767334,
|
| 342 |
+
"logps/chosen": -13.030549049377441,
|
| 343 |
+
"logps/rejected": -24.52211570739746,
|
| 344 |
+
"loss": 1.2112592458724976,
|
| 345 |
+
"memory(GiB)": 46.29,
|
| 346 |
+
"nll_loss": 0.526079535484314,
|
| 347 |
+
"rewards/accuracies": 0.5625,
|
| 348 |
+
"rewards/chosen": 0.08603789657354355,
|
| 349 |
+
"rewards/margins": 0.11102047562599182,
|
| 350 |
+
"rewards/rejected": -0.024982573464512825,
|
| 351 |
+
"step": 19,
|
| 352 |
+
"train_speed(iter/s)": 0.005649
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.0478433131494356,
|
| 356 |
+
"grad_norm": 3.4577159881591797,
|
| 357 |
+
"learning_rate": 6.349206349206349e-05,
|
| 358 |
+
"logits/chosen": -0.6348394155502319,
|
| 359 |
+
"logits/rejected": -0.6350866556167603,
|
| 360 |
+
"logps/chosen": -10.457337379455566,
|
| 361 |
+
"logps/rejected": -21.804182052612305,
|
| 362 |
+
"loss": 1.271092414855957,
|
| 363 |
+
"memory(GiB)": 46.29,
|
| 364 |
+
"nll_loss": 0.5564278364181519,
|
| 365 |
+
"rewards/accuracies": 0.5,
|
| 366 |
+
"rewards/chosen": 0.13572227954864502,
|
| 367 |
+
"rewards/margins": 0.03713224455714226,
|
| 368 |
+
"rewards/rejected": 0.09859003126621246,
|
| 369 |
+
"step": 20,
|
| 370 |
+
"train_speed(iter/s)": 0.00565
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.05023547880690738,
|
| 374 |
+
"grad_norm": 2.758153200149536,
|
| 375 |
+
"learning_rate": 6.666666666666667e-05,
|
| 376 |
+
"logits/chosen": -0.5788831114768982,
|
| 377 |
+
"logits/rejected": -0.5802679061889648,
|
| 378 |
+
"logps/chosen": -12.416455268859863,
|
| 379 |
+
"logps/rejected": -15.200132369995117,
|
| 380 |
+
"loss": 1.3435254096984863,
|
| 381 |
+
"memory(GiB)": 46.29,
|
| 382 |
+
"nll_loss": 0.5808812379837036,
|
| 383 |
+
"rewards/accuracies": 0.3125,
|
| 384 |
+
"rewards/chosen": 0.11838358640670776,
|
| 385 |
+
"rewards/margins": -0.03334012255072594,
|
| 386 |
+
"rewards/rejected": 0.1517237275838852,
|
| 387 |
+
"step": 21,
|
| 388 |
+
"train_speed(iter/s)": 0.005654
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.05262764446437916,
|
| 392 |
+
"grad_norm": 3.07053542137146,
|
| 393 |
+
"learning_rate": 6.984126984126984e-05,
|
| 394 |
+
"logits/chosen": -0.5787701606750488,
|
| 395 |
+
"logits/rejected": -0.582938551902771,
|
| 396 |
+
"logps/chosen": -6.899454116821289,
|
| 397 |
+
"logps/rejected": -21.719806671142578,
|
| 398 |
+
"loss": 1.0226894617080688,
|
| 399 |
+
"memory(GiB)": 46.29,
|
| 400 |
+
"nll_loss": 0.45206937193870544,
|
| 401 |
+
"rewards/accuracies": 0.78125,
|
| 402 |
+
"rewards/chosen": 0.30141958594322205,
|
| 403 |
+
"rewards/margins": 0.32421889901161194,
|
| 404 |
+
"rewards/rejected": -0.022799327969551086,
|
| 405 |
+
"step": 22,
|
| 406 |
+
"train_speed(iter/s)": 0.005649
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.05501981012185094,
|
| 410 |
+
"grad_norm": 3.616273880004883,
|
| 411 |
+
"learning_rate": 7.301587301587302e-05,
|
| 412 |
+
"logits/chosen": -0.5814599990844727,
|
| 413 |
+
"logits/rejected": -0.5829823017120361,
|
| 414 |
+
"logps/chosen": -10.043213844299316,
|
| 415 |
+
"logps/rejected": -16.569318771362305,
|
| 416 |
+
"loss": 1.5021785497665405,
|
| 417 |
+
"memory(GiB)": 46.29,
|
| 418 |
+
"nll_loss": 0.8470967411994934,
|
| 419 |
+
"rewards/accuracies": 0.65625,
|
| 420 |
+
"rewards/chosen": 0.15741629898548126,
|
| 421 |
+
"rewards/margins": 0.13470463454723358,
|
| 422 |
+
"rewards/rejected": 0.022711653262376785,
|
| 423 |
+
"step": 23,
|
| 424 |
+
"train_speed(iter/s)": 0.005652
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.05741197577932272,
|
| 428 |
+
"grad_norm": 2.9444239139556885,
|
| 429 |
+
"learning_rate": 7.619047619047618e-05,
|
| 430 |
+
"logits/chosen": -0.5737470388412476,
|
| 431 |
+
"logits/rejected": -0.5753998756408691,
|
| 432 |
+
"logps/chosen": -9.754992485046387,
|
| 433 |
+
"logps/rejected": -23.393245697021484,
|
| 434 |
+
"loss": 1.1848090887069702,
|
| 435 |
+
"memory(GiB)": 46.29,
|
| 436 |
+
"nll_loss": 0.5859429240226746,
|
| 437 |
+
"rewards/accuracies": 0.71875,
|
| 438 |
+
"rewards/chosen": 0.26539331674575806,
|
| 439 |
+
"rewards/margins": 0.25131887197494507,
|
| 440 |
+
"rewards/rejected": 0.01407446339726448,
|
| 441 |
+
"step": 24,
|
| 442 |
+
"train_speed(iter/s)": 0.005651
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.059804141436794495,
|
| 446 |
+
"grad_norm": 2.368903875350952,
|
| 447 |
+
"learning_rate": 7.936507936507937e-05,
|
| 448 |
+
"logits/chosen": -0.5797973871231079,
|
| 449 |
+
"logits/rejected": -0.5811963081359863,
|
| 450 |
+
"logps/chosen": -7.4247236251831055,
|
| 451 |
+
"logps/rejected": -17.2625789642334,
|
| 452 |
+
"loss": 1.1369986534118652,
|
| 453 |
+
"memory(GiB)": 46.29,
|
| 454 |
+
"nll_loss": 0.5304021835327148,
|
| 455 |
+
"rewards/accuracies": 0.6875,
|
| 456 |
+
"rewards/chosen": 0.18115447461605072,
|
| 457 |
+
"rewards/margins": 0.21577578783035278,
|
| 458 |
+
"rewards/rejected": -0.034621305763721466,
|
| 459 |
+
"step": 25,
|
| 460 |
+
"train_speed(iter/s)": 0.005651
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.062196307094266275,
|
| 464 |
+
"grad_norm": 2.0124542713165283,
|
| 465 |
+
"learning_rate": 8.253968253968255e-05,
|
| 466 |
+
"logits/chosen": -0.5834470987319946,
|
| 467 |
+
"logits/rejected": -0.5854894518852234,
|
| 468 |
+
"logps/chosen": -10.883503913879395,
|
| 469 |
+
"logps/rejected": -17.59324836730957,
|
| 470 |
+
"loss": 1.050497055053711,
|
| 471 |
+
"memory(GiB)": 46.29,
|
| 472 |
+
"nll_loss": 0.4383053183555603,
|
| 473 |
+
"rewards/accuracies": 0.59375,
|
| 474 |
+
"rewards/chosen": 0.22693531215190887,
|
| 475 |
+
"rewards/margins": 0.23587419092655182,
|
| 476 |
+
"rewards/rejected": -0.008938884362578392,
|
| 477 |
+
"step": 26,
|
| 478 |
+
"train_speed(iter/s)": 0.005654
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.06458847275173805,
|
| 482 |
+
"grad_norm": 2.3302695751190186,
|
| 483 |
+
"learning_rate": 8.571428571428571e-05,
|
| 484 |
+
"logits/chosen": -0.5692424774169922,
|
| 485 |
+
"logits/rejected": -0.5711921453475952,
|
| 486 |
+
"logps/chosen": -12.02816104888916,
|
| 487 |
+
"logps/rejected": -23.778608322143555,
|
| 488 |
+
"loss": 1.1506422758102417,
|
| 489 |
+
"memory(GiB)": 46.29,
|
| 490 |
+
"nll_loss": 0.5187042951583862,
|
| 491 |
+
"rewards/accuracies": 0.5625,
|
| 492 |
+
"rewards/chosen": 0.11056558787822723,
|
| 493 |
+
"rewards/margins": 0.18688789010047913,
|
| 494 |
+
"rewards/rejected": -0.0763222873210907,
|
| 495 |
+
"step": 27,
|
| 496 |
+
"train_speed(iter/s)": 0.005653
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.06698063840920984,
|
| 500 |
+
"grad_norm": 2.4422554969787598,
|
| 501 |
+
"learning_rate": 8.888888888888889e-05,
|
| 502 |
+
"logits/chosen": -0.5987146496772766,
|
| 503 |
+
"logits/rejected": -0.5993437767028809,
|
| 504 |
+
"logps/chosen": -9.00568675994873,
|
| 505 |
+
"logps/rejected": -17.24131965637207,
|
| 506 |
+
"loss": 1.0437039136886597,
|
| 507 |
+
"memory(GiB)": 46.29,
|
| 508 |
+
"nll_loss": 0.4295322299003601,
|
| 509 |
+
"rewards/accuracies": 0.59375,
|
| 510 |
+
"rewards/chosen": 0.19704687595367432,
|
| 511 |
+
"rewards/margins": 0.21514004468917847,
|
| 512 |
+
"rewards/rejected": -0.018093157559633255,
|
| 513 |
+
"step": 28,
|
| 514 |
+
"train_speed(iter/s)": 0.00565
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.06937280406668161,
|
| 518 |
+
"grad_norm": 3.3448095321655273,
|
| 519 |
+
"learning_rate": 9.206349206349206e-05,
|
| 520 |
+
"logits/chosen": -0.5584565997123718,
|
| 521 |
+
"logits/rejected": -0.5605788826942444,
|
| 522 |
+
"logps/chosen": -7.835123538970947,
|
| 523 |
+
"logps/rejected": -19.747636795043945,
|
| 524 |
+
"loss": 1.172892451286316,
|
| 525 |
+
"memory(GiB)": 46.29,
|
| 526 |
+
"nll_loss": 0.55903559923172,
|
| 527 |
+
"rewards/accuracies": 0.71875,
|
| 528 |
+
"rewards/chosen": 0.15631772577762604,
|
| 529 |
+
"rewards/margins": 0.2727695107460022,
|
| 530 |
+
"rewards/rejected": -0.11645177006721497,
|
| 531 |
+
"step": 29,
|
| 532 |
+
"train_speed(iter/s)": 0.005649
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.0717649697241534,
|
| 536 |
+
"grad_norm": 2.956876516342163,
|
| 537 |
+
"learning_rate": 9.523809523809524e-05,
|
| 538 |
+
"logits/chosen": -0.5594294667243958,
|
| 539 |
+
"logits/rejected": -0.5630677342414856,
|
| 540 |
+
"logps/chosen": -8.298831939697266,
|
| 541 |
+
"logps/rejected": -26.14436149597168,
|
| 542 |
+
"loss": 1.0785880088806152,
|
| 543 |
+
"memory(GiB)": 46.78,
|
| 544 |
+
"nll_loss": 0.5043998956680298,
|
| 545 |
+
"rewards/accuracies": 0.75,
|
| 546 |
+
"rewards/chosen": 0.15360432863235474,
|
| 547 |
+
"rewards/margins": 0.37323492765426636,
|
| 548 |
+
"rewards/rejected": -0.21963059902191162,
|
| 549 |
+
"step": 30,
|
| 550 |
+
"train_speed(iter/s)": 0.005646
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.07415713538162517,
|
| 554 |
+
"grad_norm": 2.969925880432129,
|
| 555 |
+
"learning_rate": 9.841269841269841e-05,
|
| 556 |
+
"logits/chosen": -0.563766360282898,
|
| 557 |
+
"logits/rejected": -0.564283549785614,
|
| 558 |
+
"logps/chosen": -9.237369537353516,
|
| 559 |
+
"logps/rejected": -23.180709838867188,
|
| 560 |
+
"loss": 1.0209345817565918,
|
| 561 |
+
"memory(GiB)": 46.78,
|
| 562 |
+
"nll_loss": 0.42502841353416443,
|
| 563 |
+
"rewards/accuracies": 0.84375,
|
| 564 |
+
"rewards/chosen": 0.18122270703315735,
|
| 565 |
+
"rewards/margins": 0.2958879768848419,
|
| 566 |
+
"rewards/rejected": -0.11466525495052338,
|
| 567 |
+
"step": 31,
|
| 568 |
+
"train_speed(iter/s)": 0.005646
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.07654930103909696,
|
| 572 |
+
"grad_norm": 3.5162458419799805,
|
| 573 |
+
"learning_rate": 0.00010158730158730159,
|
| 574 |
+
"logits/chosen": -0.5594556927680969,
|
| 575 |
+
"logits/rejected": -0.5654793977737427,
|
| 576 |
+
"logps/chosen": -5.732161045074463,
|
| 577 |
+
"logps/rejected": -32.01944351196289,
|
| 578 |
+
"loss": 0.9676632881164551,
|
| 579 |
+
"memory(GiB)": 46.78,
|
| 580 |
+
"nll_loss": 0.4746396541595459,
|
| 581 |
+
"rewards/accuracies": 0.84375,
|
| 582 |
+
"rewards/chosen": 0.2609599530696869,
|
| 583 |
+
"rewards/margins": 0.6239939332008362,
|
| 584 |
+
"rewards/rejected": -0.3630340099334717,
|
| 585 |
+
"step": 32,
|
| 586 |
+
"train_speed(iter/s)": 0.005645
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.07894146669656874,
|
| 590 |
+
"grad_norm": 3.3218283653259277,
|
| 591 |
+
"learning_rate": 0.00010476190476190477,
|
| 592 |
+
"logits/chosen": -0.5853260159492493,
|
| 593 |
+
"logits/rejected": -0.5895906686782837,
|
| 594 |
+
"logps/chosen": -10.368348121643066,
|
| 595 |
+
"logps/rejected": -20.515506744384766,
|
| 596 |
+
"loss": 1.1254382133483887,
|
| 597 |
+
"memory(GiB)": 46.78,
|
| 598 |
+
"nll_loss": 0.5297057032585144,
|
| 599 |
+
"rewards/accuracies": 0.8125,
|
| 600 |
+
"rewards/chosen": 0.15998664498329163,
|
| 601 |
+
"rewards/margins": 0.346771240234375,
|
| 602 |
+
"rewards/rejected": -0.18678458034992218,
|
| 603 |
+
"step": 33,
|
| 604 |
+
"train_speed(iter/s)": 0.005644
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.08133363235404052,
|
| 608 |
+
"grad_norm": 4.339278697967529,
|
| 609 |
+
"learning_rate": 0.00010793650793650794,
|
| 610 |
+
"logits/chosen": -0.6093581318855286,
|
| 611 |
+
"logits/rejected": -0.6088329553604126,
|
| 612 |
+
"logps/chosen": -13.336594581604004,
|
| 613 |
+
"logps/rejected": -24.60618782043457,
|
| 614 |
+
"loss": 1.1337980031967163,
|
| 615 |
+
"memory(GiB)": 46.78,
|
| 616 |
+
"nll_loss": 0.5539220571517944,
|
| 617 |
+
"rewards/accuracies": 0.71875,
|
| 618 |
+
"rewards/chosen": 0.1875077337026596,
|
| 619 |
+
"rewards/margins": 0.4340534806251526,
|
| 620 |
+
"rewards/rejected": -0.24654576182365417,
|
| 621 |
+
"step": 34,
|
| 622 |
+
"train_speed(iter/s)": 0.005642
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.0837257980115123,
|
| 626 |
+
"grad_norm": 3.8656299114227295,
|
| 627 |
+
"learning_rate": 0.00011111111111111112,
|
| 628 |
+
"logits/chosen": -0.6122190952301025,
|
| 629 |
+
"logits/rejected": -0.616493284702301,
|
| 630 |
+
"logps/chosen": -13.914678573608398,
|
| 631 |
+
"logps/rejected": -32.916744232177734,
|
| 632 |
+
"loss": 1.1174585819244385,
|
| 633 |
+
"memory(GiB)": 46.78,
|
| 634 |
+
"nll_loss": 0.5647247433662415,
|
| 635 |
+
"rewards/accuracies": 0.71875,
|
| 636 |
+
"rewards/chosen": 0.015515260398387909,
|
| 637 |
+
"rewards/margins": 0.6349303722381592,
|
| 638 |
+
"rewards/rejected": -0.6194151639938354,
|
| 639 |
+
"step": 35,
|
| 640 |
+
"train_speed(iter/s)": 0.005638
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.08611796366898408,
|
| 644 |
+
"grad_norm": 3.665461540222168,
|
| 645 |
+
"learning_rate": 0.00011428571428571428,
|
| 646 |
+
"logits/chosen": -0.6845042705535889,
|
| 647 |
+
"logits/rejected": -0.6848981380462646,
|
| 648 |
+
"logps/chosen": -12.846878051757812,
|
| 649 |
+
"logps/rejected": -25.240447998046875,
|
| 650 |
+
"loss": 1.1161198616027832,
|
| 651 |
+
"memory(GiB)": 46.78,
|
| 652 |
+
"nll_loss": 0.5125420689582825,
|
| 653 |
+
"rewards/accuracies": 0.65625,
|
| 654 |
+
"rewards/chosen": 0.049385204911231995,
|
| 655 |
+
"rewards/margins": 0.33047205209732056,
|
| 656 |
+
"rewards/rejected": -0.28108683228492737,
|
| 657 |
+
"step": 36,
|
| 658 |
+
"train_speed(iter/s)": 0.005643
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.08851012932645586,
|
| 662 |
+
"grad_norm": 3.6945273876190186,
|
| 663 |
+
"learning_rate": 0.00011746031746031746,
|
| 664 |
+
"logits/chosen": -0.6884023547172546,
|
| 665 |
+
"logits/rejected": -0.69326251745224,
|
| 666 |
+
"logps/chosen": -7.306849956512451,
|
| 667 |
+
"logps/rejected": -24.99851417541504,
|
| 668 |
+
"loss": 0.9112467765808105,
|
| 669 |
+
"memory(GiB)": 46.78,
|
| 670 |
+
"nll_loss": 0.4103700518608093,
|
| 671 |
+
"rewards/accuracies": 0.84375,
|
| 672 |
+
"rewards/chosen": 0.24825294315814972,
|
| 673 |
+
"rewards/margins": 0.5438191294670105,
|
| 674 |
+
"rewards/rejected": -0.295566201210022,
|
| 675 |
+
"step": 37,
|
| 676 |
+
"train_speed(iter/s)": 0.005646
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.09090229498392764,
|
| 680 |
+
"grad_norm": 2.759567975997925,
|
| 681 |
+
"learning_rate": 0.00012063492063492063,
|
| 682 |
+
"logits/chosen": -0.6944407224655151,
|
| 683 |
+
"logits/rejected": -0.6986436247825623,
|
| 684 |
+
"logps/chosen": -8.414443969726562,
|
| 685 |
+
"logps/rejected": -29.627033233642578,
|
| 686 |
+
"loss": 0.9807862043380737,
|
| 687 |
+
"memory(GiB)": 46.78,
|
| 688 |
+
"nll_loss": 0.464620977640152,
|
| 689 |
+
"rewards/accuracies": 0.78125,
|
| 690 |
+
"rewards/chosen": 0.205428808927536,
|
| 691 |
+
"rewards/margins": 0.6206997036933899,
|
| 692 |
+
"rewards/rejected": -0.41527092456817627,
|
| 693 |
+
"step": 38,
|
| 694 |
+
"train_speed(iter/s)": 0.005647
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.09329446064139942,
|
| 698 |
+
"grad_norm": 3.707176685333252,
|
| 699 |
+
"learning_rate": 0.0001238095238095238,
|
| 700 |
+
"logits/chosen": -0.6549853086471558,
|
| 701 |
+
"logits/rejected": -0.6607218384742737,
|
| 702 |
+
"logps/chosen": -13.402036666870117,
|
| 703 |
+
"logps/rejected": -29.2238826751709,
|
| 704 |
+
"loss": 1.1863672733306885,
|
| 705 |
+
"memory(GiB)": 46.78,
|
| 706 |
+
"nll_loss": 0.5955373048782349,
|
| 707 |
+
"rewards/accuracies": 0.75,
|
| 708 |
+
"rewards/chosen": 0.0599823072552681,
|
| 709 |
+
"rewards/margins": 0.4598371684551239,
|
| 710 |
+
"rewards/rejected": -0.3998548984527588,
|
| 711 |
+
"step": 39,
|
| 712 |
+
"train_speed(iter/s)": 0.005645
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.0956866262988712,
|
| 716 |
+
"grad_norm": 5.015953540802002,
|
| 717 |
+
"learning_rate": 0.00012698412698412698,
|
| 718 |
+
"logits/chosen": -0.6966100931167603,
|
| 719 |
+
"logits/rejected": -0.7002891302108765,
|
| 720 |
