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).*\\.(q_proj|down_proj|k_proj|v_proj|up_proj|gate_proj|o_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 sha256:4b295bd3ba86e0fbfac5a3dc6d844083ac05a6cd607d03afe149efc475b00d78
|
| 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": "###Task You are an expert at converting visual questions into effective search queries. \nYour goal is to create a comprehensive search query that will help find the most relevant information. \nFor each image-based question, you must create a search query that combines: \n1. Key visual elements from the image (objects, text, logos, scenes, actions, etc.) \n2. The core question being asked \n3. Potential answer terms or relevant context \nFor example: \n- If asking about a logo: include company name, industry, and visual description \n- If asking about an object: include its appearance, category, and possible brands/models \n- If asking about an event/scene: include location hints, activities, and time period clues",
|
| 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 |
+
"task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.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/task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.json/v0-20250614-165413",
|
| 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": 1.0,
|
| 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/task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.json/v0-20250614-165413/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 0x749104b4ac20>, 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/task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.json/v0-20250614-165413', 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=1.0, 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/task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.json/v0-20250614-165413/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/task3_dpo_gt6_task2_gt30.3+task3gt5_0.1vp_idk0614.json/v0-20250614-165413', 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:c94be3dde8c0aa1eb49095eb9d4c3e374518355fb77322b22676eb4fa32b4e69
|
| 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": 0.12102569274602254,
|
| 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": 0.0008068379516401503,
|
| 14 |
+
"grad_norm": 9.261910438537598,
|
| 15 |
+
"learning_rate": 3.225806451612903e-06,
|
| 16 |
+
"logits/chosen": -0.6039718985557556,
|
| 17 |
+
"logits/rejected": -0.6058337688446045,
|
| 18 |
+
"logps/chosen": -7.412725925445557,
|
| 19 |
+
"logps/rejected": -11.413676261901855,
|
| 20 |
+
"loss": 1.2381761074066162,
|
| 21 |
+
"memory(GiB)": 46.32,
|
| 22 |
+
"nll_loss": 0.5450288653373718,
|
| 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)": 0.005524
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.0016136759032803006,
|
| 32 |
+
"grad_norm": 5.391567230224609,
|
| 33 |
+
"learning_rate": 6.451612903225806e-06,
|
| 34 |
+
"logits/chosen": -0.6199994087219238,
|
| 35 |
+
"logits/rejected": -0.6217825412750244,
|
| 36 |
+
"logps/chosen": -7.599736213684082,
|
| 37 |
+
"logps/rejected": -15.575040817260742,
|
| 38 |
+
"loss": 1.1066936254501343,
|
| 39 |
+
"memory(GiB)": 46.83,
|
| 40 |
+
"nll_loss": 0.41354650259017944,
|
| 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)": 0.005546
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.002420513854920451,
|
| 50 |
+
"grad_norm": 6.5534772872924805,
|
| 51 |
+
"learning_rate": 9.67741935483871e-06,
|
| 52 |
+
"logits/chosen": -0.6068329811096191,
|
| 53 |
+
"logits/rejected": -0.6103144288063049,
|
| 54 |
+
"logps/chosen": -11.938759803771973,
|
| 55 |
+
"logps/rejected": -14.32825756072998,
|
| 56 |
+
"loss": 1.267569661140442,
|
| 57 |
+
"memory(GiB)": 46.83,
|
| 58 |
+
"nll_loss": 0.5742746591567993,
|
| 59 |
+
"rewards/accuracies": 0.5,