+
"logps/chosen": -7.736152172088623,
|
| 721 |
+
"logps/rejected": -27.045543670654297,
|
| 722 |
+
"loss": 0.977799654006958,
|
| 723 |
+
"memory(GiB)": 46.78,
|
| 724 |
+
"nll_loss": 0.5642723441123962,
|
| 725 |
+
"rewards/accuracies": 0.84375,
|
| 726 |
+
"rewards/chosen": 0.3473990261554718,
|
| 727 |
+
"rewards/margins": 1.0138760805130005,
|
| 728 |
+
"rewards/rejected": -0.6664770245552063,
|
| 729 |
+
"step": 40,
|
| 730 |
+
"train_speed(iter/s)": 0.005645
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.09807879195634298,
|
| 734 |
+
"grad_norm": 4.407506465911865,
|
| 735 |
+
"learning_rate": 0.00013015873015873017,
|
| 736 |
+
"logits/chosen": -0.692713737487793,
|
| 737 |
+
"logits/rejected": -0.6956310272216797,
|
| 738 |
+
"logps/chosen": -9.848102569580078,
|
| 739 |
+
"logps/rejected": -24.158405303955078,
|
| 740 |
+
"loss": 1.0173840522766113,
|
| 741 |
+
"memory(GiB)": 46.78,
|
| 742 |
+
"nll_loss": 0.5190764665603638,
|
| 743 |
+
"rewards/accuracies": 0.84375,
|
| 744 |
+
"rewards/chosen": 0.1344590187072754,
|
| 745 |
+
"rewards/margins": 0.6329586505889893,
|
| 746 |
+
"rewards/rejected": -0.49849969148635864,
|
| 747 |
+
"step": 41,
|
| 748 |
+
"train_speed(iter/s)": 0.005645
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.10047095761381476,
|
| 752 |
+
"grad_norm": 3.482964515686035,
|
| 753 |
+
"learning_rate": 0.00013333333333333334,
|
| 754 |
+
"logits/chosen": -0.6898783445358276,
|
| 755 |
+
"logits/rejected": -0.694994330406189,
|
| 756 |
+
"logps/chosen": -10.68550968170166,
|
| 757 |
+
"logps/rejected": -28.72760772705078,
|
| 758 |
+
"loss": 1.0436055660247803,
|
| 759 |
+
"memory(GiB)": 46.78,
|
| 760 |
+
"nll_loss": 0.5537880659103394,
|
| 761 |
+
"rewards/accuracies": 0.71875,
|
| 762 |
+
"rewards/chosen": 0.1501934826374054,
|
| 763 |
+
"rewards/margins": 0.8055344820022583,
|
| 764 |
+
"rewards/rejected": -0.6553410291671753,
|
| 765 |
+
"step": 42,
|
| 766 |
+
"train_speed(iter/s)": 0.005645
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.10286312327128654,
|
| 770 |
+
"grad_norm": 3.4329941272735596,
|
| 771 |
+
"learning_rate": 0.0001365079365079365,
|
| 772 |
+
"logits/chosen": -0.7011124491691589,
|
| 773 |
+
"logits/rejected": -0.7064176797866821,
|
| 774 |
+
"logps/chosen": -15.171253204345703,
|
| 775 |
+
"logps/rejected": -29.2159481048584,
|
| 776 |
+
"loss": 1.0602749586105347,
|
| 777 |
+
"memory(GiB)": 46.78,
|
| 778 |
+
"nll_loss": 0.5264595746994019,
|
| 779 |
+
"rewards/accuracies": 0.6875,
|
| 780 |
+
"rewards/chosen": 0.06841089576482773,
|
| 781 |
+
"rewards/margins": 0.6964786052703857,
|
| 782 |
+
"rewards/rejected": -0.6280677318572998,
|
| 783 |
+
"step": 43,
|
| 784 |
+
"train_speed(iter/s)": 0.005644
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.10525528892875832,
|
| 788 |
+
"grad_norm": 2.459496259689331,
|
| 789 |
+
"learning_rate": 0.00013968253968253967,
|
| 790 |
+
"logits/chosen": -0.7666099667549133,
|
| 791 |
+
"logits/rejected": -0.771510899066925,
|
| 792 |
+
"logps/chosen": -5.71858549118042,
|
| 793 |
+
"logps/rejected": -27.757709503173828,
|
| 794 |
+
"loss": 0.8418532609939575,
|
| 795 |
+
"memory(GiB)": 46.78,
|
| 796 |
+
"nll_loss": 0.43147915601730347,
|
| 797 |
+
"rewards/accuracies": 0.84375,
|
| 798 |
+
"rewards/chosen": 0.2912772297859192,
|
| 799 |
+
"rewards/margins": 0.8343029022216797,
|
| 800 |
+
"rewards/rejected": -0.5430256724357605,
|
| 801 |
+
"step": 44,
|
| 802 |
+
"train_speed(iter/s)": 0.005645
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.1076474545862301,
|
| 806 |
+
"grad_norm": 2.439218759536743,
|
| 807 |
+
"learning_rate": 0.00014285714285714287,
|
| 808 |
+
"logits/chosen": -0.7568098902702332,
|
| 809 |
+
"logits/rejected": -0.7595718502998352,
|
| 810 |
+
"logps/chosen": -10.82071304321289,
|
| 811 |
+
"logps/rejected": -26.015369415283203,
|
| 812 |
+
"loss": 0.9067925214767456,
|
| 813 |
+
"memory(GiB)": 46.78,
|
| 814 |
+
"nll_loss": 0.4222971796989441,
|
| 815 |
+
"rewards/accuracies": 0.71875,
|
| 816 |
+
"rewards/chosen": 0.23031610250473022,
|
| 817 |
+
"rewards/margins": 0.6913963556289673,
|
| 818 |
+
"rewards/rejected": -0.46108025312423706,
|
| 819 |
+
"step": 45,
|
| 820 |
+
"train_speed(iter/s)": 0.005643
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.11003962024370188,
|
| 824 |
+
"grad_norm": 2.0914173126220703,
|
| 825 |
+
"learning_rate": 0.00014603174603174603,
|
| 826 |
+
"logits/chosen": -0.8171934485435486,
|
| 827 |
+
"logits/rejected": -0.8208147287368774,
|
| 828 |
+
"logps/chosen": -7.6959333419799805,
|
| 829 |
+
"logps/rejected": -26.958995819091797,
|
| 830 |
+
"loss": 0.8884089589118958,
|
| 831 |
+
"memory(GiB)": 46.78,
|
| 832 |
+
"nll_loss": 0.4511643052101135,
|
| 833 |
+
"rewards/accuracies": 0.78125,
|
| 834 |
+
"rewards/chosen": 0.29817235469818115,
|
| 835 |
+
"rewards/margins": 0.9990655183792114,
|
| 836 |
+
"rewards/rejected": -0.700893223285675,
|
| 837 |
+
"step": 46,
|
| 838 |
+
"train_speed(iter/s)": 0.005641
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.11243178590117366,
|
| 842 |
+
"grad_norm": 2.7921295166015625,
|
| 843 |
+
"learning_rate": 0.00014920634920634923,
|
| 844 |
+
"logits/chosen": -0.8353937864303589,
|
| 845 |
+
"logits/rejected": -0.8423863649368286,
|
| 846 |
+
"logps/chosen": -11.051738739013672,
|
| 847 |
+
"logps/rejected": -29.439268112182617,
|
| 848 |
+
"loss": 0.9526035189628601,
|
| 849 |
+
"memory(GiB)": 46.78,
|
| 850 |
+
"nll_loss": 0.5338684916496277,
|
| 851 |
+
"rewards/accuracies": 0.84375,
|
| 852 |
+
"rewards/chosen": 0.15457165241241455,
|
| 853 |
+
"rewards/margins": 0.8811558485031128,
|
| 854 |
+
"rewards/rejected": -0.7265841960906982,
|
| 855 |
+
"step": 47,
|
| 856 |
+
"train_speed(iter/s)": 0.00564
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.11482395155864544,
|
| 860 |
+
"grad_norm": 3.267803907394409,
|
| 861 |
+
"learning_rate": 0.00015238095238095237,
|
| 862 |
+
"logits/chosen": -0.8806051015853882,
|
| 863 |
+
"logits/rejected": -0.8819608688354492,
|
| 864 |
+
"logps/chosen": -15.4578275680542,
|
| 865 |
+
"logps/rejected": -28.783483505249023,
|
| 866 |
+
"loss": 1.037742018699646,
|
| 867 |
+
"memory(GiB)": 46.78,
|
| 868 |
+
"nll_loss": 0.5505475997924805,
|
| 869 |
+
"rewards/accuracies": 0.78125,
|
| 870 |
+
"rewards/chosen": 0.205413818359375,
|
| 871 |
+
"rewards/margins": 0.9907757043838501,
|
| 872 |
+
"rewards/rejected": -0.7853618264198303,
|
| 873 |
+
"step": 48,
|
| 874 |
+
"train_speed(iter/s)": 0.005641
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.11721611721611722,
|
| 878 |
+
"grad_norm": 5.174523830413818,
|
| 879 |
+
"learning_rate": 0.00015555555555555556,
|
| 880 |
+
"logits/chosen": -0.9132509827613831,
|
| 881 |
+
"logits/rejected": -0.9135016202926636,
|
| 882 |
+
"logps/chosen": -16.307157516479492,
|
| 883 |
+
"logps/rejected": -32.103309631347656,
|
| 884 |
+
"loss": 1.1124699115753174,
|
| 885 |
+
"memory(GiB)": 46.78,
|
| 886 |
+
"nll_loss": 0.6212166547775269,
|
| 887 |
+
"rewards/accuracies": 0.78125,
|
| 888 |
+
"rewards/chosen": 0.16737183928489685,
|
| 889 |
+
"rewards/margins": 0.9279904365539551,
|
| 890 |
+
"rewards/rejected": -0.7606185078620911,
|
| 891 |
+
"step": 49,
|
| 892 |
+
"train_speed(iter/s)": 0.005639
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.11960828287358899,
|
| 896 |
+
"grad_norm": 3.378316879272461,
|
| 897 |
+
"learning_rate": 0.00015873015873015873,
|
| 898 |
+
"logits/chosen": -0.9473586678504944,
|
| 899 |
+
"logits/rejected": -0.958495020866394,
|
| 900 |
+
"logps/chosen": -7.192904472351074,
|
| 901 |
+
"logps/rejected": -38.12461853027344,
|
| 902 |
+
"loss": 0.8130012154579163,
|
| 903 |
+
"memory(GiB)": 46.78,
|
| 904 |
+
"nll_loss": 0.4022335410118103,
|
| 905 |
+
"rewards/accuracies": 0.78125,
|
| 906 |
+
"rewards/chosen": 0.14782631397247314,
|
| 907 |
+
"rewards/margins": 1.2623234987258911,
|
| 908 |
+
"rewards/rejected": -1.114497184753418,
|
| 909 |
+
"step": 50,
|
| 910 |
+
"train_speed(iter/s)": 0.005639
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.12200044853106078,
|
| 914 |
+
"grad_norm": 3.7333006858825684,
|
| 915 |
+
"learning_rate": 0.00016190476190476192,
|
| 916 |
+
"logits/chosen": -0.9992814064025879,
|
| 917 |
+
"logits/rejected": -1.0156452655792236,
|
| 918 |
+
"logps/chosen": -10.031129837036133,
|
| 919 |
+
"logps/rejected": -41.15076446533203,
|
| 920 |
+
"loss": 0.8442965745925903,
|
| 921 |
+
"memory(GiB)": 46.78,
|
| 922 |
+
"nll_loss": 0.538231611251831,
|
| 923 |
+
"rewards/accuracies": 0.96875,
|
| 924 |
+
"rewards/chosen": 0.17759491503238678,
|
| 925 |
+
"rewards/margins": 1.8191030025482178,
|
| 926 |
+
"rewards/rejected": -1.6415079832077026,
|
| 927 |
+
"step": 51,
|
| 928 |
+
"train_speed(iter/s)": 0.005631
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.12439261418853255,
|
| 932 |
+
"grad_norm": 5.16239070892334,
|
| 933 |
+
"learning_rate": 0.0001650793650793651,
|
| 934 |
+
"logits/chosen": -1.0308305025100708,
|
| 935 |
+
"logits/rejected": -1.0365955829620361,
|
| 936 |
+
"logps/chosen": -12.618029594421387,
|
| 937 |
+
"logps/rejected": -30.713321685791016,
|
| 938 |
+
"loss": 1.0460460186004639,
|
| 939 |
+
"memory(GiB)": 46.78,
|
| 940 |
+
"nll_loss": 0.6516051292419434,
|
| 941 |
+
"rewards/accuracies": 0.8125,
|
| 942 |
+
"rewards/chosen": 0.07658500969409943,
|
| 943 |
+
"rewards/margins": 1.1642749309539795,
|
| 944 |
+
"rewards/rejected": -1.0876898765563965,
|
| 945 |
+
"step": 52,
|
| 946 |
+
"train_speed(iter/s)": 0.005632
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.12678477984600434,
|
| 950 |
+
"grad_norm": 8.370670318603516,
|
| 951 |
+
"learning_rate": 0.00016825396825396826,
|
| 952 |
+
"logits/chosen": -0.9728441834449768,
|
| 953 |
+
"logits/rejected": -0.9805364012718201,
|
| 954 |
+
"logps/chosen": -15.092721939086914,
|
| 955 |
+
"logps/rejected": -43.16249465942383,
|
| 956 |
+
"loss": 1.3046923875808716,
|
| 957 |
+
"memory(GiB)": 46.78,
|
| 958 |
+
"nll_loss": 0.8397946953773499,
|
| 959 |
+
"rewards/accuracies": 0.71875,
|
| 960 |
+
"rewards/chosen": -0.006958425045013428,
|
| 961 |
+
"rewards/margins": 1.5226078033447266,
|
| 962 |
+
"rewards/rejected": -1.5295664072036743,
|
| 963 |
+
"step": 53,
|
| 964 |
+
"train_speed(iter/s)": 0.005632
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.1291769455034761,
|
| 968 |
+
"grad_norm": 4.0541462898254395,
|
| 969 |
+
"learning_rate": 0.00017142857142857143,
|
| 970 |
+
"logits/chosen": -1.0004466772079468,
|
| 971 |
+
"logits/rejected": -1.0036160945892334,
|
| 972 |
+
"logps/chosen": -10.588321685791016,
|
| 973 |
+
"logps/rejected": -31.380657196044922,
|
| 974 |
+
"loss": 0.784751296043396,
|
| 975 |
+
"memory(GiB)": 46.78,
|
| 976 |
+
"nll_loss": 0.43249669671058655,
|
| 977 |
+
"rewards/accuracies": 0.84375,
|
| 978 |
+
"rewards/chosen": 0.19054383039474487,
|
| 979 |
+
"rewards/margins": 1.5502821207046509,
|
| 980 |
+
"rewards/rejected": -1.3597382307052612,
|
| 981 |
+
"step": 54,
|
| 982 |
+
"train_speed(iter/s)": 0.005632
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.13156911116094788,
|
| 986 |
+
"grad_norm": 5.073096752166748,
|
| 987 |
+
"learning_rate": 0.00017460317460317462,
|
| 988 |
+
"logits/chosen": -1.0498322248458862,
|
| 989 |
+
"logits/rejected": -1.0584659576416016,
|
| 990 |
+
"logps/chosen": -15.89664363861084,
|
| 991 |
+
"logps/rejected": -31.34156608581543,
|
| 992 |
+
"loss": 1.277337908744812,
|
| 993 |
+
"memory(GiB)": 46.78,
|
| 994 |
+
"nll_loss": 0.8222866654396057,
|
| 995 |
+
"rewards/accuracies": 0.78125,
|
| 996 |
+
"rewards/chosen": -0.005467001348733902,
|
| 997 |
+
"rewards/margins": 1.1543742418289185,
|
| 998 |
+
"rewards/rejected": -1.1598412990570068,
|
| 999 |
+
"step": 55,
|
| 1000 |
+
"train_speed(iter/s)": 0.005634
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.13396127681841968,
|
| 1004 |
+
"grad_norm": 9.963640213012695,
|
| 1005 |
+
"learning_rate": 0.00017777777777777779,
|
| 1006 |
+
"logits/chosen": -0.9661000370979309,
|
| 1007 |
+
"logits/rejected": -0.9783132672309875,
|
| 1008 |
+
"logps/chosen": -10.20169448852539,
|
| 1009 |
+
"logps/rejected": -36.709903717041016,
|
| 1010 |
+
"loss": 1.0616700649261475,
|
| 1011 |
+
"memory(GiB)": 46.78,
|
| 1012 |
+
"nll_loss": 0.6122196912765503,
|
| 1013 |
+
"rewards/accuracies": 0.78125,
|
| 1014 |
+
"rewards/chosen": 0.12718632817268372,
|
| 1015 |
+
"rewards/margins": 1.107292890548706,
|
| 1016 |
+
"rewards/rejected": -0.9801065921783447,
|
| 1017 |
+
"step": 56,
|
| 1018 |
+
"train_speed(iter/s)": 0.005633
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.13635344247589146,
|
| 1022 |
+
"grad_norm": 4.3275227546691895,
|
| 1023 |
+
"learning_rate": 0.00018095238095238095,
|
| 1024 |
+
"logits/chosen": -1.0224848985671997,
|
| 1025 |
+
"logits/rejected": -1.0346163511276245,
|
| 1026 |
+
"logps/chosen": -12.20643424987793,
|
| 1027 |
+
"logps/rejected": -46.608978271484375,
|
| 1028 |
+
"loss": 0.8705050349235535,
|
| 1029 |
+
"memory(GiB)": 46.78,
|
| 1030 |
+
"nll_loss": 0.501814067363739,
|
| 1031 |
+
"rewards/accuracies": 0.75,
|
| 1032 |
+
"rewards/chosen": 0.06085830181837082,
|
| 1033 |
+
"rewards/margins": 1.5555297136306763,
|
| 1034 |
+
"rewards/rejected": -1.49467134475708,
|
| 1035 |
+
"step": 57,
|
| 1036 |
+
"train_speed(iter/s)": 0.005632
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.13874560813336323,
|
| 1040 |
+
"grad_norm": 7.873358726501465,
|
| 1041 |
+
"learning_rate": 0.00018412698412698412,
|
| 1042 |
+
"logits/chosen": -0.9996501207351685,
|
| 1043 |
+
"logits/rejected": -1.008366346359253,
|
| 1044 |
+
"logps/chosen": -8.571969985961914,
|
| 1045 |
+
"logps/rejected": -36.11326599121094,
|
| 1046 |
+
"loss": 0.8140489459037781,
|
| 1047 |
+
"memory(GiB)": 46.78,
|
| 1048 |
+
"nll_loss": 0.4696846008300781,
|
| 1049 |
+
"rewards/accuracies": 0.8125,
|
| 1050 |
+
"rewards/chosen": 0.10605719685554504,
|
| 1051 |
+
"rewards/margins": 1.5816770792007446,
|
| 1052 |
+
"rewards/rejected": -1.475619912147522,
|
| 1053 |
+
"step": 58,
|
| 1054 |
+
"train_speed(iter/s)": 0.005632
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.141137773790835,
|
| 1058 |
+
"grad_norm": 4.191744804382324,
|
| 1059 |
+
"learning_rate": 0.00018730158730158731,
|
| 1060 |
+
"logits/chosen": -0.9680933952331543,
|
| 1061 |
+
"logits/rejected": -0.9731469750404358,
|
| 1062 |
+
"logps/chosen": -11.40235710144043,
|
| 1063 |
+
"logps/rejected": -45.129981994628906,
|
| 1064 |
+
"loss": 0.8071048259735107,
|
| 1065 |
+
"memory(GiB)": 46.78,
|
| 1066 |
+
"nll_loss": 0.45748206973075867,
|
| 1067 |
+
"rewards/accuracies": 0.875,
|
| 1068 |
+
"rewards/chosen": -0.023274360224604607,
|
| 1069 |
+
"rewards/margins": 2.327709913253784,
|
| 1070 |
+
"rewards/rejected": -2.3509843349456787,
|
| 1071 |
+
"step": 59,
|
| 1072 |
+
"train_speed(iter/s)": 0.005631
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 0.1435299394483068,
|
| 1076 |
+
"grad_norm": 3.393984317779541,
|
| 1077 |
+
"learning_rate": 0.00019047619047619048,
|
| 1078 |
+
"logits/chosen": -0.9939959645271301,
|
| 1079 |
+
"logits/rejected": -0.9985793232917786,
|
| 1080 |
+
"logps/chosen": -10.470319747924805,
|
| 1081 |
+
"logps/rejected": -34.314205169677734,
|
| 1082 |
+
"loss": 0.8613054156303406,
|
| 1083 |
+
"memory(GiB)": 46.78,
|
| 1084 |
+
"nll_loss": 0.5560427904129028,
|
| 1085 |
+
"rewards/accuracies": 0.84375,
|
| 1086 |
+
"rewards/chosen": 0.24216137826442719,
|
| 1087 |
+
"rewards/margins": 1.7053587436676025,
|
| 1088 |
+
"rewards/rejected": -1.4631972312927246,
|
| 1089 |
+
"step": 60,
|
| 1090 |
+
"train_speed(iter/s)": 0.005629
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.14592210510577858,
|
| 1094 |
+
"grad_norm": 2.9272122383117676,
|
| 1095 |
+
"learning_rate": 0.00019365079365079365,
|
| 1096 |
+
"logits/chosen": -0.9692378044128418,
|
| 1097 |
+
"logits/rejected": -0.9806127548217773,
|
| 1098 |
+
"logps/chosen": -9.072213172912598,
|
| 1099 |
+
"logps/rejected": -40.46186828613281,
|
| 1100 |
+
"loss": 0.7413874864578247,
|
| 1101 |
+
"memory(GiB)": 46.78,
|
| 1102 |
+
"nll_loss": 0.42558133602142334,
|
| 1103 |
+
"rewards/accuracies": 0.84375,
|
| 1104 |
+
"rewards/chosen": 0.12333947420120239,
|
| 1105 |
+
"rewards/margins": 1.8263578414916992,
|
| 1106 |
+
"rewards/rejected": -1.703018307685852,
|
| 1107 |
+
"step": 61,
|
| 1108 |
+
"train_speed(iter/s)": 0.005632
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.14831427076325035,
|
| 1112 |
+
"grad_norm": 6.832886219024658,
|
| 1113 |
+
"learning_rate": 0.00019682539682539682,
|
| 1114 |
+
"logits/chosen": -0.9575268030166626,
|
| 1115 |
+
"logits/rejected": -0.9714451432228088,
|
| 1116 |
+
"logps/chosen": -12.573237419128418,
|
| 1117 |
+
"logps/rejected": -46.78509521484375,
|
| 1118 |
+
"loss": 1.1069118976593018,
|
| 1119 |
+
"memory(GiB)": 46.78,
|
| 1120 |
+
"nll_loss": 0.7730302214622498,
|
| 1121 |
+
"rewards/accuracies": 0.84375,
|
| 1122 |
+
"rewards/chosen": -0.055254869163036346,
|
| 1123 |
+
"rewards/margins": 1.9246735572814941,
|
| 1124 |
+
"rewards/rejected": -1.9799282550811768,
|
| 1125 |
+
"step": 62,
|
| 1126 |
+
"train_speed(iter/s)": 0.00563
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 0.15070643642072212,
|
| 1130 |
+
"grad_norm": 3.042998790740967,
|
| 1131 |
+
"learning_rate": 0.0002,
|
| 1132 |
+
"logits/chosen": -0.982741117477417,
|
| 1133 |
+
"logits/rejected": -0.9914990663528442,
|
| 1134 |
+
"logps/chosen": -7.9191131591796875,
|
| 1135 |
+
"logps/rejected": -37.919273376464844,
|
| 1136 |
+
"loss": 0.7625957131385803,
|
| 1137 |
+
"memory(GiB)": 46.78,
|
| 1138 |
+
"nll_loss": 0.39874428510665894,
|
| 1139 |
+
"rewards/accuracies": 0.78125,
|
| 1140 |
+
"rewards/chosen": 0.21871888637542725,
|
| 1141 |
+
"rewards/margins": 1.7384941577911377,
|
| 1142 |
+
"rewards/rejected": -1.5197750329971313,
|
| 1143 |
+
"step": 63,
|
| 1144 |
+
"train_speed(iter/s)": 0.005632
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.15309860207819392,
|
| 1148 |
+
"grad_norm": 4.333725452423096,
|
| 1149 |
+
"learning_rate": 0.00019999965210677203,
|
| 1150 |
+
"logits/chosen": -0.923725962638855,
|
| 1151 |
+
"logits/rejected": -0.9260438084602356,
|
| 1152 |
+
"logps/chosen": -10.865621566772461,
|
| 1153 |
+
"logps/rejected": -28.435291290283203,
|
| 1154 |
+
"loss": 1.0750609636306763,
|
| 1155 |
+
"memory(GiB)": 46.78,
|
| 1156 |
+
"nll_loss": 0.5927442908287048,
|
| 1157 |
+
"rewards/accuracies": 0.6875,
|
| 1158 |
+
"rewards/chosen": 0.0741436630487442,
|
| 1159 |
+
"rewards/margins": 0.9901498556137085,
|
| 1160 |
+
"rewards/rejected": -0.9160062074661255,
|
| 1161 |
+
"step": 64,
|
| 1162 |
+
"train_speed(iter/s)": 0.005632
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.1554907677356657,
|
| 1166 |
+
"grad_norm": 3.861036539077759,
|
| 1167 |
+
"learning_rate": 0.00019999860842950868,
|
| 1168 |
+
"logits/chosen": -0.9479286670684814,
|
| 1169 |
+
"logits/rejected": -0.9511004686355591,
|
| 1170 |
+
"logps/chosen": -14.088654518127441,
|
| 1171 |
+
"logps/rejected": -34.085487365722656,
|
| 1172 |
+
"loss": 0.8642847537994385,
|
| 1173 |
+
"memory(GiB)": 46.78,
|
| 1174 |
+
"nll_loss": 0.4594627618789673,
|
| 1175 |
+
"rewards/accuracies": 0.8125,
|
| 1176 |
+
"rewards/chosen": -0.02415481209754944,
|
| 1177 |
+
"rewards/margins": 1.243844747543335,
|
| 1178 |
+
"rewards/rejected": -1.2679994106292725,
|
| 1179 |
+
"step": 65,
|
| 1180 |
+
"train_speed(iter/s)": 0.005631
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.15788293339313747,
|
| 1184 |
+
"grad_norm": 2.964956283569336,
|
| 1185 |
+
"learning_rate": 0.00019999686897547169,
|
| 1186 |
+
"logits/chosen": -0.8965855240821838,
|
| 1187 |
+
"logits/rejected": -0.9040545225143433,
|
| 1188 |
+
"logps/chosen": -11.898802757263184,
|
| 1189 |
+
"logps/rejected": -30.001935958862305,
|
| 1190 |
+
"loss": 0.9386196136474609,
|
| 1191 |
+
"memory(GiB)": 46.78,
|
| 1192 |
+
"nll_loss": 0.5237320065498352,
|
| 1193 |
+
"rewards/accuracies": 0.78125,
|
| 1194 |
+
"rewards/chosen": 0.20598752796649933,
|
| 1195 |
+
"rewards/margins": 1.1080772876739502,
|
| 1196 |
+
"rewards/rejected": -0.9020897150039673,
|
| 1197 |
+
"step": 66,
|
| 1198 |
+
"train_speed(iter/s)": 0.005632
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.16027509905060924,
|
| 1202 |
+
"grad_norm": 2.645886182785034,
|
| 1203 |
+
"learning_rate": 0.000199994433756764,
|
| 1204 |
+
"logits/chosen": -0.9297692775726318,
|
| 1205 |
+
"logits/rejected": -0.9331662058830261,
|
| 1206 |
+
"logps/chosen": -9.838333129882812,
|
| 1207 |
+
"logps/rejected": -31.623937606811523,
|
| 1208 |
+
"loss": 0.9248643517494202,
|
| 1209 |
+
"memory(GiB)": 46.78,
|
| 1210 |
+
"nll_loss": 0.5388163328170776,
|
| 1211 |
+
"rewards/accuracies": 0.875,
|
| 1212 |
+
"rewards/chosen": 0.30869799852371216,
|
| 1213 |
+
"rewards/margins": 1.5470086336135864,
|
| 1214 |
+
"rewards/rejected": -1.2383108139038086,
|
| 1215 |
+
"step": 67,
|
| 1216 |
+
"train_speed(iter/s)": 0.005633
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.16266726470808104,
|
| 1220 |
+
"grad_norm": 4.206631660461426,
|
| 1221 |
+
"learning_rate": 0.00019999130279032952,
|
| 1222 |
+
"logits/chosen": -0.8769949674606323,
|
| 1223 |
+
"logits/rejected": -0.8823909759521484,
|
| 1224 |
+
"logps/chosen": -18.815860748291016,
|
| 1225 |
+
"logps/rejected": -40.8411865234375,
|
| 1226 |
+
"loss": 1.2017946243286133,
|
| 1227 |
+
"memory(GiB)": 46.78,
|
| 1228 |
+
"nll_loss": 0.7082366347312927,
|
| 1229 |
+
"rewards/accuracies": 0.75,
|
| 1230 |
+
"rewards/chosen": -0.15441593527793884,
|
| 1231 |
+
"rewards/margins": 1.270005702972412,
|
| 1232 |
+
"rewards/rejected": -1.4244215488433838,
|
| 1233 |
+
"step": 68,
|
| 1234 |
+
"train_speed(iter/s)": 0.005631
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.16505943036555282,
|
| 1238 |
+
"grad_norm": 2.9063844680786133,
|
| 1239 |
+
"learning_rate": 0.00019998747609795308,
|
| 1240 |
+
"logits/chosen": -0.9327405691146851,
|
| 1241 |
+
"logits/rejected": -0.9352924823760986,
|
| 1242 |
+
"logps/chosen": -8.574593544006348,
|
| 1243 |
+
"logps/rejected": -26.808048248291016,
|
| 1244 |
+
"loss": 0.9598045349121094,
|
| 1245 |
+
"memory(GiB)": 46.78,
|
| 1246 |
+
"nll_loss": 0.48059219121932983,
|
| 1247 |
+
"rewards/accuracies": 0.6875,
|
| 1248 |
+
"rewards/chosen": 0.14917421340942383,
|
| 1249 |
+
"rewards/margins": 1.1800099611282349,
|
| 1250 |
+
"rewards/rejected": -1.0308358669281006,
|
| 1251 |
+
"step": 69,
|
| 1252 |
+
"train_speed(iter/s)": 0.005633
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.1674515960230246,
|
| 1256 |
+
"grad_norm": 3.173595428466797,
|
| 1257 |
+
"learning_rate": 0.00019998295370626026,
|
| 1258 |
+
"logits/chosen": -0.9241899847984314,
|
| 1259 |
+
"logits/rejected": -0.9277300834655762,
|
| 1260 |
+
"logps/chosen": -8.667724609375,
|
| 1261 |
+
"logps/rejected": -33.7216911315918,
|
| 1262 |
+
"loss": 0.7956899404525757,
|
| 1263 |
+
"memory(GiB)": 46.78,
|
| 1264 |
+
"nll_loss": 0.46254533529281616,
|
| 1265 |
+
"rewards/accuracies": 0.84375,
|
| 1266 |
+
"rewards/chosen": 0.3548792004585266,
|
| 1267 |
+
"rewards/margins": 1.690018653869629,
|
| 1268 |
+
"rewards/rejected": -1.3351393938064575,
|
| 1269 |
+
"step": 70,
|
| 1270 |
+
"train_speed(iter/s)": 0.005632
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.16984376168049636,
|
| 1274 |
+
"grad_norm": 2.7610597610473633,
|
| 1275 |
+
"learning_rate": 0.00019997773564671732,
|
| 1276 |
+
"logits/chosen": -0.9446437358856201,
|
| 1277 |
+
"logits/rejected": -0.948798656463623,
|
| 1278 |
+
"logps/chosen": -7.366062164306641,
|
| 1279 |
+
"logps/rejected": -33.738685607910156,
|
| 1280 |
+
"loss": 0.7108196020126343,
|
| 1281 |
+
"memory(GiB)": 46.78,
|
| 1282 |
+
"nll_loss": 0.34850752353668213,
|
| 1283 |
+
"rewards/accuracies": 0.78125,
|
| 1284 |
+
"rewards/chosen": 0.19515307247638702,
|
| 1285 |
+
"rewards/margins": 1.521824836730957,
|
| 1286 |
+
"rewards/rejected": -1.3266719579696655,
|
| 1287 |
+
"step": 71,
|
| 1288 |
+
"train_speed(iter/s)": 0.00563
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 0.17223592733796816,
|
| 1292 |
+
"grad_norm": 2.687474250793457,
|
| 1293 |
+
"learning_rate": 0.00019997182195563072,
|
| 1294 |
+
"logits/chosen": -0.9068199396133423,
|
| 1295 |
+
"logits/rejected": -0.9117379188537598,
|
| 1296 |
+
"logps/chosen": -11.47612476348877,
|
| 1297 |
+
"logps/rejected": -36.21818161010742,
|
| 1298 |
+
"loss": 0.7809538245201111,
|
| 1299 |
+
"memory(GiB)": 46.78,
|
| 1300 |
+
"nll_loss": 0.4381601810455322,
|
| 1301 |
+
"rewards/accuracies": 0.90625,
|
| 1302 |
+
"rewards/chosen": 0.23770935833454132,
|
| 1303 |
+
"rewards/margins": 1.4757200479507446,
|
| 1304 |
+
"rewards/rejected": -1.2380106449127197,
|
| 1305 |
+
"step": 72,
|
| 1306 |
+
"train_speed(iter/s)": 0.005633
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.17462809299543994,
|
| 1310 |
+
"grad_norm": 4.630646228790283,
|
| 1311 |
+
"learning_rate": 0.00019996521267414717,
|
| 1312 |
+
"logits/chosen": -0.9765245914459229,
|
| 1313 |
+
"logits/rejected": -0.9796733856201172,
|
| 1314 |
+
"logps/chosen": -18.482873916625977,
|
| 1315 |
+
"logps/rejected": -29.491580963134766,
|
| 1316 |
+
"loss": 1.1687482595443726,
|
| 1317 |
+
"memory(GiB)": 46.78,
|
| 1318 |
+
"nll_loss": 0.703139066696167,
|
| 1319 |
+
"rewards/accuracies": 0.71875,
|
| 1320 |
+
"rewards/chosen": 0.22344376146793365,
|
| 1321 |
+
"rewards/margins": 1.3359349966049194,
|
| 1322 |
+
"rewards/rejected": -1.1124913692474365,
|
| 1323 |
+
"step": 73,
|
| 1324 |
+
"train_speed(iter/s)": 0.005635
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.1770202586529117,
|
| 1328 |
+
"grad_norm": 2.8936681747436523,
|
| 1329 |
+
"learning_rate": 0.00019995790784825318,
|
| 1330 |
+
"logits/chosen": -0.9390259385108948,
|
| 1331 |
+
"logits/rejected": -0.9489994049072266,
|
| 1332 |
+
"logps/chosen": -7.563755035400391,
|
| 1333 |
+
"logps/rejected": -45.333255767822266,
|
| 1334 |
+
"loss": 0.6042292714118958,
|
| 1335 |
+
"memory(GiB)": 46.78,
|
| 1336 |
+
"nll_loss": 0.29600095748901367,
|
| 1337 |
+
"rewards/accuracies": 0.84375,
|
| 1338 |
+
"rewards/chosen": 0.2876776158809662,
|
| 1339 |
+
"rewards/margins": 2.2362325191497803,
|
| 1340 |
+
"rewards/rejected": -1.9485547542572021,
|
| 1341 |
+
"step": 74,
|
| 1342 |
+
"train_speed(iter/s)": 0.005634
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.17941242431038348,
|
| 1346 |
+
"grad_norm": 7.294376850128174,
|
| 1347 |
+
"learning_rate": 0.00019994990752877472,
|
| 1348 |
+
"logits/chosen": -0.9351106882095337,
|
| 1349 |
+
"logits/rejected": -0.9450594186782837,
|
| 1350 |
+
"logps/chosen": -8.70926570892334,
|
| 1351 |
+
"logps/rejected": -40.92999267578125,
|
| 1352 |
+
"loss": 0.8546091914176941,
|
| 1353 |
+
"memory(GiB)": 46.78,
|
| 1354 |
+
"nll_loss": 0.4996742010116577,
|
| 1355 |
+
"rewards/accuracies": 0.78125,
|
| 1356 |
+
"rewards/chosen": 0.15468531847000122,
|
| 1357 |
+
"rewards/margins": 1.8762543201446533,
|
| 1358 |
+
"rewards/rejected": -1.7215690612792969,
|
| 1359 |
+
"step": 75,
|
| 1360 |
+
"train_speed(iter/s)": 0.005635
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.18180458996785528,
|
| 1364 |
+
"grad_norm": 3.972262382507324,
|
| 1365 |
+
"learning_rate": 0.0001999412117713769,
|
| 1366 |
+
"logits/chosen": -1.0458097457885742,
|
| 1367 |
+
"logits/rejected": -1.0525654554367065,
|
| 1368 |
+
"logps/chosen": -14.451887130737305,
|
| 1369 |
+
"logps/rejected": -50.781612396240234,
|
| 1370 |
+
"loss": 1.1873810291290283,
|
| 1371 |
+
"memory(GiB)": 46.78,
|
| 1372 |
+
"nll_loss": 0.7520004510879517,
|
| 1373 |
+
"rewards/accuracies": 0.78125,
|
| 1374 |
+
"rewards/chosen": -0.15246914327144623,
|
| 1375 |
+
"rewards/margins": 1.8416060209274292,
|
| 1376 |
+
"rewards/rejected": -1.9940751791000366,
|
| 1377 |
+
"step": 76,
|
| 1378 |
+
"train_speed(iter/s)": 0.005634
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.18419675562532706,
|
| 1382 |
+
"grad_norm": 4.1366729736328125,
|
| 1383 |
+
"learning_rate": 0.00019993182063656365,
|
| 1384 |
+
"logits/chosen": -0.9752227663993835,
|
| 1385 |
+
"logits/rejected": -0.9808940887451172,
|
| 1386 |
+
"logps/chosen": -13.090173721313477,
|
| 1387 |
+
"logps/rejected": -46.20813751220703,
|
| 1388 |
+
"loss": 0.9597315788269043,
|
| 1389 |
+
"memory(GiB)": 46.78,
|
| 1390 |
+
"nll_loss": 0.564085841178894,
|
| 1391 |
+
"rewards/accuracies": 0.75,
|
| 1392 |
+
"rewards/chosen": 0.06369027495384216,
|
| 1393 |
+
"rewards/margins": 2.183162212371826,
|
| 1394 |
+
"rewards/rejected": -2.119471788406372,
|
| 1395 |
+
"step": 77,
|
| 1396 |
+
"train_speed(iter/s)": 0.005635
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.18658892128279883,
|
| 1400 |
+
"grad_norm": 7.812252044677734,
|
| 1401 |
+
"learning_rate": 0.00019992173418967722,
|
| 1402 |
+
"logits/chosen": -1.0128633975982666,
|
| 1403 |
+
"logits/rejected": -1.0182214975357056,
|
| 1404 |
+
"logps/chosen": -8.333160400390625,
|
| 1405 |
+
"logps/rejected": -49.87608337402344,
|
| 1406 |
+
"loss": 0.7557947635650635,
|
| 1407 |
+
"memory(GiB)": 46.78,
|
| 1408 |
+
"nll_loss": 0.5276219248771667,
|
| 1409 |
+
"rewards/accuracies": 0.875,
|
| 1410 |
+
"rewards/chosen": 0.05898723006248474,
|
| 1411 |
+
"rewards/margins": 3.0209884643554688,
|
| 1412 |
+
"rewards/rejected": -2.962001323699951,
|
| 1413 |
+
"step": 78,
|
| 1414 |
+
"train_speed(iter/s)": 0.005637
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.1889810869402706,
|
| 1418 |
+
"grad_norm": 5.186205863952637,
|
| 1419 |
+
"learning_rate": 0.0001999109525008977,
|
| 1420 |
+
"logits/chosen": -1.006530523300171,
|
| 1421 |
+
"logits/rejected": -1.0176763534545898,
|
| 1422 |
+
"logps/chosen": -16.32330894470215,
|
| 1423 |
+
"logps/rejected": -42.1322135925293,
|
| 1424 |
+
"loss": 1.0281399488449097,
|
| 1425 |
+
"memory(GiB)": 46.78,
|
| 1426 |
+
"nll_loss": 0.6972037553787231,
|
| 1427 |
+
"rewards/accuracies": 0.84375,
|
| 1428 |
+
"rewards/chosen": 0.12587852776050568,
|
| 1429 |
+
"rewards/margins": 2.273554801940918,
|
| 1430 |
+
"rewards/rejected": -2.1476762294769287,
|
| 1431 |
+
"step": 79,
|
| 1432 |
+
"train_speed(iter/s)": 0.005635
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 0.1913732525977424,
|
| 1436 |
+
"grad_norm": 3.2977113723754883,
|
| 1437 |
+
"learning_rate": 0.00019989947564524267,
|
| 1438 |
+
"logits/chosen": -0.9115131497383118,
|
| 1439 |
+
"logits/rejected": -0.9150525331497192,
|
| 1440 |
+
"logps/chosen": -14.462236404418945,
|
| 1441 |
+
"logps/rejected": -37.0766487121582,
|
| 1442 |
+
"loss": 0.9416806697845459,
|
| 1443 |
+
"memory(GiB)": 46.78,
|
| 1444 |
+
"nll_loss": 0.618413507938385,
|
| 1445 |
+
"rewards/accuracies": 0.875,
|
| 1446 |
+
"rewards/chosen": 0.1239091008901596,
|
| 1447 |
+
"rewards/margins": 1.5363889932632446,
|
| 1448 |
+
"rewards/rejected": -1.4124797582626343,
|
| 1449 |
+
"step": 80,
|
| 1450 |
+
"train_speed(iter/s)": 0.005635
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.19376541825521418,
|
| 1454 |
+
"grad_norm": 2.5254037380218506,
|
| 1455 |
+
"learning_rate": 0.00019988730370256653,
|
| 1456 |
+
"logits/chosen": -0.923732340335846,
|
| 1457 |
+
"logits/rejected": -0.9374791383743286,
|
| 1458 |
+
"logps/chosen": -7.155979156494141,
|
| 1459 |
+
"logps/rejected": -48.05502700805664,
|
| 1460 |
+
"loss": 0.7125514149665833,
|
| 1461 |
+
"memory(GiB)": 46.78,
|
| 1462 |
+
"nll_loss": 0.3992992639541626,
|
| 1463 |
+
"rewards/accuracies": 0.84375,
|
| 1464 |
+
"rewards/chosen": 0.3067445456981659,
|
| 1465 |
+
"rewards/margins": 2.08925461769104,
|
| 1466 |
+
"rewards/rejected": -1.7825100421905518,
|
| 1467 |
+
"step": 81,
|
| 1468 |
+
"train_speed(iter/s)": 0.005636
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 0.19615758391268595,
|
| 1472 |
+
"grad_norm": 9.468795776367188,
|
| 1473 |
+
"learning_rate": 0.00019987443675756,
|
| 1474 |
+
"logits/chosen": -0.9325875043869019,
|
| 1475 |
+
"logits/rejected": -0.9395610094070435,
|
| 1476 |
+
"logps/chosen": -10.340322494506836,
|
| 1477 |
+
"logps/rejected": -39.463584899902344,
|
| 1478 |
+
"loss": 0.9755638241767883,
|
| 1479 |
+
"memory(GiB)": 46.78,
|
| 1480 |
+
"nll_loss": 0.6087434887886047,
|
| 1481 |
+
"rewards/accuracies": 0.90625,
|
| 1482 |
+
"rewards/chosen": 0.17063137888908386,
|
| 1483 |
+
"rewards/margins": 1.9463139772415161,
|
| 1484 |
+
"rewards/rejected": -1.775682806968689,
|
| 1485 |
+
"step": 82,
|
| 1486 |
+
"train_speed(iter/s)": 0.005637
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.19854974957015772,
|
| 1490 |
+
"grad_norm": 3.6070539951324463,
|
| 1491 |
+
"learning_rate": 0.00019986087489974952,
|
| 1492 |
+
"logits/chosen": -0.953080952167511,
|
| 1493 |
+
"logits/rejected": -0.9536972045898438,
|
| 1494 |
+
"logps/chosen": -12.9752197265625,
|
| 1495 |
+
"logps/rejected": -32.04649353027344,
|
| 1496 |
+
"loss": 1.0608779191970825,
|
| 1497 |
+
"memory(GiB)": 46.78,
|
| 1498 |
+
"nll_loss": 0.6077963709831238,
|
| 1499 |
+
"rewards/accuracies": 0.75,
|
| 1500 |
+
"rewards/chosen": 0.20461547374725342,
|
| 1501 |
+
"rewards/margins": 1.1314504146575928,
|
| 1502 |
+
"rewards/rejected": -0.9268349409103394,
|
| 1503 |
+
"step": 83,
|
| 1504 |
+
"train_speed(iter/s)": 0.005636
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 0.20094191522762953,
|
| 1508 |
+
"grad_norm": 3.0724594593048096,
|
| 1509 |
+
"learning_rate": 0.00019984661822349667,
|
| 1510 |
+
"logits/chosen": -0.9149019718170166,
|
| 1511 |
+
"logits/rejected": -0.9223135113716125,
|
| 1512 |
+
"logps/chosen": -11.019997596740723,
|
| 1513 |
+
"logps/rejected": -34.67535400390625,
|
| 1514 |
+
"loss": 0.959977924823761,
|
| 1515 |
+
"memory(GiB)": 46.78,
|
| 1516 |
+
"nll_loss": 0.5666155815124512,
|
| 1517 |
+
"rewards/accuracies": 0.8125,
|
| 1518 |
+
"rewards/chosen": 0.07833094894886017,
|
| 1519 |
+
"rewards/margins": 1.359312891960144,
|
| 1520 |
+
"rewards/rejected": -1.2809820175170898,
|
| 1521 |
+
"step": 84,
|
| 1522 |
+
"train_speed(iter/s)": 0.005636
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.2033340808851013,
|
| 1526 |
+
"grad_norm": 4.832602024078369,
|
| 1527 |
+
"learning_rate": 0.0001998316668279975,
|
| 1528 |
+
"logits/chosen": -0.9071481823921204,
|
| 1529 |
+
"logits/rejected": -0.9128391742706299,
|
| 1530 |
+
"logps/chosen": -10.688645362854004,
|
| 1531 |
+
"logps/rejected": -35.94822311401367,
|
| 1532 |
+
"loss": 0.8470439314842224,
|
| 1533 |
+
"memory(GiB)": 46.78,
|
| 1534 |
+
"nll_loss": 0.4921894371509552,
|
| 1535 |
+
"rewards/accuracies": 0.78125,
|
| 1536 |
+
"rewards/chosen": 0.15914583206176758,
|
| 1537 |
+
"rewards/margins": 1.6146492958068848,
|
| 1538 |
+
"rewards/rejected": -1.4555033445358276,
|
| 1539 |
+
"step": 85,
|
| 1540 |
+
"train_speed(iter/s)": 0.005637
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.20572624654257307,
|
| 1544 |
+
"grad_norm": 2.5131633281707764,
|
| 1545 |
+
"learning_rate": 0.00019981602081728178,
|
| 1546 |
+
"logits/chosen": -0.9089372754096985,
|
| 1547 |
+
"logits/rejected": -0.9133983850479126,
|
| 1548 |
+
"logps/chosen": -9.597051620483398,
|
| 1549 |
+
"logps/rejected": -31.3452205657959,
|
| 1550 |
+
"loss": 0.8543159365653992,
|
| 1551 |
+
"memory(GiB)": 46.78,
|
| 1552 |
+
"nll_loss": 0.49588966369628906,
|
| 1553 |
+
"rewards/accuracies": 0.875,
|
| 1554 |
+
"rewards/chosen": 0.37841910123825073,
|
| 1555 |
+
"rewards/margins": 1.7511966228485107,
|
| 1556 |
+
"rewards/rejected": -1.3727774620056152,
|
| 1557 |
+
"step": 86,
|
| 1558 |
+
"train_speed(iter/s)": 0.005636
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.20811841220004484,
|
| 1562 |
+
"grad_norm": 2.388970136642456,
|
| 1563 |
+
"learning_rate": 0.0001997996803002123,
|
| 1564 |
+
"logits/chosen": -0.9073879718780518,
|
| 1565 |
+
"logits/rejected": -0.9152944684028625,
|
| 1566 |
+
"logps/chosen": -9.81510066986084,
|
| 1567 |
+
"logps/rejected": -44.42865753173828,
|
| 1568 |
+
"loss": 0.7038865685462952,
|
| 1569 |
+
"memory(GiB)": 46.78,
|
| 1570 |
+
"nll_loss": 0.4832083582878113,
|
| 1571 |
+
"rewards/accuracies": 0.90625,
|
| 1572 |
+
"rewards/chosen": 0.40937453508377075,
|
| 1573 |
+
"rewards/margins": 2.5496997833251953,
|
| 1574 |
+
"rewards/rejected": -2.1403250694274902,
|
| 1575 |
+
"step": 87,
|
| 1576 |
+
"train_speed(iter/s)": 0.005637
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 0.21051057785751665,
|
| 1580 |
+
"grad_norm": 4.166887283325195,
|
| 1581 |
+
"learning_rate": 0.0001997826453904842,
|
| 1582 |
+
"logits/chosen": -0.9330130815505981,
|
| 1583 |
+
"logits/rejected": -0.9432638883590698,
|
| 1584 |
+
"logps/chosen": -8.299344062805176,
|
| 1585 |
+
"logps/rejected": -45.63969421386719,
|
| 1586 |
+
"loss": 0.8084508180618286,
|
| 1587 |
+
"memory(GiB)": 46.78,
|
| 1588 |
+
"nll_loss": 0.5791301727294922,
|
| 1589 |
+
"rewards/accuracies": 0.9375,
|
| 1590 |
+
"rewards/chosen": 0.28511884808540344,
|
| 1591 |
+
"rewards/margins": 2.360887050628662,
|
| 1592 |
+
"rewards/rejected": -2.07576847076416,
|
| 1593 |
+
"step": 88,
|
| 1594 |
+
"train_speed(iter/s)": 0.005636
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.21290274351498842,
|
| 1598 |
+
"grad_norm": 2.945251703262329,
|
| 1599 |
+
"learning_rate": 0.00019976491620662406,
|
| 1600 |
+
"logits/chosen": -0.9447211623191833,
|
| 1601 |
+
"logits/rejected": -0.9499524235725403,
|
| 1602 |
+
"logps/chosen": -11.17585277557373,
|
| 1603 |
+
"logps/rejected": -39.13389205932617,
|
| 1604 |
+
"loss": 0.8930118680000305,
|
| 1605 |
+
"memory(GiB)": 46.78,
|
| 1606 |
+
"nll_loss": 0.5336206555366516,
|
| 1607 |
+
"rewards/accuracies": 0.84375,
|
| 1608 |
+
"rewards/chosen": 0.2357935607433319,
|
| 1609 |
+
"rewards/margins": 1.8502506017684937,
|
| 1610 |
+
"rewards/rejected": -1.6144568920135498,
|
| 1611 |
+
"step": 89,
|
| 1612 |
+
"train_speed(iter/s)": 0.005651
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.2152949091724602,
|
| 1616 |
+
"grad_norm": 2.779130697250366,
|
| 1617 |
+
"learning_rate": 0.00019974649287198917,
|
| 1618 |
+
"logits/chosen": -0.9709842205047607,
|
| 1619 |
+
"logits/rejected": -0.9703356623649597,
|
| 1620 |
+
"logps/chosen": -14.912643432617188,
|
| 1621 |
+
"logps/rejected": -38.99196243286133,
|
| 1622 |
+
"loss": 0.8809282779693604,
|
| 1623 |
+
"memory(GiB)": 46.78,
|
| 1624 |
+
"nll_loss": 0.5660299062728882,
|
| 1625 |
+
"rewards/accuracies": 0.84375,
|
| 1626 |
+
"rewards/chosen": 0.05061885714530945,
|
| 1627 |
+
"rewards/margins": 2.167189598083496,
|
| 1628 |
+
"rewards/rejected": -2.116570472717285,
|
| 1629 |
+
"step": 90,
|
| 1630 |
+
"train_speed(iter/s)": 0.00565
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 0.21768707482993196,
|
| 1634 |
+
"grad_norm": 18.56264305114746,
|
| 1635 |
+
"learning_rate": 0.00019972737551476657,
|
| 1636 |
+
"logits/chosen": -0.9611991047859192,
|
| 1637 |
+
"logits/rejected": -0.9779796600341797,
|
| 1638 |
+
"logps/chosen": -12.885236740112305,
|
| 1639 |
+
"logps/rejected": -61.122474670410156,
|
| 1640 |
+
"loss": 1.059419870376587,
|
| 1641 |
+
"memory(GiB)": 46.78,
|
| 1642 |
+
"nll_loss": 0.7570096254348755,
|
| 1643 |
+
"rewards/accuracies": 0.8125,
|
| 1644 |
+
"rewards/chosen": -0.2121628373861313,
|
| 1645 |
+
"rewards/margins": 3.2702770233154297,
|
| 1646 |
+
"rewards/rejected": -3.4824395179748535,
|
| 1647 |
+
"step": 91,
|
| 1648 |
+
"train_speed(iter/s)": 0.005648
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.22007924048740377,
|
| 1652 |
+
"grad_norm": 12.53586196899414,
|
| 1653 |
+
"learning_rate": 0.00019970756426797222,
|
| 1654 |
+
"logits/chosen": -0.9786768555641174,
|
| 1655 |
+
"logits/rejected": -0.9855670928955078,
|
| 1656 |
+
"logps/chosen": -11.347864151000977,
|
| 1657 |
+
"logps/rejected": -53.41314697265625,
|
| 1658 |
+
"loss": 1.0533485412597656,
|
| 1659 |
+
"memory(GiB)": 46.78,
|
| 1660 |
+
"nll_loss": 0.7857727408409119,
|
| 1661 |
+
"rewards/accuracies": 0.90625,
|
| 1662 |
+
"rewards/chosen": -0.0066121164709329605,
|
| 1663 |
+
"rewards/margins": 3.5626165866851807,
|
| 1664 |
+
"rewards/rejected": -3.5692288875579834,
|
| 1665 |
+
"step": 92,
|
| 1666 |
+
"train_speed(iter/s)": 0.005648
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.22247140614487554,
|
| 1670 |
+
"grad_norm": 5.693381309509277,
|
| 1671 |
+
"learning_rate": 0.00019968705926945015,
|
| 1672 |
+
"logits/chosen": -0.9693127870559692,
|
| 1673 |
+
"logits/rejected": -0.9756276607513428,
|
| 1674 |
+
"logps/chosen": -12.194255828857422,
|
| 1675 |
+
"logps/rejected": -57.608524322509766,
|
| 1676 |
+
"loss": 0.8848276138305664,
|
| 1677 |
+
"memory(GiB)": 46.78,
|
| 1678 |
+
"nll_loss": 0.5927705764770508,
|
| 1679 |
+
"rewards/accuracies": 0.84375,
|
| 1680 |
+
"rewards/chosen": -0.047529980540275574,
|
| 1681 |
+
"rewards/margins": 3.575960397720337,
|
| 1682 |
+
"rewards/rejected": -3.623490333557129,
|
| 1683 |
+
"step": 93,
|
| 1684 |
+
"train_speed(iter/s)": 0.005648
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.2248635718023473,
|
| 1688 |
+
"grad_norm": 6.475986957550049,
|
| 1689 |
+
"learning_rate": 0.00019966586066187133,
|
| 1690 |
+
"logits/chosen": -0.9738945960998535,
|
| 1691 |
+
"logits/rejected": -0.9793066382408142,
|
| 1692 |
+
"logps/chosen": -13.28403377532959,
|
| 1693 |
+
"logps/rejected": -44.68003845214844,
|
| 1694 |
+
"loss": 1.1366519927978516,
|
| 1695 |
+
"memory(GiB)": 46.78,
|
| 1696 |
+
"nll_loss": 0.6431254148483276,
|
| 1697 |
+
"rewards/accuracies": 0.75,
|
| 1698 |
+
"rewards/chosen": -0.027368444949388504,
|
| 1699 |
+
"rewards/margins": 2.1656975746154785,
|
| 1700 |
+
"rewards/rejected": -2.193066120147705,
|
| 1701 |
+
"step": 94,
|
| 1702 |
+
"train_speed(iter/s)": 0.005648
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.22725573745981908,
|
| 1706 |
+
"grad_norm": 3.099266529083252,
|
| 1707 |
+
"learning_rate": 0.00019964396859273277,
|
| 1708 |
+
"logits/chosen": -0.911146879196167,
|
| 1709 |
+
"logits/rejected": -0.918366551399231,
|
| 1710 |
+
"logps/chosen": -11.675085067749023,
|
| 1711 |
+
"logps/rejected": -43.39706802368164,
|
| 1712 |
+
"loss": 1.0351701974868774,
|
| 1713 |
+
"memory(GiB)": 46.78,
|
| 1714 |
+
"nll_loss": 0.6883416175842285,
|
| 1715 |
+
"rewards/accuracies": 0.875,
|
| 1716 |
+
"rewards/chosen": 0.11412099003791809,
|
| 1717 |
+
"rewards/margins": 2.480591297149658,
|
| 1718 |
+
"rewards/rejected": -2.3664703369140625,
|
| 1719 |
+
"step": 95,
|
| 1720 |
+
"train_speed(iter/s)": 0.005648
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 0.22964790311729089,
|
| 1724 |
+
"grad_norm": 2.586851119995117,
|
| 1725 |
+
"learning_rate": 0.00019962138321435657,
|
| 1726 |
+
"logits/chosen": -0.8519182205200195,
|
| 1727 |
+
"logits/rejected": -0.8646640181541443,
|
| 1728 |
+
"logps/chosen": -9.719616889953613,
|
| 1729 |
+
"logps/rejected": -47.84657669067383,
|
| 1730 |
+
"loss": 0.7686633467674255,
|
| 1731 |
+
"memory(GiB)": 46.78,
|
| 1732 |
+
"nll_loss": 0.5021700859069824,
|
| 1733 |
+
"rewards/accuracies": 0.90625,
|
| 1734 |
+
"rewards/chosen": 0.3157491087913513,
|
| 1735 |
+
"rewards/margins": 2.579385280609131,
|
| 1736 |
+
"rewards/rejected": -2.2636361122131348,
|
| 1737 |
+
"step": 96,
|
| 1738 |
+
"train_speed(iter/s)": 0.005649
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.23204006877476266,
|
| 1742 |
+
"grad_norm": 2.8380484580993652,
|
| 1743 |
+
"learning_rate": 0.0001995981046838887,
|
| 1744 |
+
"logits/chosen": -0.8343847990036011,
|
| 1745 |
+
"logits/rejected": -0.8459703326225281,
|
| 1746 |
+
"logps/chosen": -18.732688903808594,
|
| 1747 |
+
"logps/rejected": -45.6234016418457,
|
| 1748 |
+
"loss": 0.8623945116996765,
|
| 1749 |
+
"memory(GiB)": 46.78,
|
| 1750 |
+
"nll_loss": 0.5681560635566711,
|
| 1751 |
+
"rewards/accuracies": 0.78125,
|
| 1752 |
+
"rewards/chosen": 0.1901763379573822,
|
| 1753 |
+
"rewards/margins": 2.299340009689331,
|
| 1754 |
+
"rewards/rejected": -2.109163522720337,
|
| 1755 |
+
"step": 97,
|
| 1756 |
+
"train_speed(iter/s)": 0.005648
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.23443223443223443,
|
| 1760 |
+
"grad_norm": 4.969798564910889,
|
| 1761 |
+
"learning_rate": 0.00019957413316329807,
|
| 1762 |
+
"logits/chosen": -0.747201681137085,
|
| 1763 |
+
"logits/rejected": -0.7517687082290649,
|
| 1764 |
+
"logps/chosen": -17.974029541015625,
|
| 1765 |
+
"logps/rejected": -45.19165802001953,
|
| 1766 |
+
"loss": 1.0327142477035522,
|
| 1767 |
+
"memory(GiB)": 46.78,
|
| 1768 |
+
"nll_loss": 0.5473703145980835,
|
| 1769 |
+
"rewards/accuracies": 0.75,
|
| 1770 |
+
"rewards/chosen": -0.12275683879852295,
|
| 1771 |
+
"rewards/margins": 2.03201961517334,
|
| 1772 |
+
"rewards/rejected": -2.1547763347625732,
|
| 1773 |
+
"step": 98,
|
| 1774 |
+
"train_speed(iter/s)": 0.005647
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.2368244000897062,
|
| 1778 |
+
"grad_norm": 4.590868949890137,
|
| 1779 |
+
"learning_rate": 0.00019954946881937523,
|
| 1780 |
+
"logits/chosen": -0.7821736335754395,
|
| 1781 |
+
"logits/rejected": -0.7837796807289124,
|
| 1782 |
+
"logps/chosen": -10.349934577941895,
|
| 1783 |
+
"logps/rejected": -31.049577713012695,
|
| 1784 |
+
"loss": 0.8570771217346191,
|
| 1785 |
+
"memory(GiB)": 46.78,
|
| 1786 |
+
"nll_loss": 0.5170512199401855,
|
| 1787 |
+
"rewards/accuracies": 0.84375,
|
| 1788 |
+
"rewards/chosen": 0.12493617832660675,
|
| 1789 |
+
"rewards/margins": 1.5637221336364746,
|
| 1790 |
+
"rewards/rejected": -1.4387857913970947,
|
| 1791 |
+
"step": 99,
|
| 1792 |
+
"train_speed(iter/s)": 0.005648
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 0.23921656574717798,
|
| 1796 |
+
"grad_norm": 3.0355851650238037,
|
| 1797 |
+
"learning_rate": 0.00019952411182373137,
|
| 1798 |
+
"logits/chosen": -0.7416669726371765,
|
| 1799 |
+
"logits/rejected": -0.7514837980270386,
|
| 1800 |
+
"logps/chosen": -12.758575439453125,
|
| 1801 |
+
"logps/rejected": -52.49340057373047,
|
| 1802 |
+
"loss": 0.7629787921905518,
|
| 1803 |
+
"memory(GiB)": 46.78,
|
| 1804 |
+
"nll_loss": 0.4688795804977417,
|
| 1805 |
+
"rewards/accuracies": 0.90625,
|
| 1806 |
+
"rewards/chosen": 0.32235589623451233,
|
| 1807 |
+
"rewards/margins": 2.829197406768799,
|
| 1808 |
+
"rewards/rejected": -2.5068414211273193,
|
| 1809 |
+
"step": 100,
|
| 1810 |
+
"train_speed(iter/s)": 0.005647
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.24160873140464978,
|
| 1814 |
+
"grad_norm": 4.2240214347839355,
|
| 1815 |
+
"learning_rate": 0.000199498062352797,
|
| 1816 |
+
"logits/chosen": -0.7358923554420471,
|
| 1817 |
+
"logits/rejected": -0.7415207624435425,
|
| 1818 |
+
"logps/chosen": -12.469839096069336,
|
| 1819 |
+
"logps/rejected": -39.323734283447266,
|
| 1820 |
+
"loss": 0.8776499629020691,
|
| 1821 |
+
"memory(GiB)": 46.78,
|
| 1822 |
+
"nll_loss": 0.5749576091766357,
|
| 1823 |
+
"rewards/accuracies": 0.875,
|
| 1824 |
+
"rewards/chosen": 0.2892524003982544,
|
| 1825 |
+
"rewards/margins": 1.7897101640701294,
|
| 1826 |
+
"rewards/rejected": -1.5004578828811646,
|
| 1827 |
+
"step": 101,
|
| 1828 |
+
"train_speed(iter/s)": 0.00564
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 0.24400089706212155,
|
| 1832 |
+
"grad_norm": 4.133398532867432,
|
| 1833 |
+
"learning_rate": 0.00019947132058782083,
|
| 1834 |
+
"logits/chosen": -0.7967073321342468,
|
| 1835 |
+
"logits/rejected": -0.7993899583816528,
|
| 1836 |
+
"logps/chosen": -9.669886589050293,
|
| 1837 |
+
"logps/rejected": -25.658836364746094,
|
| 1838 |
+
"loss": 0.9530386924743652,
|
| 1839 |
+
"memory(GiB)": 46.78,
|
| 1840 |
+
"nll_loss": 0.5376937985420227,
|
| 1841 |
+
"rewards/accuracies": 0.78125,
|
| 1842 |
+
"rewards/chosen": 0.09255746752023697,
|
| 1843 |
+
"rewards/margins": 1.1425445079803467,
|
| 1844 |
+
"rewards/rejected": -1.0499870777130127,
|
| 1845 |
+
"step": 102,
|
| 1846 |
+
"train_speed(iter/s)": 0.00564
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.24639306271959333,
|
| 1850 |
+
"grad_norm": 4.5498857498168945,
|
| 1851 |
+
"learning_rate": 0.00019944388671486844,
|
| 1852 |
+
"logits/chosen": -0.7290014028549194,
|
| 1853 |
+
"logits/rejected": -0.7364125847816467,
|
| 1854 |
+
"logps/chosen": -10.035048484802246,
|
| 1855 |
+
"logps/rejected": -38.11687088012695,
|
| 1856 |
+
"loss": 0.8911282420158386,
|
| 1857 |
+
"memory(GiB)": 46.78,
|
| 1858 |
+
"nll_loss": 0.4538458585739136,
|
| 1859 |
+
"rewards/accuracies": 0.78125,
|
| 1860 |
+
"rewards/chosen": 0.18342246115207672,
|
| 1861 |
+
"rewards/margins": 1.3529313802719116,
|
| 1862 |
+
"rewards/rejected": -1.169508934020996,
|
| 1863 |
+
"step": 103,
|
| 1864 |
+
"train_speed(iter/s)": 0.00564
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.2487852283770651,
|
| 1868 |
+
"grad_norm": 2.2867326736450195,
|
| 1869 |
+
"learning_rate": 0.00019941576092482102,
|
| 1870 |
+
"logits/chosen": -0.7742420434951782,
|
| 1871 |
+
"logits/rejected": -0.7834679484367371,
|
| 1872 |
+
"logps/chosen": -16.7891788482666,
|
| 1873 |
+
"logps/rejected": -41.97465133666992,
|
| 1874 |
+
"loss": 0.8852651119232178,
|
| 1875 |
+
"memory(GiB)": 46.78,
|
| 1876 |
+
"nll_loss": 0.5684996843338013,
|
| 1877 |
+
"rewards/accuracies": 0.9375,
|
| 1878 |
+
"rewards/chosen": 0.13892318308353424,
|
| 1879 |
+
"rewards/margins": 1.5296766757965088,
|
| 1880 |
+
"rewards/rejected": -1.3907533884048462,
|
| 1881 |
+
"step": 104,
|
| 1882 |
+
"train_speed(iter/s)": 0.005638
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.25117739403453687,
|
| 1886 |
+
"grad_norm": 2.7168843746185303,
|
| 1887 |
+
"learning_rate": 0.00019938694341337397,
|
| 1888 |
+
"logits/chosen": -0.761275053024292,
|
| 1889 |
+
"logits/rejected": -0.7697631120681763,
|
| 1890 |
+
"logps/chosen": -7.466435432434082,
|
| 1891 |
+
"logps/rejected": -35.15258026123047,
|
| 1892 |
+
"loss": 1.0209513902664185,
|
| 1893 |
+
"memory(GiB)": 46.78,
|
| 1894 |
+
"nll_loss": 0.636327862739563,
|
| 1895 |
+
"rewards/accuracies": 0.84375,
|
| 1896 |
+
"rewards/chosen": 0.23773828148841858,
|
| 1897 |
+
"rewards/margins": 1.3149192333221436,
|
| 1898 |
+
"rewards/rejected": -1.0771809816360474,
|
| 1899 |
+
"step": 105,
|
| 1900 |
+
"train_speed(iter/s)": 0.005638
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.2535695596920087,
|
| 1904 |
+
"grad_norm": 3.2692759037017822,
|
| 1905 |
+
"learning_rate": 0.00019935743438103564,
|
| 1906 |
+
"logits/chosen": -0.7422237992286682,
|
| 1907 |
+
"logits/rejected": -0.7514990568161011,
|
| 1908 |
+
"logps/chosen": -10.624504089355469,
|
| 1909 |
+
"logps/rejected": -45.264095306396484,
|
| 1910 |
+
"loss": 0.860304057598114,
|
| 1911 |
+
"memory(GiB)": 46.78,
|
| 1912 |
+
"nll_loss": 0.5044058561325073,
|
| 1913 |
+
"rewards/accuracies": 0.875,
|
| 1914 |
+
"rewards/chosen": 0.3707730770111084,
|
| 1915 |
+
"rewards/margins": 1.6010938882827759,
|
| 1916 |
+
"rewards/rejected": -1.230320930480957,
|
| 1917 |
+
"step": 106,
|
| 1918 |
+
"train_speed(iter/s)": 0.005637
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.2559617253494805,
|
| 1922 |
+
"grad_norm": 2.6627414226531982,
|
| 1923 |
+
"learning_rate": 0.0001993272340331259,
|
| 1924 |
+
"logits/chosen": -0.75310879945755,
|
| 1925 |
+
"logits/rejected": -0.7589280605316162,
|
| 1926 |
+
"logps/chosen": -11.713288307189941,
|
| 1927 |
+
"logps/rejected": -42.20861053466797,
|
| 1928 |
+
"loss": 0.7711183428764343,
|
| 1929 |
+
"memory(GiB)": 46.78,
|
| 1930 |
+
"nll_loss": 0.43505772948265076,
|
| 1931 |
+
"rewards/accuracies": 0.8125,
|
| 1932 |
+
"rewards/chosen": 0.3599666357040405,
|
| 1933 |
+
"rewards/margins": 1.9370594024658203,
|
| 1934 |
+
"rewards/rejected": -1.5770928859710693,
|
| 1935 |
+
"step": 107,
|
| 1936 |
+
"train_speed(iter/s)": 0.005638
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 0.2583538910069522,
|
| 1940 |
+
"grad_norm": 10.524235725402832,
|
| 1941 |
+
"learning_rate": 0.00019929634257977467,
|
| 1942 |
+
"logits/chosen": -0.7487523555755615,
|
| 1943 |
+
"logits/rejected": -0.7474150657653809,
|
| 1944 |
+
"logps/chosen": -13.211384773254395,
|
| 1945 |
+
"logps/rejected": -29.434967041015625,
|
| 1946 |
+
"loss": 1.4064631462097168,
|
| 1947 |
+
"memory(GiB)": 46.78,
|
| 1948 |
+
"nll_loss": 0.9639731049537659,
|
| 1949 |
+
"rewards/accuracies": 0.75,
|
| 1950 |
+
"rewards/chosen": 0.05272696912288666,
|
| 1951 |
+
"rewards/margins": 1.349626064300537,
|
| 1952 |
+
"rewards/rejected": -1.2968993186950684,
|
| 1953 |
+
"step": 108,
|
| 1954 |
+
"train_speed(iter/s)": 0.005638
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.260746056664424,
|
| 1958 |
+
"grad_norm": 3.4797213077545166,
|
| 1959 |
+
"learning_rate": 0.00019926476023592052,
|
| 1960 |
+
"logits/chosen": -0.7358635067939758,
|
| 1961 |
+
"logits/rejected": -0.7399594783782959,
|
| 1962 |
+
"logps/chosen": -10.68916130065918,
|
| 1963 |
+
"logps/rejected": -39.55656814575195,
|
| 1964 |
+
"loss": 0.8483824133872986,
|
| 1965 |
+
"memory(GiB)": 46.78,
|
| 1966 |
+
"nll_loss": 0.512445867061615,
|
| 1967 |
+
"rewards/accuracies": 0.875,
|
| 1968 |
+
"rewards/chosen": 0.26777422428131104,
|
| 1969 |
+
"rewards/margins": 1.7832790613174438,
|
| 1970 |
+
"rewards/rejected": -1.5155047178268433,
|
| 1971 |
+
"step": 109,
|
| 1972 |
+
"train_speed(iter/s)": 0.00564
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 0.26313822232189576,
|
| 1976 |
+
"grad_norm": 5.318943023681641,
|
| 1977 |
+
"learning_rate": 0.00019923248722130907,
|
| 1978 |
+
"logits/chosen": -0.8071874976158142,
|
| 1979 |
+
"logits/rejected": -0.8125062584877014,
|
| 1980 |
+
"logps/chosen": -13.162405967712402,
|
| 1981 |
+
"logps/rejected": -41.4532356262207,
|
| 1982 |
+
"loss": 0.6827958822250366,
|
| 1983 |
+
"memory(GiB)": 46.78,
|
| 1984 |
+
"nll_loss": 0.39053356647491455,
|
| 1985 |
+
"rewards/accuracies": 0.875,
|
| 1986 |
+
"rewards/chosen": 0.33915120363235474,
|
| 1987 |
+
"rewards/margins": 1.944946050643921,
|
| 1988 |
+
"rewards/rejected": -1.605794906616211,
|
| 1989 |
+
"step": 110,
|
| 1990 |
+
"train_speed(iter/s)": 0.00564
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.26553038797936757,
|
| 1994 |
+
"grad_norm": 6.246517658233643,
|
| 1995 |
+
"learning_rate": 0.0001991995237604916,
|
| 1996 |
+
"logits/chosen": -0.7993965148925781,
|
| 1997 |
+
"logits/rejected": -0.8035753965377808,
|
| 1998 |
+
"logps/chosen": -7.477054595947266,
|
| 1999 |
+
"logps/rejected": -43.46379089355469,
|
| 2000 |
+
"loss": 0.6327071785926819,
|
| 2001 |
+
"memory(GiB)": 46.78,
|
| 2002 |
+
"nll_loss": 0.3665684461593628,
|
| 2003 |
+
"rewards/accuracies": 0.8125,
|
| 2004 |
+
"rewards/chosen": 0.18014857172966003,
|
| 2005 |
+
"rewards/margins": 2.2636728286743164,
|
| 2006 |
+
"rewards/rejected": -2.083524227142334,
|
| 2007 |
+
"step": 111,
|
| 2008 |
+
"train_speed(iter/s)": 0.005641
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.26792255363683937,
|
| 2012 |
+
"grad_norm": 2.495997905731201,
|
| 2013 |
+
"learning_rate": 0.00019916587008282345,
|
| 2014 |
+
"logits/chosen": -0.8536009192466736,
|
| 2015 |
+
"logits/rejected": -0.8555616736412048,
|
| 2016 |
+
"logps/chosen": -12.577634811401367,
|
| 2017 |
+
"logps/rejected": -40.71754837036133,
|
| 2018 |
+
"loss": 0.7153143286705017,
|
| 2019 |
+
"memory(GiB)": 46.78,
|
| 2020 |
+
"nll_loss": 0.4246208667755127,
|
| 2021 |
+
"rewards/accuracies": 0.875,
|
| 2022 |
+
"rewards/chosen": 0.16410964727401733,
|
| 2023 |
+
"rewards/margins": 2.231029748916626,
|
| 2024 |
+
"rewards/rejected": -2.066920042037964,
|
| 2025 |
+
"step": 112,
|
| 2026 |
+
"train_speed(iter/s)": 0.005641
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.2703147192943111,
|
| 2030 |
+
"grad_norm": 2.619420289993286,
|
| 2031 |
+
"learning_rate": 0.00019913152642246233,
|
| 2032 |
+
"logits/chosen": -0.8602064847946167,
|
| 2033 |
+
"logits/rejected": -0.870837926864624,
|
| 2034 |
+
"logps/chosen": -10.31131362915039,
|
| 2035 |
+
"logps/rejected": -69.16264343261719,
|
| 2036 |
+
"loss": 0.6141177415847778,
|
| 2037 |
+
"memory(GiB)": 46.78,
|
| 2038 |
+
"nll_loss": 0.4619813561439514,
|
| 2039 |
+
"rewards/accuracies": 0.96875,
|
| 2040 |
+
"rewards/chosen": 0.4243915379047394,
|
| 2041 |
+
"rewards/margins": 3.666055202484131,
|
| 2042 |
+
"rewards/rejected": -3.241663694381714,
|
| 2043 |
+
"step": 113,
|
| 2044 |
+
"train_speed(iter/s)": 0.00564
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.2727068849517829,
|
| 2048 |
+
"grad_norm": 3.2425930500030518,
|
| 2049 |
+
"learning_rate": 0.00019909649301836675,
|
| 2050 |
+
"logits/chosen": -0.8935564756393433,
|
| 2051 |
+
"logits/rejected": -0.8967443108558655,
|
| 2052 |
+
"logps/chosen": -11.790775299072266,
|
| 2053 |
+
"logps/rejected": -54.12834930419922,
|
| 2054 |
+
"loss": 0.6750085353851318,
|
| 2055 |
+
"memory(GiB)": 46.78,
|
| 2056 |
+
"nll_loss": 0.44180190563201904,
|
| 2057 |
+
"rewards/accuracies": 0.90625,
|
| 2058 |
+
"rewards/chosen": 0.022756237536668777,
|
| 2059 |
+
"rewards/margins": 2.9651176929473877,
|
| 2060 |
+
"rewards/rejected": -2.942361354827881,
|
| 2061 |
+
"step": 114,
|
| 2062 |
+
"train_speed(iter/s)": 0.005639
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.2750990506092547,
|
| 2066 |
+
"grad_norm": 24.632802963256836,
|
| 2067 |
+
"learning_rate": 0.0001990607701142944,
|
| 2068 |
+
"logits/chosen": -0.8810749650001526,
|
| 2069 |
+
"logits/rejected": -0.8888768553733826,
|
| 2070 |
+
"logps/chosen": -23.208826065063477,
|
| 2071 |
+
"logps/rejected": -53.65020751953125,
|
| 2072 |
+
"loss": 1.851314902305603,
|
| 2073 |
+
"memory(GiB)": 46.78,
|
| 2074 |
+
"nll_loss": 1.315173625946045,
|
| 2075 |
+
"rewards/accuracies": 0.78125,
|
| 2076 |
+
"rewards/chosen": -0.5939430594444275,
|
| 2077 |
+
"rewards/margins": 2.5491535663604736,
|
| 2078 |
+
"rewards/rejected": -3.143096685409546,
|
| 2079 |
+
"step": 115,
|
| 2080 |
+
"train_speed(iter/s)": 0.005639
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 0.27749121626672646,
|
| 2084 |
+
"grad_norm": 17.350990295410156,
|
| 2085 |
+
"learning_rate": 0.00019902435795880044,
|
| 2086 |
+
"logits/chosen": -0.9093245267868042,
|
| 2087 |
+
"logits/rejected": -0.9126628637313843,
|
| 2088 |
+
"logps/chosen": -16.813339233398438,
|
| 2089 |
+
"logps/rejected": -48.75463104248047,
|
| 2090 |
+
"loss": 1.4077650308609009,
|
| 2091 |
+
"memory(GiB)": 46.78,
|
| 2092 |
+
"nll_loss": 1.050284743309021,
|
| 2093 |
+
"rewards/accuracies": 0.8125,
|
| 2094 |
+
"rewards/chosen": -0.2693726122379303,
|
| 2095 |
+
"rewards/margins": 2.530968427658081,
|
| 2096 |
+
"rewards/rejected": -2.8003406524658203,
|
| 2097 |
+
"step": 116,
|
| 2098 |
+
"train_speed(iter/s)": 0.005638
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.27988338192419826,
|
| 2102 |
+
"grad_norm": 4.22681188583374,
|
| 2103 |
+
"learning_rate": 0.00019898725680523568,
|
| 2104 |
+
"logits/chosen": -0.8980717062950134,
|
| 2105 |
+
"logits/rejected": -0.9024423956871033,
|
| 2106 |
+
"logps/chosen": -17.1817626953125,
|
| 2107 |
+
"logps/rejected": -39.5125732421875,
|
| 2108 |
+
"loss": 1.0394302606582642,
|
| 2109 |
+
"memory(GiB)": 46.78,
|
| 2110 |
+
"nll_loss": 0.6766522526741028,
|
| 2111 |
+
"rewards/accuracies": 0.78125,
|
| 2112 |
+
"rewards/chosen": 0.015612306073307991,
|
| 2113 |
+
"rewards/margins": 2.2480926513671875,
|
| 2114 |
+
"rewards/rejected": -2.23248028755188,
|
| 2115 |
+
"step": 117,
|
| 2116 |
+
"train_speed(iter/s)": 0.005637
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.28227554758167,
|
| 2120 |
+
"grad_norm": 4.9705047607421875,
|
| 2121 |
+
"learning_rate": 0.00019894946691174494,
|
| 2122 |
+
"logits/chosen": -0.9122129082679749,
|
| 2123 |
+
"logits/rejected": -0.9150456190109253,
|
| 2124 |
+
"logps/chosen": -14.797954559326172,
|
| 2125 |
+
"logps/rejected": -38.31578826904297,
|
| 2126 |
+
"loss": 1.0148959159851074,
|
| 2127 |
+
"memory(GiB)": 46.78,
|
| 2128 |
+
"nll_loss": 0.6429183483123779,
|
| 2129 |
+
"rewards/accuracies": 0.78125,
|
| 2130 |
+
"rewards/chosen": -0.06372372806072235,
|
| 2131 |
+
"rewards/margins": 2.232508659362793,
|
| 2132 |
+
"rewards/rejected": -2.296232223510742,
|
| 2133 |
+
"step": 118,
|
| 2134 |
+
"train_speed(iter/s)": 0.005637
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.2846677132391418,
|
| 2138 |
+
"grad_norm": 5.455178737640381,
|
| 2139 |
+
"learning_rate": 0.00019891098854126518,
|
| 2140 |
+
"logits/chosen": -0.8895492553710938,
|
| 2141 |
+
"logits/rejected": -0.8904259204864502,
|
| 2142 |
+
"logps/chosen": -15.091131210327148,
|
| 2143 |
+
"logps/rejected": -40.248023986816406,
|
| 2144 |
+
"loss": 1.0479692220687866,
|
| 2145 |
+
"memory(GiB)": 46.78,
|
| 2146 |
+
"nll_loss": 0.5974382758140564,
|
| 2147 |
+
"rewards/accuracies": 0.78125,
|
| 2148 |
+
"rewards/chosen": -0.1267915964126587,
|
| 2149 |
+
"rewards/margins": 1.8017388582229614,
|
| 2150 |
+
"rewards/rejected": -1.9285303354263306,
|
| 2151 |
+
"step": 119,
|
| 2152 |
+
"train_speed(iter/s)": 0.005634
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.2870598788966136,
|
| 2156 |
+
"grad_norm": 3.6146156787872314,
|
| 2157 |
+
"learning_rate": 0.00019887182196152368,
|
| 2158 |
+
"logits/chosen": -0.8824920654296875,
|
| 2159 |
+
"logits/rejected": -0.8917353749275208,
|
| 2160 |
+
"logps/chosen": -7.259885311126709,
|
| 2161 |
+
"logps/rejected": -55.807106018066406,
|
| 2162 |
+
"loss": 0.6745078563690186,
|
| 2163 |
+
"memory(GiB)": 46.78,
|
| 2164 |
+
"nll_loss": 0.3909413814544678,
|
| 2165 |
+
"rewards/accuracies": 0.875,
|
| 2166 |
+
"rewards/chosen": 0.12902025878429413,
|
| 2167 |
+
"rewards/margins": 2.945380210876465,
|
| 2168 |
+
"rewards/rejected": -2.8163604736328125,
|
| 2169 |
+
"step": 120,
|
| 2170 |
+
"train_speed(iter/s)": 0.005634
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.28945204455408535,
|
| 2174 |
+
"grad_norm": 2.7159321308135986,
|
| 2175 |
+
"learning_rate": 0.0001988319674450362,
|
| 2176 |
+
"logits/chosen": -0.8625448942184448,
|
| 2177 |
+
"logits/rejected": -0.8625946640968323,
|
| 2178 |
+
"logps/chosen": -14.481998443603516,
|
| 2179 |
+
"logps/rejected": -32.343990325927734,
|
| 2180 |
+
"loss": 0.8893653750419617,
|
| 2181 |
+
"memory(GiB)": 46.78,
|
| 2182 |
+
"nll_loss": 0.5466328859329224,
|
| 2183 |
+
"rewards/accuracies": 0.8125,
|
| 2184 |
+
"rewards/chosen": 0.1943284571170807,
|
| 2185 |
+
"rewards/margins": 1.5906785726547241,
|
| 2186 |
+
"rewards/rejected": -1.3963501453399658,
|
| 2187 |
+
"step": 121,
|
| 2188 |
+
"train_speed(iter/s)": 0.005634
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 0.29184421021155715,
|
| 2192 |
+
"grad_norm": 3.480830430984497,
|
| 2193 |
+
"learning_rate": 0.0001987914252691051,
|
| 2194 |
+
"logits/chosen": -0.854420006275177,
|
| 2195 |
+
"logits/rejected": -0.8605158925056458,
|
| 2196 |
+
"logps/chosen": -11.787839889526367,
|
| 2197 |
+
"logps/rejected": -43.2979850769043,
|
| 2198 |
+
"loss": 0.7196720838546753,
|
| 2199 |
+
"memory(GiB)": 46.78,
|
| 2200 |
+
"nll_loss": 0.5012603402137756,
|
| 2201 |
+
"rewards/accuracies": 0.96875,
|
| 2202 |
+
"rewards/chosen": 0.2540440559387207,
|
| 2203 |
+
"rewards/margins": 2.362658977508545,
|
| 2204 |
+
"rewards/rejected": -2.1086151599884033,
|
| 2205 |
+
"step": 122,
|
| 2206 |
+
"train_speed(iter/s)": 0.005633
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.29423637586902895,
|
| 2210 |
+
"grad_norm": 3.526841163635254,
|
| 2211 |
+
"learning_rate": 0.00019875019571581728,
|
| 2212 |
+
"logits/chosen": -0.8358349204063416,
|
| 2213 |
+
"logits/rejected": -0.8408463597297668,
|
| 2214 |
+
"logps/chosen": -10.75222110748291,
|
| 2215 |
+
"logps/rejected": -39.87123489379883,
|
| 2216 |
+
"loss": 0.8948889970779419,
|
| 2217 |
+
"memory(GiB)": 46.78,
|
| 2218 |
+
"nll_loss": 0.5788209438323975,
|
| 2219 |
+
"rewards/accuracies": 0.84375,
|
| 2220 |
+
"rewards/chosen": 0.15971140563488007,
|
| 2221 |
+
"rewards/margins": 2.309971809387207,
|
| 2222 |
+
"rewards/rejected": -2.1502604484558105,
|
| 2223 |
+
"step": 123,
|
| 2224 |
+
"train_speed(iter/s)": 0.005633
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.2966285415265007,
|
| 2228 |
+
"grad_norm": 2.6714415550231934,
|
| 2229 |
+
"learning_rate": 0.00019870827907204243,
|
| 2230 |
+
"logits/chosen": -0.8233751654624939,
|
| 2231 |
+
"logits/rejected": -0.8268409967422485,
|
| 2232 |
+
"logps/chosen": -7.600090503692627,
|
| 2233 |
+
"logps/rejected": -23.002704620361328,
|
| 2234 |
+
"loss": 0.8616309762001038,
|
| 2235 |
+
"memory(GiB)": 46.78,
|
| 2236 |
+
"nll_loss": 0.43116894364356995,
|
| 2237 |
+
"rewards/accuracies": 0.78125,
|
| 2238 |
+
"rewards/chosen": 0.28889453411102295,
|
| 2239 |
+
"rewards/margins": 1.081323504447937,
|
| 2240 |
+
"rewards/rejected": -0.7924288511276245,
|
| 2241 |
+
"step": 124,
|
| 2242 |
+
"train_speed(iter/s)": 0.005633
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.2990207071839725,
|
| 2246 |
+
"grad_norm": 2.5394530296325684,
|
| 2247 |
+
"learning_rate": 0.0001986656756294309,
|
| 2248 |
+
"logits/chosen": -0.8398711085319519,
|
| 2249 |
+
"logits/rejected": -0.8451254963874817,
|
| 2250 |
+
"logps/chosen": -9.005830764770508,
|
| 2251 |
+
"logps/rejected": -33.04253387451172,
|
| 2252 |
+
"loss": 0.8945890665054321,
|
| 2253 |
+
"memory(GiB)": 46.78,
|
| 2254 |
+
"nll_loss": 0.4701317250728607,
|
| 2255 |
+
"rewards/accuracies": 0.75,
|
| 2256 |
+
"rewards/chosen": 0.3293447494506836,
|
| 2257 |
+
"rewards/margins": 1.3195619583129883,
|
| 2258 |
+
"rewards/rejected": -0.9902173280715942,
|
| 2259 |
+
"step": 125,
|
| 2260 |
+
"train_speed(iter/s)": 0.005633
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.30141287284144425,
|
| 2264 |
+
"grad_norm": 2.8723111152648926,
|
| 2265 |
+
"learning_rate": 0.00019862238568441165,
|
| 2266 |
+
"logits/chosen": -0.815192699432373,
|
| 2267 |
+
"logits/rejected": -0.8217712044715881,
|
| 2268 |
+
"logps/chosen": -9.070804595947266,
|
| 2269 |
+
"logps/rejected": -33.22805404663086,
|
| 2270 |
+
"loss": 0.7637823820114136,
|
| 2271 |
+
"memory(GiB)": 46.78,
|
| 2272 |
+
"nll_loss": 0.38797256350517273,
|
| 2273 |
+
"rewards/accuracies": 0.84375,
|
| 2274 |
+
"rewards/chosen": 0.17756348848342896,
|
| 2275 |
+
"rewards/margins": 1.2646557092666626,
|
| 2276 |
+
"rewards/rejected": -1.0870922803878784,
|
| 2277 |
+
"step": 126,
|
| 2278 |
+
"train_speed(iter/s)": 0.005633
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.30380503849891605,
|
| 2282 |
+
"grad_norm": 2.79227614402771,
|
| 2283 |
+
"learning_rate": 0.00019857840953819026,
|
| 2284 |
+
"logits/chosen": -0.8166335821151733,
|
| 2285 |
+
"logits/rejected": -0.8273878693580627,
|
| 2286 |
+
"logps/chosen": -9.302844047546387,
|
| 2287 |
+
"logps/rejected": -32.995567321777344,
|
| 2288 |
+
"loss": 0.7958816289901733,
|
| 2289 |
+
"memory(GiB)": 46.78,
|
| 2290 |
+
"nll_loss": 0.4939688444137573,
|
| 2291 |
+
"rewards/accuracies": 0.875,
|
| 2292 |
+
"rewards/chosen": 0.3217732310295105,
|
| 2293 |
+
"rewards/margins": 1.6436134576797485,
|
| 2294 |
+
"rewards/rejected": -1.3218401670455933,
|
| 2295 |
+
"step": 127,
|
| 2296 |
+
"train_speed(iter/s)": 0.005633
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 0.30619720415638785,
|
| 2300 |
+
"grad_norm": 2.4456770420074463,
|
| 2301 |
+
"learning_rate": 0.00019853374749674677,
|
| 2302 |
+
"logits/chosen": -0.8173456192016602,
|
| 2303 |
+
"logits/rejected": -0.8242702484130859,
|
| 2304 |
+
"logps/chosen": -10.526348114013672,
|
| 2305 |
+
"logps/rejected": -49.73170471191406,
|
| 2306 |
+
"loss": 0.6932899951934814,
|
| 2307 |
+
"memory(GiB)": 46.78,
|
| 2308 |
+
"nll_loss": 0.410495400428772,
|
| 2309 |
+
"rewards/accuracies": 0.84375,
|
| 2310 |
+
"rewards/chosen": 0.2942225933074951,
|
| 2311 |
+
"rewards/margins": 2.532844305038452,
|
| 2312 |
+
"rewards/rejected": -2.238621711730957,
|
| 2313 |
+
"step": 128,
|
| 2314 |
+
"train_speed(iter/s)": 0.005634
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.3085893698138596,
|
| 2318 |
+
"grad_norm": 9.124894142150879,
|
| 2319 |
+
"learning_rate": 0.00019848839987083366,
|
| 2320 |
+
"logits/chosen": -0.8869707584381104,
|
| 2321 |
+
"logits/rejected": -0.8904052376747131,
|
| 2322 |
+
"logps/chosen": -15.062342643737793,
|
| 2323 |
+
"logps/rejected": -45.34114456176758,
|
| 2324 |
+
"loss": 1.0228577852249146,
|
| 2325 |
+
"memory(GiB)": 46.78,
|
| 2326 |
+
"nll_loss": 0.5322936773300171,
|
| 2327 |
+
"rewards/accuracies": 0.84375,
|
| 2328 |
+
"rewards/chosen": -0.14518728852272034,
|
| 2329 |
+
"rewards/margins": 2.259458541870117,
|
| 2330 |
+
"rewards/rejected": -2.4046459197998047,
|
| 2331 |
+
"step": 129,
|
| 2332 |
+
"train_speed(iter/s)": 0.005633
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.3109815354713314,
|
| 2336 |
+
"grad_norm": 3.8776865005493164,
|
| 2337 |
+
"learning_rate": 0.00019844236697597354,
|
| 2338 |
+
"logits/chosen": -0.8990283012390137,
|
| 2339 |
+
"logits/rejected": -0.912631630897522,
|
| 2340 |
+
"logps/chosen": -9.924826622009277,
|
| 2341 |
+
"logps/rejected": -59.806114196777344,
|
| 2342 |
+
"loss": 0.6793578863143921,
|
| 2343 |
+
"memory(GiB)": 46.78,
|
| 2344 |
+
"nll_loss": 0.4696520268917084,
|
| 2345 |
+
"rewards/accuracies": 0.90625,
|
| 2346 |
+
"rewards/chosen": 0.21068614721298218,
|
| 2347 |
+
"rewards/margins": 3.6280405521392822,
|
| 2348 |
+
"rewards/rejected": -3.4173545837402344,
|
| 2349 |
+
"step": 130,
|
| 2350 |
+
"train_speed(iter/s)": 0.005633
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 0.3133737011288032,
|
| 2354 |
+
"grad_norm": 5.517581462860107,
|
| 2355 |
+
"learning_rate": 0.0001983956491324571,
|
| 2356 |
+
"logits/chosen": -0.890454888343811,
|
| 2357 |
+
"logits/rejected": -0.8954933881759644,
|
| 2358 |
+
"logps/chosen": -11.391611099243164,
|
| 2359 |
+
"logps/rejected": -50.16078567504883,
|
| 2360 |
+
"loss": 0.9601025581359863,
|
| 2361 |
+
"memory(GiB)": 46.78,
|
| 2362 |
+
"nll_loss": 0.588812530040741,
|
| 2363 |
+
"rewards/accuracies": 0.78125,
|
| 2364 |
+
"rewards/chosen": 0.01075834035873413,
|
| 2365 |
+
"rewards/margins": 3.3035926818847656,
|
| 2366 |
+
"rewards/rejected": -3.292834758758545,
|
| 2367 |
+
"step": 131,
|
| 2368 |
+
"train_speed(iter/s)": 0.005634
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.31576586678627494,
|
| 2372 |
+
"grad_norm": 2.991635322570801,
|
| 2373 |
+
"learning_rate": 0.0001983482466653407,
|
| 2374 |
+
"logits/chosen": -0.8738813400268555,
|
| 2375 |
+
"logits/rejected": -0.870830774307251,
|
| 2376 |
+
"logps/chosen": -8.894745826721191,
|
| 2377 |
+
"logps/rejected": -39.2451171875,
|
| 2378 |
+
"loss": 0.6562129855155945,
|
| 2379 |
+
"memory(GiB)": 46.78,
|
| 2380 |
+
"nll_loss": 0.34973058104515076,
|
| 2381 |
+
"rewards/accuracies": 0.9375,
|
| 2382 |
+
"rewards/chosen": 0.2841479778289795,
|
| 2383 |
+
"rewards/margins": 2.604597330093384,
|
| 2384 |
+
"rewards/rejected": -2.3204493522644043,
|
| 2385 |
+
"step": 132,
|
| 2386 |
+
"train_speed(iter/s)": 0.005635
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.31815803244374674,
|
| 2390 |
+
"grad_norm": 8.708478927612305,
|
| 2391 |
+
"learning_rate": 0.00019830015990444434,
|
| 2392 |
+
"logits/chosen": -0.8821411728858948,
|
| 2393 |
+
"logits/rejected": -0.891589879989624,
|
| 2394 |
+
"logps/chosen": -18.36916732788086,
|
| 2395 |
+
"logps/rejected": -58.93795394897461,
|
| 2396 |
+
"loss": 0.8125941753387451,
|
| 2397 |
+
"memory(GiB)": 46.78,
|
| 2398 |
+
"nll_loss": 0.6013258695602417,
|
| 2399 |
+
"rewards/accuracies": 0.9375,
|
| 2400 |
+
"rewards/chosen": -0.01710733026266098,
|
| 2401 |
+
"rewards/margins": 2.8880860805511475,
|
| 2402 |
+
"rewards/rejected": -2.905193567276001,
|
| 2403 |
+
"step": 133,
|
| 2404 |
+
"train_speed(iter/s)": 0.005635
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.3205501981012185,
|
| 2408 |
+
"grad_norm": 5.8355326652526855,
|
| 2409 |
+
"learning_rate": 0.00019825138918434915,
|
| 2410 |
+
"logits/chosen": -0.8907016515731812,
|
| 2411 |
+
"logits/rejected": -0.9023689031600952,
|
| 2412 |
+
"logps/chosen": -7.357381343841553,
|
| 2413 |
+
"logps/rejected": -43.13167190551758,
|
| 2414 |
+
"loss": 0.8315655589103699,
|
| 2415 |
+
"memory(GiB)": 46.78,
|
| 2416 |
+
"nll_loss": 0.5653954148292542,
|
| 2417 |
+
"rewards/accuracies": 0.90625,
|
| 2418 |
+
"rewards/chosen": 0.18780827522277832,
|
| 2419 |
+
"rewards/margins": 2.5148556232452393,
|
| 2420 |
+
"rewards/rejected": -2.327047348022461,
|
| 2421 |
+
"step": 134,
|
| 2422 |
+
"train_speed(iter/s)": 0.005634
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.3229423637586903,
|
| 2426 |
+
"grad_norm": 5.084504127502441,
|
| 2427 |
+
"learning_rate": 0.0001982019348443952,
|
| 2428 |
+
"logits/chosen": -0.8744340538978577,
|
| 2429 |
+
"logits/rejected": -0.8839950561523438,
|
| 2430 |
+
"logps/chosen": -11.753833770751953,
|
| 2431 |
+
"logps/rejected": -50.23076248168945,
|
| 2432 |
+
"loss": 0.900967001914978,
|
| 2433 |
+
"memory(GiB)": 46.78,
|
| 2434 |
+
"nll_loss": 0.6376667618751526,
|
| 2435 |
+
"rewards/accuracies": 0.90625,
|
| 2436 |
+
"rewards/chosen": 0.2352873831987381,
|
| 2437 |
+
"rewards/margins": 2.9518003463745117,
|
| 2438 |
+
"rewards/rejected": -2.71651291847229,
|
| 2439 |
+
"step": 135,
|
| 2440 |
+
"train_speed(iter/s)": 0.005634
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 0.3253345294161621,
|
| 2444 |
+
"grad_norm": 7.443116664886475,
|
| 2445 |
+
"learning_rate": 0.00019815179722867913,
|
| 2446 |
+
"logits/chosen": -0.8482657670974731,
|
| 2447 |
+
"logits/rejected": -0.8486264944076538,
|
| 2448 |
+
"logps/chosen": -13.961491584777832,
|
| 2449 |
+
"logps/rejected": -33.860713958740234,
|
| 2450 |
+
"loss": 1.0923020839691162,
|
| 2451 |
+
"memory(GiB)": 46.78,
|
| 2452 |
+
"nll_loss": 0.626233696937561,
|
| 2453 |
+
"rewards/accuracies": 0.75,
|
| 2454 |
+
"rewards/chosen": 0.035439833998680115,
|
| 2455 |
+
"rewards/margins": 1.6097826957702637,
|
| 2456 |
+
"rewards/rejected": -1.574342966079712,
|
| 2457 |
+
"step": 136,
|
| 2458 |
+
"train_speed(iter/s)": 0.005634
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.32772669507363383,
|
| 2462 |
+
"grad_norm": 6.069121360778809,
|
| 2463 |
+
"learning_rate": 0.0001981009766860517,
|
| 2464 |
+
"logits/chosen": -0.8712698817253113,
|
| 2465 |
+
"logits/rejected": -0.8716377019882202,
|
| 2466 |
+
"logps/chosen": -12.955726623535156,
|
| 2467 |
+
"logps/rejected": -39.7027587890625,
|
| 2468 |
+
"loss": 1.063010811805725,
|
| 2469 |
+
"memory(GiB)": 46.78,
|
| 2470 |
+
"nll_loss": 0.7573559284210205,
|
| 2471 |
+
"rewards/accuracies": 0.84375,
|
| 2472 |
+
"rewards/chosen": -0.1307298243045807,
|
| 2473 |
+
"rewards/margins": 2.164278507232666,
|
| 2474 |
+
"rewards/rejected": -2.295008420944214,
|
| 2475 |
+
"step": 137,
|
| 2476 |
+
"train_speed(iter/s)": 0.005634
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.33011886073110563,
|
| 2480 |
+
"grad_norm": 4.233928203582764,
|
| 2481 |
+
"learning_rate": 0.00019804947357011522,
|
| 2482 |
+
"logits/chosen": -0.8236069679260254,
|
| 2483 |
+
"logits/rejected": -0.834182858467102,
|
| 2484 |
+
"logps/chosen": -11.049253463745117,
|
| 2485 |
+
"logps/rejected": -44.08565902709961,
|
| 2486 |
+
"loss": 0.8465083241462708,
|
| 2487 |
+
"memory(GiB)": 46.78,
|
| 2488 |
+
"nll_loss": 0.5441975593566895,
|
| 2489 |
+
"rewards/accuracies": 0.875,
|
| 2490 |
+
"rewards/chosen": 0.11046641319990158,
|
| 2491 |
+
"rewards/margins": 2.231780767440796,
|
| 2492 |
+
"rewards/rejected": -2.121314525604248,
|
| 2493 |
+
"step": 138,
|
| 2494 |
+
"train_speed(iter/s)": 0.005633
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.33251102638857744,
|
| 2498 |
+
"grad_norm": 4.2061309814453125,
|
| 2499 |
+
"learning_rate": 0.00019799728823922155,
|
| 2500 |
+
"logits/chosen": -0.8435903191566467,
|
| 2501 |
+
"logits/rejected": -0.8470354080200195,
|
| 2502 |
+
"logps/chosen": -8.918343544006348,
|
| 2503 |
+
"logps/rejected": -36.74750900268555,
|
| 2504 |
+
"loss": 0.7144109010696411,
|
| 2505 |
+
"memory(GiB)": 46.78,
|
| 2506 |
+
"nll_loss": 0.40702298283576965,
|
| 2507 |
+
"rewards/accuracies": 0.875,
|
| 2508 |
+
"rewards/chosen": 0.38820797204971313,
|
| 2509 |
+
"rewards/margins": 1.9121365547180176,
|
| 2510 |
+
"rewards/rejected": -1.5239286422729492,
|
| 2511 |
+
"step": 139,
|
| 2512 |
+
"train_speed(iter/s)": 0.005634
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.3349031920460492,
|
| 2516 |
+
"grad_norm": 4.416022300720215,
|
| 2517 |
+
"learning_rate": 0.0001979444210564691,
|
| 2518 |
+
"logits/chosen": -0.7840769290924072,
|
| 2519 |
+
"logits/rejected": -0.7934098839759827,
|
| 2520 |
+
"logps/chosen": -8.80458927154541,
|
| 2521 |
+
"logps/rejected": -47.96394348144531,
|
| 2522 |
+
"loss": 0.6305175423622131,
|
| 2523 |
+
"memory(GiB)": 46.78,
|
| 2524 |
+
"nll_loss": 0.46931201219558716,
|
| 2525 |
+
"rewards/accuracies": 1.0,
|
| 2526 |
+
"rewards/chosen": 0.1097404956817627,
|
| 2527 |
+
"rewards/margins": 2.367628335952759,
|
| 2528 |
+
"rewards/rejected": -2.257887840270996,
|
| 2529 |
+
"step": 140,
|
| 2530 |
+
"train_speed(iter/s)": 0.005636
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.337295357703521,
|
| 2534 |
+
"grad_norm": 4.785271644592285,
|
| 2535 |
+
"learning_rate": 0.00019789087238970054,
|
| 2536 |
+
"logits/chosen": -0.8002479076385498,
|
| 2537 |
+
"logits/rejected": -0.8085525035858154,
|
| 2538 |
+
"logps/chosen": -13.70367431640625,
|
| 2539 |
+
"logps/rejected": -50.61127853393555,
|
| 2540 |
+
"loss": 0.9849874377250671,
|
| 2541 |
+
"memory(GiB)": 46.78,
|
| 2542 |
+
"nll_loss": 0.6505732536315918,
|
| 2543 |
+
"rewards/accuracies": 0.875,
|
| 2544 |
+
"rewards/chosen": 0.002065986394882202,
|
| 2545 |
+
"rewards/margins": 2.2779290676116943,
|
| 2546 |
+
"rewards/rejected": -2.2758634090423584,
|
| 2547 |
+
"step": 141,
|
| 2548 |
+
"train_speed(iter/s)": 0.005636
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.3396875233609927,
|
| 2552 |
+
"grad_norm": 4.302982330322266,
|
| 2553 |
+
"learning_rate": 0.0001978366426115003,
|
| 2554 |
+
"logits/chosen": -0.7700361013412476,
|
| 2555 |
+
"logits/rejected": -0.7705273628234863,
|
| 2556 |
+
"logps/chosen": -7.580388069152832,
|
| 2557 |
+
"logps/rejected": -34.4312744140625,
|
| 2558 |
+
"loss": 0.5671791434288025,
|
| 2559 |
+
"memory(GiB)": 46.78,
|
| 2560 |
+
"nll_loss": 0.33011525869369507,
|
| 2561 |
+
"rewards/accuracies": 0.90625,
|
| 2562 |
+
"rewards/chosen": 0.3449672758579254,
|
| 2563 |
+
"rewards/margins": 2.0864133834838867,
|
| 2564 |
+
"rewards/rejected": -1.7414460182189941,
|
| 2565 |
+
"step": 142,
|
| 2566 |
+
"train_speed(iter/s)": 0.005637
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 0.34207968901846453,
|
| 2570 |
+
"grad_norm": 4.768328666687012,
|
| 2571 |
+
"learning_rate": 0.00019778173209919174,
|
| 2572 |
+
"logits/chosen": -0.7594372034072876,
|
| 2573 |
+
"logits/rejected": -0.7637089490890503,
|
| 2574 |
+
"logps/chosen": -14.413496971130371,
|
| 2575 |
+
"logps/rejected": -35.399288177490234,
|
| 2576 |
+
"loss": 0.8721976280212402,
|
| 2577 |
+
"memory(GiB)": 46.78,
|
| 2578 |
+
"nll_loss": 0.6180708408355713,
|
| 2579 |
+
"rewards/accuracies": 0.90625,
|
| 2580 |
+
"rewards/chosen": 0.4894520044326782,
|
| 2581 |
+
"rewards/margins": 1.8526508808135986,
|
| 2582 |
+
"rewards/rejected": -1.3631988763809204,
|
| 2583 |
+
"step": 143,
|
| 2584 |
+
"train_speed(iter/s)": 0.005636
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.34447185467593633,
|
| 2588 |
+
"grad_norm": 2.74961519241333,
|
| 2589 |
+
"learning_rate": 0.00019772614123483487,
|
| 2590 |
+
"logits/chosen": -0.778110146522522,
|
| 2591 |
+
"logits/rejected": -0.7844111919403076,
|
| 2592 |
+
"logps/chosen": -15.620856285095215,
|
| 2593 |
+
"logps/rejected": -41.484710693359375,
|
| 2594 |
+
"loss": 0.7825542092323303,
|
| 2595 |
+
"memory(GiB)": 46.78,
|
| 2596 |
+
"nll_loss": 0.49021872878074646,
|
| 2597 |
+
"rewards/accuracies": 0.875,
|
| 2598 |
+
"rewards/chosen": 0.1902773678302765,
|
| 2599 |
+
"rewards/margins": 1.998608112335205,
|
| 2600 |
+
"rewards/rejected": -1.808330774307251,
|
| 2601 |
+
"step": 144,
|
| 2602 |
+
"train_speed(iter/s)": 0.005634
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.3468640203334081,
|
| 2606 |
+
"grad_norm": 4.053314208984375,
|
| 2607 |
+
"learning_rate": 0.00019766987040522333,
|
| 2608 |
+
"logits/chosen": -0.8054400682449341,
|
| 2609 |
+
"logits/rejected": -0.8061996698379517,
|
| 2610 |
+
"logps/chosen": -11.948256492614746,
|
| 2611 |
+
"logps/rejected": -25.110414505004883,
|
| 2612 |
+
"loss": 1.033211350440979,
|
| 2613 |
+
"memory(GiB)": 46.78,
|
| 2614 |
+
"nll_loss": 0.6106947064399719,
|
| 2615 |
+
"rewards/accuracies": 0.78125,
|
| 2616 |
+
"rewards/chosen": 0.13057959079742432,
|
| 2617 |
+
"rewards/margins": 1.2587168216705322,
|
| 2618 |
+
"rewards/rejected": -1.128137230873108,
|
| 2619 |
+
"step": 145,
|
| 2620 |
+
"train_speed(iter/s)": 0.005634
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.3492561859908799,
|
| 2624 |
+
"grad_norm": 2.9867749214172363,
|
| 2625 |
+
"learning_rate": 0.00019761292000188197,
|
| 2626 |
+
"logits/chosen": -0.8127353191375732,
|
| 2627 |
+
"logits/rejected": -0.8201886415481567,
|
| 2628 |
+
"logps/chosen": -16.549510955810547,
|
| 2629 |
+
"logps/rejected": -38.11656188964844,
|
| 2630 |
+
"loss": 0.9068831205368042,
|
| 2631 |
+
"memory(GiB)": 46.78,
|
| 2632 |
+
"nll_loss": 0.5614981055259705,
|
| 2633 |
+
"rewards/accuracies": 0.84375,
|
| 2634 |
+
"rewards/chosen": 0.20250406861305237,
|
| 2635 |
+
"rewards/margins": 1.7395099401474,
|
| 2636 |
+
"rewards/rejected": -1.5370057821273804,
|
| 2637 |
+
"step": 146,
|
| 2638 |
+
"train_speed(iter/s)": 0.005634
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.3516483516483517,
|
| 2642 |
+
"grad_norm": 2.311192750930786,
|
| 2643 |
+
"learning_rate": 0.00019755529042106394,
|
| 2644 |
+
"logits/chosen": -0.7973526120185852,
|
| 2645 |
+
"logits/rejected": -0.800200343132019,
|
| 2646 |
+
"logps/chosen": -7.481082916259766,
|
| 2647 |
+
"logps/rejected": -31.512645721435547,
|
| 2648 |
+
"loss": 0.7267748713493347,
|
| 2649 |
+
"memory(GiB)": 46.78,
|
| 2650 |
+
"nll_loss": 0.36258465051651,
|
| 2651 |
+
"rewards/accuracies": 0.84375,
|
| 2652 |
+
"rewards/chosen": 0.21067243814468384,
|
| 2653 |
+
"rewards/margins": 1.607694149017334,
|
| 2654 |
+
"rewards/rejected": -1.3970215320587158,
|
| 2655 |
+
"step": 147,
|
| 2656 |
+
"train_speed(iter/s)": 0.005632
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.3540405173058234,
|
| 2660 |
+
"grad_norm": 2.8598074913024902,
|
| 2661 |
+
"learning_rate": 0.00019749698206374808,
|
| 2662 |
+
"logits/chosen": -0.849799633026123,
|
| 2663 |
+
"logits/rejected": -0.8563861846923828,
|
| 2664 |
+
"logps/chosen": -9.258435249328613,
|
| 2665 |
+
"logps/rejected": -43.992244720458984,
|
| 2666 |
+
"loss": 0.6600675582885742,
|
| 2667 |
+
"memory(GiB)": 46.78,
|
| 2668 |
+
"nll_loss": 0.4354017972946167,
|
| 2669 |
+
"rewards/accuracies": 0.9375,
|
| 2670 |
+
"rewards/chosen": 0.33977681398391724,
|
| 2671 |
+
"rewards/margins": 2.25696063041687,
|
| 2672 |
+
"rewards/rejected": -1.9171836376190186,
|
| 2673 |
+
"step": 148,
|
| 2674 |
+
"train_speed(iter/s)": 0.005633
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.3564326829632952,
|
| 2678 |
+
"grad_norm": 2.3342740535736084,
|
| 2679 |
+
"learning_rate": 0.00019743799533563603,
|
| 2680 |
+
"logits/chosen": -0.8605116605758667,
|
| 2681 |
+
"logits/rejected": -0.8653464913368225,
|
| 2682 |
+
"logps/chosen": -7.782684326171875,
|
| 2683 |
+
"logps/rejected": -37.683738708496094,
|
| 2684 |
+
"loss": 0.6275639533996582,
|
| 2685 |
+
"memory(GiB)": 46.78,
|
| 2686 |
+
"nll_loss": 0.4273594915866852,
|
| 2687 |
+
"rewards/accuracies": 0.96875,
|
| 2688 |
+
"rewards/chosen": 0.4285268187522888,
|
| 2689 |
+
"rewards/margins": 2.251387357711792,
|
| 2690 |
+
"rewards/rejected": -1.8228604793548584,
|
| 2691 |
+
"step": 149,
|
| 2692 |
+
"train_speed(iter/s)": 0.005632
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.35882484862076697,
|
| 2696 |
+
"grad_norm": 4.0247483253479,
|
| 2697 |
+
"learning_rate": 0.0001973783306471495,
|
| 2698 |
+
"logits/chosen": -0.8887797594070435,
|
| 2699 |
+
"logits/rejected": -0.898193895816803,
|
| 2700 |
+
"logps/chosen": -8.930159568786621,
|
| 2701 |
+
"logps/rejected": -43.03559112548828,
|
| 2702 |
+
"loss": 0.7508974671363831,
|
| 2703 |
+
"memory(GiB)": 46.78,
|
| 2704 |
+
"nll_loss": 0.518282413482666,
|
| 2705 |
+
"rewards/accuracies": 0.96875,
|
| 2706 |
+
"rewards/chosen": 0.28456175327301025,
|
| 2707 |
+
"rewards/margins": 2.084869146347046,
|
| 2708 |
+
"rewards/rejected": -1.8003073930740356,
|
| 2709 |
+
"step": 150,
|
| 2710 |
+
"train_speed(iter/s)": 0.005634
|
| 2711 |
+
}
|
| 2712 |
+
],
|
| 2713 |
+
"logging_steps": 1,
|
| 2714 |
+
"max_steps": 1254,
|
| 2715 |
+
"num_input_tokens_seen": 0,
|
| 2716 |
+
"num_train_epochs": 3,
|
| 2717 |
+
"save_steps": 50,
|
| 2718 |
+
"stateful_callbacks": {
|
| 2719 |
+
"TrainerControl": {
|
| 2720 |
+
"args": {
|
| 2721 |
+
"should_epoch_stop": false,
|
| 2722 |
+
"should_evaluate": false,
|
| 2723 |
+
"should_log": false,
|
| 2724 |
+
"should_save": true,
|
| 2725 |
+
"should_training_stop": false
|
| 2726 |
+
},
|
| 2727 |
+
"attributes": {}
|
| 2728 |
+
}
|
| 2729 |
+
},
|
| 2730 |
+
"total_flos": 1.0456400650462822e+18,
|
| 2731 |
+
"train_batch_size": 1,
|
| 2732 |
+
"trial_name": null,
|
| 2733 |
+
"trial_params": null
|
| 2734 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:30a677136073f487609898cc6113311ce9918fac33f99d44c163da40fe6035b2
|
| 3 |
+
size 8696
|
zero_to_fp32.py
ADDED
|
@@ -0,0 +1,760 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 gc
|
| 25 |
+
import json
|
| 26 |
+
import numpy as np
|
| 27 |
+
from tqdm import tqdm
|
| 28 |
+
from collections import OrderedDict
|
| 29 |
+
from dataclasses import dataclass
|
| 30 |
+
|
| 31 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 32 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
| 33 |
+
from deepspeed.utils import logger
|
| 34 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 35 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 36 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
@dataclass
|
| 40 |
+
class zero_model_state:
|
| 41 |
+
buffers: dict()
|
| 42 |
+
param_shapes: dict()
|
| 43 |
+
shared_params: list
|
| 44 |
+
ds_version: int
|
| 45 |
+
frozen_param_shapes: dict()
|
| 46 |
+
frozen_param_fragments: dict()
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
debug = 0
|
| 50 |
+
|
| 51 |
+
# load to cpu
|
| 52 |
+
device = torch.device('cpu')
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
def atoi(text):
|
| 56 |
+
return int(text) if text.isdigit() else text
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def natural_keys(text):
|
| 60 |
+
'''
|
| 61 |
+
alist.sort(key=natural_keys) sorts in human order
|
| 62 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 63 |
+
(See Toothy's implementation in the comments)
|
| 64 |
+
'''
|
| 65 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 69 |
+
if not os.path.isdir(checkpoint_dir):
|
| 70 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 71 |
+
|
| 72 |
+
# there should be only one file
|
| 73 |
+
if zero_stage <= 2:
|
| 74 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 75 |
+
elif zero_stage == 3:
|
| 76 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 77 |
+
|
| 78 |
+
if not os.path.exists(file):
|
| 79 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 80 |
+
|
| 81 |
+
return file
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 85 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 86 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 87 |
+
|
| 88 |
+
if len(ckpt_files) == 0:
|
| 89 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 90 |
+
|
| 91 |
+
return ckpt_files
|
| 92 |
+
|
| 93 |
+
|
| 94 |
+
def get_optim_files(checkpoint_dir):
|
| 95 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
def get_model_state_files(checkpoint_dir):
|
| 99 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
def parse_model_states(files):
|
| 103 |
+
zero_model_states = []
|
| 104 |
+
for file in files:
|
| 105 |
+
state_dict = torch.load(file, map_location=device, weights_only=False)
|
| 106 |
+
|
| 107 |
+
if BUFFER_NAMES not in state_dict:
|
| 108 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
| 109 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
| 110 |
+
if debug:
|
| 111 |
+
print("Found buffers:", buffer_names)
|
| 112 |
+
|
| 113 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 114 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 115 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
| 116 |
+
|
| 117 |
+
# collect parameters that are included in param_shapes
|
| 118 |
+
param_names = []
|
| 119 |
+
for s in param_shapes:
|
| 120 |
+
for name in s.keys():
|
| 121 |
+
param_names.append(name)
|
| 122 |
+
|
| 123 |
+
# update with frozen parameters
|
| 124 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 125 |
+
if frozen_param_shapes is not None:
|
| 126 |
+
if debug:
|
| 127 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 128 |
+
param_names += list(frozen_param_shapes.keys())
|
| 129 |
+
|
| 130 |
+
# handle shared params
|
| 131 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 132 |
+
|
| 133 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
| 134 |
+
|
| 135 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 136 |
+
|
| 137 |
+
z_model_state = zero_model_state(buffers=buffers,
|
| 138 |
+
param_shapes=param_shapes,
|
| 139 |
+
shared_params=shared_params,
|
| 140 |
+
ds_version=ds_version,
|
| 141 |
+
frozen_param_shapes=frozen_param_shapes,
|
| 142 |
+
frozen_param_fragments=frozen_param_fragments)
|
| 143 |
+
zero_model_states.append(z_model_state)
|
| 144 |
+
|
| 145 |
+
return zero_model_states
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
| 149 |
+
total_files = len(files)
|
| 150 |
+
state_dicts = []
|
| 151 |
+
for f in tqdm(files, desc='Loading checkpoint shards'):
|
| 152 |
+
state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
|
| 153 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 154 |
+
# and also handle the case where it was already removed by another helper script
|
| 155 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 156 |
+
state_dicts.append(state_dict)
|
| 157 |
+
|
| 158 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 159 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 160 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 161 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 162 |
+
|
| 163 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 164 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 165 |
+
# use the max of the partition_count to get the dp world_size.
|
| 166 |
+
|
| 167 |
+
if type(world_size) is list:
|
| 168 |
+
world_size = max(world_size)
|
| 169 |
+
|
| 170 |
+
if world_size != total_files:
|
| 171 |
+
raise ValueError(
|
| 172 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 173 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 174 |
+
)
|
| 175 |
+
|
| 176 |
+
# the groups are named differently in each stage
|
| 177 |
+
if zero_stage <= 2:
|
| 178 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 179 |
+
elif zero_stage == 3:
|
| 180 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
| 181 |
+
else:
|
| 182 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 183 |
+
|
| 184 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 185 |
+
return zero_stage, world_size, fp32_flat_groups
|
| 186 |
+
|
| 187 |
+
|
| 188 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
|
| 189 |
+
"""
|
| 190 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 191 |
+
|
| 192 |
+
Args:
|
| 193 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 194 |
+
|
| 195 |
+
"""
|
| 196 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 197 |
+
|
| 198 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 199 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 200 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 201 |
+
|
| 202 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 203 |
+
|
| 204 |
+
zero_model_states = parse_model_states(model_files)
|
| 205 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 206 |
+
|
| 207 |
+
if zero_stage <= 2:
|
| 208 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 209 |
+
exclude_frozen_parameters)
|
| 210 |
+
elif zero_stage == 3:
|
| 211 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 212 |
+
exclude_frozen_parameters)
|
| 213 |
+
|
| 214 |
+
|
| 215 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 216 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 217 |
+
return
|
| 218 |
+
|
| 219 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 220 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 221 |
+
|
| 222 |
+
if debug:
|
| 223 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 224 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 225 |
+
|
| 226 |
+
wanted_params = len(frozen_param_shapes)
|
| 227 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 228 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 229 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 230 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 231 |
+
|
| 232 |
+
total_params = 0
|
| 233 |
+
total_numel = 0
|
| 234 |
+
for name, shape in frozen_param_shapes.items():
|
| 235 |
+
total_params += 1
|
| 236 |
+
unpartitioned_numel = shape.numel()
|
| 237 |
+
total_numel += unpartitioned_numel
|
| 238 |
+
|
| 239 |
+
state_dict[name] = frozen_param_fragments[name]
|
| 240 |
+
|
| 241 |
+
if debug:
|
| 242 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 243 |
+
|
| 244 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 245 |
+
|
| 246 |
+
|
| 247 |
+
def _has_callable(obj, fn):
|
| 248 |
+
attr = getattr(obj, fn, None)
|
| 249 |
+
return callable(attr)
|
| 250 |
+
|
| 251 |
+
|
| 252 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 253 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 254 |
+
|
| 255 |
+
# Reconstruction protocol:
|
| 256 |
+
#
|
| 257 |
+
# XXX: document this
|
| 258 |
+
|
| 259 |
+
if debug:
|
| 260 |
+
for i in range(world_size):
|
| 261 |
+
for j in range(len(fp32_flat_groups[0])):
|
| 262 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 263 |
+
|
| 264 |
+
# XXX: memory usage doubles here (zero2)
|
| 265 |
+
num_param_groups = len(fp32_flat_groups[0])
|
| 266 |
+
merged_single_partition_of_fp32_groups = []
|
| 267 |
+
for i in range(num_param_groups):
|
| 268 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 269 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 270 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 271 |
+
avail_numel = sum(
|
| 272 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 273 |
+
|
| 274 |
+
if debug:
|
| 275 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 276 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 277 |
+
# not asserting if there is a mismatch due to possible padding
|
| 278 |
+
print(f"Have {avail_numel} numels to process.")
|
| 279 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 280 |
+
|
| 281 |
+
# params
|
| 282 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 283 |
+
# out-of-core computing solution
|
| 284 |
+
total_numel = 0
|
| 285 |
+
total_params = 0
|
| 286 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 287 |
+
offset = 0
|
| 288 |
+
avail_numel = full_single_fp32_vector.numel()
|
| 289 |
+
for name, shape in shapes.items():
|
| 290 |
+
|
| 291 |
+
unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
|
| 292 |
+
total_numel += unpartitioned_numel
|
| 293 |
+
total_params += 1
|
| 294 |
+
|
| 295 |
+
if debug:
|
| 296 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 297 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 298 |
+
offset += unpartitioned_numel
|
| 299 |
+
|
| 300 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 301 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 302 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 303 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
| 304 |
+
align_to = 2 * world_size
|
| 305 |
+
|
| 306 |
+
def zero2_align(x):
|
| 307 |
+
return align_to * math.ceil(x / align_to)
|
| 308 |
+
|
| 309 |
+
if debug:
|
| 310 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 311 |
+
|
| 312 |
+
offset = zero2_align(offset)
|
| 313 |
+
avail_numel = zero2_align(avail_numel)
|
| 314 |
+
|
| 315 |
+
if debug:
|
| 316 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 317 |
+
|
| 318 |
+
# Sanity check
|
| 319 |
+
if offset != avail_numel:
|
| 320 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 321 |
+
|
| 322 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 323 |
+
|
| 324 |
+
|
| 325 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 326 |
+
exclude_frozen_parameters):
|
| 327 |
+
state_dict = OrderedDict()
|
| 328 |
+
|
| 329 |
+
# buffers
|
| 330 |
+
buffers = zero_model_states[0].buffers
|
| 331 |
+
state_dict.update(buffers)
|
| 332 |
+
if debug:
|
| 333 |
+
print(f"added {len(buffers)} buffers")
|
| 334 |
+
|
| 335 |
+
if not exclude_frozen_parameters:
|
| 336 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 337 |
+
|
| 338 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 339 |
+
|
| 340 |
+
# recover shared parameters
|
| 341 |
+
for pair in zero_model_states[0].shared_params:
|
| 342 |
+
if pair[1] in state_dict:
|
| 343 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 344 |
+
|
| 345 |
+
return state_dict
|
| 346 |
+
|
| 347 |
+
|
| 348 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 349 |
+
remainder = unpartitioned_numel % world_size
|
| 350 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
| 351 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 352 |
+
return partitioned_numel, padding_numel
|
| 353 |
+
|
| 354 |
+
|
| 355 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 356 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 357 |
+
return
|
| 358 |
+
|
| 359 |
+
if debug:
|
| 360 |
+
for i in range(world_size):
|
| 361 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 362 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 363 |
+
|
| 364 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 365 |
+
wanted_params = len(frozen_param_shapes)
|
| 366 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 367 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 368 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 369 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 370 |
+
|
| 371 |
+
total_params = 0
|
| 372 |
+
total_numel = 0
|
| 373 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 374 |
+
total_params += 1
|
| 375 |
+
unpartitioned_numel = shape.numel()
|
| 376 |
+
total_numel += unpartitioned_numel
|
| 377 |
+
|
| 378 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 379 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 380 |
+
|
| 381 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 382 |
+
|
| 383 |
+
if debug:
|
| 384 |
+
print(
|
| 385 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 386 |
+
)
|
| 387 |
+
|
| 388 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 389 |
+
|
| 390 |
+
|
| 391 |
+
class GatheredTensor:
|
| 392 |
+
"""
|
| 393 |
+
A pseudo tensor that collects partitioned weights.
|
| 394 |
+
It is more memory efficient when there are multiple groups.
|
| 395 |
+
"""
|
| 396 |
+
|
| 397 |
+
def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
|
| 398 |
+
self.flat_groups = flat_groups
|
| 399 |
+
self.flat_groups_offset = flat_groups_offset
|
| 400 |
+
self.offset = offset
|
| 401 |
+
self.partitioned_numel = partitioned_numel
|
| 402 |
+
self.shape = shape
|
| 403 |
+
self.dtype = self.flat_groups[0][0].dtype
|
| 404 |
+
|
| 405 |
+
def contiguous(self):
|
| 406 |
+
"""
|
| 407 |
+
Merge partitioned weights from flat_groups into a single tensor.
|
| 408 |
+
"""
|
| 409 |
+
end_idx = self.offset + self.partitioned_numel
|
| 410 |
+
world_size = len(self.flat_groups)
|
| 411 |
+
pad_flat_param_chunks = []
|
| 412 |
+
|
| 413 |
+
for rank_i in range(world_size):
|
| 414 |
+
# for each rank, we need to collect weights from related group/groups
|
| 415 |
+
flat_groups_at_rank_i = self.flat_groups[rank_i]
|
| 416 |
+
start_group_id = None
|
| 417 |
+
end_group_id = None
|
| 418 |
+
for group_id in range(len(self.flat_groups_offset)):
|
| 419 |
+
if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
|
| 420 |
+
start_group_id = group_id
|
| 421 |
+
if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
|
| 422 |
+
end_group_id = group_id
|
| 423 |
+
break
|
| 424 |
+
# collect weights from related group/groups
|
| 425 |
+
for group_id in range(start_group_id, end_group_id + 1):
|
| 426 |
+
flat_tensor = flat_groups_at_rank_i[group_id]
|
| 427 |
+
start_offset = self.offset - self.flat_groups_offset[group_id]
|
| 428 |
+
end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
|
| 429 |
+
pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
|
| 430 |
+
|
| 431 |
+
# collect weights from all ranks
|
| 432 |
+
pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
|
| 433 |
+
param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
|
| 434 |
+
return param
|
| 435 |
+
|
| 436 |
+
|
| 437 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 438 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 439 |
+
avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
|
| 440 |
+
|
| 441 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 442 |
+
# param, re-consolidating each param, while dealing with padding if any
|
| 443 |
+
|
| 444 |
+
# merge list of dicts, preserving order
|
| 445 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 446 |
+
|
| 447 |
+
if debug:
|
| 448 |
+
for i in range(world_size):
|
| 449 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 450 |
+
|
| 451 |
+
wanted_params = len(param_shapes)
|
| 452 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 453 |
+
# not asserting if there is a mismatch due to possible padding
|
| 454 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 455 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 456 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 457 |
+
|
| 458 |
+
# params
|
| 459 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 460 |
+
# out-of-core computing solution
|
| 461 |
+
offset = 0
|
| 462 |
+
total_numel = 0
|
| 463 |
+
total_params = 0
|
| 464 |
+
flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
|
| 465 |
+
for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
|
| 466 |
+
unpartitioned_numel = shape.numel()
|
| 467 |
+
total_numel += unpartitioned_numel
|
| 468 |
+
total_params += 1
|
| 469 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 470 |
+
|
| 471 |
+
if debug:
|
| 472 |
+
print(
|
| 473 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 474 |
+
)
|
| 475 |
+
|
| 476 |
+
# memory efficient tensor
|
| 477 |
+
tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
|
| 478 |
+
state_dict[name] = tensor
|
| 479 |
+
offset += partitioned_numel
|
| 480 |
+
|
| 481 |
+
offset *= world_size
|
| 482 |
+
|
| 483 |
+
# Sanity check
|
| 484 |
+
if offset != avail_numel:
|
| 485 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 486 |
+
|
| 487 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 488 |
+
|
| 489 |
+
|
| 490 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 491 |
+
exclude_frozen_parameters):
|
| 492 |
+
state_dict = OrderedDict()
|
| 493 |
+
|
| 494 |
+
# buffers
|
| 495 |
+
buffers = zero_model_states[0].buffers
|
| 496 |
+
state_dict.update(buffers)
|
| 497 |
+
if debug:
|
| 498 |
+
print(f"added {len(buffers)} buffers")
|
| 499 |
+
|
| 500 |
+
if not exclude_frozen_parameters:
|
| 501 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 502 |
+
|
| 503 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 504 |
+
|
| 505 |
+
# recover shared parameters
|
| 506 |
+
for pair in zero_model_states[0].shared_params:
|
| 507 |
+
if pair[1] in state_dict:
|
| 508 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 509 |
+
|
| 510 |
+
return state_dict
|
| 511 |
+
|
| 512 |
+
|
| 513 |
+
def to_torch_tensor(state_dict, return_empty_tensor=False):
|
| 514 |
+
"""
|
| 515 |
+
Convert state_dict of GatheredTensor to torch tensor
|
| 516 |
+
"""
|
| 517 |
+
torch_state_dict = {}
|
| 518 |
+
converted_tensors = {}
|
| 519 |
+
for name, tensor in state_dict.items():
|
| 520 |
+
tensor_id = id(tensor)
|
| 521 |
+
if tensor_id in converted_tensors: # shared tensors
|
| 522 |
+
shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
|
| 523 |
+
torch_state_dict[name] = shared_tensor
|
| 524 |
+
else:
|
| 525 |
+
converted_tensors[tensor_id] = name
|
| 526 |
+
if return_empty_tensor:
|
| 527 |
+
torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
|
| 528 |
+
else:
|
| 529 |
+
torch_state_dict[name] = tensor.contiguous()
|
| 530 |
+
return torch_state_dict
|
| 531 |
+
|
| 532 |
+
|
| 533 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
|
| 534 |
+
tag=None,
|
| 535 |
+
exclude_frozen_parameters=False,
|
| 536 |
+
lazy_mode=False):
|
| 537 |
+
"""
|
| 538 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 539 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 540 |
+
via a model hub.
|
| 541 |
+
|
| 542 |
+
Args:
|
| 543 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 544 |
+
- ``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``
|
| 545 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 546 |
+
- ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
|
| 547 |
+
Convert the pesduo tensor to torch tensor by ``.contiguous()``
|
| 548 |
+
|
| 549 |
+
Returns:
|
| 550 |
+
- pytorch ``state_dict``
|
| 551 |
+
|
| 552 |
+
A typical usage might be ::
|
| 553 |
+
|
| 554 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 555 |
+
# do the training and checkpoint saving
|
| 556 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 557 |
+
model = model.cpu() # move to cpu
|
| 558 |
+
model.load_state_dict(state_dict)
|
| 559 |
+
# submit to model hub or save the model to share with others
|
| 560 |
+
|
| 561 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 562 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 563 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 564 |
+
|
| 565 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 566 |
+
|
| 567 |
+
Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
|
| 568 |
+
You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 569 |
+
the checkpoint. Or you can load state_dict in lazy mode ::
|
| 570 |
+
|
| 571 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 572 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
|
| 573 |
+
for name, lazy_tensor in state_dict.item():
|
| 574 |
+
tensor = lazy_tensor.contiguous() # to cpu
|
| 575 |
+
print(name, tensor)
|
| 576 |
+
# del tensor to release memory if it no longer in use
|
| 577 |
+
"""
|
| 578 |
+
if tag is None:
|
| 579 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 580 |
+
if os.path.isfile(latest_path):
|
| 581 |
+
with open(latest_path, 'r') as fd:
|
| 582 |
+
tag = fd.read().strip()
|
| 583 |
+
else:
|
| 584 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 585 |
+
|
| 586 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 587 |
+
|
| 588 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
| 589 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 590 |
+
|
| 591 |
+
state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
|
| 592 |
+
if lazy_mode:
|
| 593 |
+
return state_dict
|
| 594 |
+
else:
|
| 595 |
+
return to_torch_tensor(state_dict)
|
| 596 |
+
|
| 597 |
+
|
| 598 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
|
| 599 |
+
output_dir,
|
| 600 |
+
max_shard_size="5GB",
|
| 601 |
+
safe_serialization=False,
|
| 602 |
+
tag=None,
|
| 603 |
+
exclude_frozen_parameters=False):
|
| 604 |
+
"""
|
| 605 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 606 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 607 |
+
|
| 608 |
+
Args:
|
| 609 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 610 |
+
- ``output_dir``: directory to the pytorch fp32 state_dict output files
|
| 611 |
+
- ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
|
| 612 |
+
- ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
|
| 613 |
+
- ``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``
|
| 614 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 615 |
+
"""
|
| 616 |
+
|
| 617 |
+
# Dependency pre-check
|
| 618 |
+
if safe_serialization:
|
| 619 |
+
try:
|
| 620 |
+
from safetensors.torch import save_file
|
| 621 |
+
except ImportError:
|
| 622 |
+
print('If you want to use `safe_serialization`, please `pip install safetensors`')
|
| 623 |
+
raise
|
| 624 |
+
if max_shard_size is not None:
|
| 625 |
+
try:
|
| 626 |
+
from huggingface_hub import split_torch_state_dict_into_shards
|
| 627 |
+
except ImportError:
|
| 628 |
+
print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
|
| 629 |
+
raise
|
| 630 |
+
|
| 631 |
+
# Convert zero checkpoint to state_dict
|
| 632 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
|
| 633 |
+
tag,
|
| 634 |
+
exclude_frozen_parameters,
|
| 635 |
+
lazy_mode=True)
|
| 636 |
+
|
| 637 |
+
# Shard the model if it is too big.
|
| 638 |
+
weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
|
| 639 |
+
if max_shard_size is not None:
|
| 640 |
+
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
|
| 641 |
+
# an memory-efficient approach for sharding
|
| 642 |
+
empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
|
| 643 |
+
state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
|
| 644 |
+
filename_pattern=filename_pattern,
|
| 645 |
+
max_shard_size=max_shard_size)
|
| 646 |
+
else:
|
| 647 |
+
from collections import namedtuple
|
| 648 |
+
StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
|
| 649 |
+
state_dict_split = StateDictSplit(is_sharded=False,
|
| 650 |
+
filename_to_tensors={weights_name: list(state_dict.keys())})
|
| 651 |
+
|
| 652 |
+
# Save the model by shard
|
| 653 |
+
os.makedirs(output_dir, exist_ok=True)
|
| 654 |
+
filename_to_tensors = state_dict_split.filename_to_tensors.items()
|
| 655 |
+
for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
|
| 656 |
+
shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
|
| 657 |
+
shard_state_dict = to_torch_tensor(shard_state_dict)
|
| 658 |
+
output_path = os.path.join(output_dir, shard_file)
|
| 659 |
+
if safe_serialization:
|
| 660 |
+
save_file(shard_state_dict, output_path, metadata={"format": "pt"})
|
| 661 |
+
else:
|
| 662 |
+
torch.save(shard_state_dict, output_path)
|
| 663 |
+
# release the memory of current shard
|
| 664 |
+
for tensor_name in list(shard_state_dict.keys()):
|
| 665 |
+
del state_dict[tensor_name]
|
| 666 |
+
del shard_state_dict[tensor_name]
|
| 667 |
+
del shard_state_dict
|
| 668 |
+
gc.collect()
|
| 669 |
+
|
| 670 |
+
# Save index if sharded
|
| 671 |
+
if state_dict_split.is_sharded:
|
| 672 |
+
index = {
|
| 673 |
+
"metadata": state_dict_split.metadata,
|
| 674 |
+
"weight_map": state_dict_split.tensor_to_filename,
|
| 675 |
+
}
|
| 676 |
+
save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
|
| 677 |
+
save_index_file = os.path.join(output_dir, save_index_file)
|
| 678 |
+
with open(save_index_file, "w", encoding="utf-8") as f:
|
| 679 |
+
content = json.dumps(index, indent=2, sort_keys=True) + "\n"
|
| 680 |
+
f.write(content)
|
| 681 |
+
|
| 682 |
+
|
| 683 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 684 |
+
"""
|
| 685 |
+
1. Put the provided model to cpu
|
| 686 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 687 |
+
3. Load it into the provided model
|
| 688 |
+
|
| 689 |
+
Args:
|
| 690 |
+
- ``model``: the model object to update
|
| 691 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 692 |
+
- ``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``
|
| 693 |
+
|
| 694 |
+
Returns:
|
| 695 |
+
- ``model`: modified model
|
| 696 |
+
|
| 697 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 698 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 699 |
+
conveniently placed for you in the checkpoint folder.
|
| 700 |
+
|
| 701 |
+
A typical usage might be ::
|
| 702 |
+
|
| 703 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 704 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 705 |
+
# submit to model hub or save the model to share with others
|
| 706 |
+
|
| 707 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 708 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 709 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 710 |
+
|
| 711 |
+
"""
|
| 712 |
+
logger.info(f"Extracting fp32 weights")
|
| 713 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 714 |
+
|
| 715 |
+
logger.info(f"Overwriting model with fp32 weights")
|
| 716 |
+
model = model.cpu()
|
| 717 |
+
model.load_state_dict(state_dict, strict=False)
|
| 718 |
+
|
| 719 |
+
return model
|
| 720 |
+
|
| 721 |
+
|
| 722 |
+
if __name__ == "__main__":
|
| 723 |
+
parser = argparse.ArgumentParser()
|
| 724 |
+
parser.add_argument("checkpoint_dir",
|
| 725 |
+
type=str,
|
| 726 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 727 |
+
parser.add_argument("output_dir",
|
| 728 |
+
type=str,
|
| 729 |
+
help="directory to the pytorch fp32 state_dict output files"
|
| 730 |
+
"(e.g. path/checkpoint-12-output/)")
|
| 731 |
+
parser.add_argument(
|
| 732 |
+
"--max_shard_size",
|
| 733 |
+
type=str,
|
| 734 |
+
default="5GB",
|
| 735 |
+
help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
|
| 736 |
+
"lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
|
| 737 |
+
"We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
|
| 738 |
+
"without CPU OOM issues.")
|
| 739 |
+
parser.add_argument(
|
| 740 |
+
"--safe_serialization",
|
| 741 |
+
default=False,
|
| 742 |
+
action='store_true',
|
| 743 |
+
help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
|
| 744 |
+
parser.add_argument("-t",
|
| 745 |
+
"--tag",
|
| 746 |
+
type=str,
|
| 747 |
+
default=None,
|
| 748 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 749 |
+
parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
|
| 750 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 751 |
+
args = parser.parse_args()
|
| 752 |
+
|
| 753 |
+
debug = args.debug
|
| 754 |
+
|
| 755 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
|
| 756 |
+
args.output_dir,
|
| 757 |
+
max_shard_size=args.max_shard_size,
|
| 758 |
+
safe_serialization=args.safe_serialization,
|
| 759 |
+
tag=args.tag,
|
| 760 |
+
exclude_frozen_parameters=args.exclude_frozen_parameters)
|