|
| 60 |
+
"rewards/chosen": 0.0005358309717848897,
|
| 61 |
+
"rewards/margins": -0.00011114418157376349,
|
| 62 |
+
"rewards/rejected": 0.0006469750078395009,
|
| 63 |
+
"step": 3,
|
| 64 |
+
"train_speed(iter/s)": 0.005487
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.003227351806560601,
|
| 68 |
+
"grad_norm": 3.4633655548095703,
|
| 69 |
+
"learning_rate": 1.2903225806451613e-05,
|
| 70 |
+
"logits/chosen": -0.6417700052261353,
|
| 71 |
+
"logits/rejected": -0.638687789440155,
|
| 72 |
+
"logps/chosen": -10.244372367858887,
|
| 73 |
+
"logps/rejected": -8.448144912719727,
|
| 74 |
+
"loss": 1.1101956367492676,
|
| 75 |
+
"memory(GiB)": 46.83,
|
| 76 |
+
"nll_loss": 0.4145403802394867,
|
| 77 |
+
"rewards/accuracies": 0.40625,
|
| 78 |
+
"rewards/chosen": -0.002526381518691778,
|
| 79 |
+
"rewards/margins": -0.004837517160922289,
|
| 80 |
+
"rewards/rejected": 0.0023111351765692234,
|
| 81 |
+
"step": 4,
|
| 82 |
+
"train_speed(iter/s)": 0.005482
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.004034189758200751,
|
| 86 |
+
"grad_norm": 9.884061813354492,
|
| 87 |
+
"learning_rate": 1.6129032258064517e-05,
|
| 88 |
+
"logits/chosen": -0.6263395547866821,
|
| 89 |
+
"logits/rejected": -0.6316207647323608,
|
| 90 |
+
"logps/chosen": -8.768255233764648,
|
| 91 |
+
"logps/rejected": -12.074126243591309,
|
| 92 |
+
"loss": 1.2384487390518188,
|
| 93 |
+
"memory(GiB)": 46.83,
|
| 94 |
+
"nll_loss": 0.5502734780311584,
|
| 95 |
+
"rewards/accuracies": 0.53125,
|
| 96 |
+
"rewards/chosen": 0.011151458136737347,
|
| 97 |
+
"rewards/margins": 0.010235130786895752,
|
| 98 |
+
"rewards/rejected": 0.0009163276990875602,
|
| 99 |
+
"step": 5,
|
| 100 |
+
"train_speed(iter/s)": 0.005496
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.004841027709840902,
|
| 104 |
+
"grad_norm": 9.879764556884766,
|
| 105 |
+
"learning_rate": 1.935483870967742e-05,
|
| 106 |
+
"logits/chosen": -0.6110924482345581,
|
| 107 |
+
"logits/rejected": -0.6128084063529968,
|
| 108 |
+
"logps/chosen": -11.146334648132324,
|
| 109 |
+
"logps/rejected": -12.724503517150879,
|
| 110 |
+
"loss": 1.246738314628601,
|
| 111 |
+
"memory(GiB)": 46.83,
|
| 112 |
+
"nll_loss": 0.5631918907165527,
|
| 113 |
+
"rewards/accuracies": 0.65625,
|
| 114 |
+
"rewards/chosen": 0.02471097745001316,
|
| 115 |
+
"rewards/margins": 0.02056196704506874,
|
| 116 |
+
"rewards/rejected": 0.004149014595896006,
|
| 117 |
+
"step": 6,
|
| 118 |
+
"train_speed(iter/s)": 0.005505
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.0056478656614810515,
|
| 122 |
+
"grad_norm": 3.7321648597717285,
|
| 123 |
+
"learning_rate": 2.258064516129032e-05,
|
| 124 |
+
"logits/chosen": -0.6434396505355835,
|
| 125 |
+
"logits/rejected": -0.6456248164176941,
|
| 126 |
+
"logps/chosen": -12.346940994262695,
|
| 127 |
+
"logps/rejected": -13.585854530334473,
|
| 128 |
+
"loss": 1.25950026512146,
|
| 129 |
+
"memory(GiB)": 46.83,
|
| 130 |
+
"nll_loss": 0.5602214932441711,
|
| 131 |
+
"rewards/accuracies": 0.46875,
|
| 132 |
+
"rewards/chosen": 0.01709746941924095,
|
| 133 |
+
"rewards/margins": -0.004833616316318512,
|
| 134 |
+
"rewards/rejected": 0.021931089460849762,
|
| 135 |
+
"step": 7,
|
| 136 |
+
"train_speed(iter/s)": 0.005511
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.006454703613121202,
|
| 140 |
+
"grad_norm": 6.577890872955322,
|
| 141 |
+
"learning_rate": 2.5806451612903226e-05,
|
| 142 |
+
"logits/chosen": -0.6166916489601135,
|
| 143 |
+
"logits/rejected": -0.6146227717399597,
|
| 144 |
+
"logps/chosen": -16.807830810546875,
|
| 145 |
+
"logps/rejected": -11.38271427154541,
|
| 146 |
+
"loss": 1.2828420400619507,
|
| 147 |
+
"memory(GiB)": 46.83,
|
| 148 |
+
"nll_loss": 0.5101535320281982,
|
| 149 |
+
"rewards/accuracies": 0.25,
|
| 150 |
+
"rewards/chosen": -0.04314221069216728,
|
| 151 |
+
"rewards/margins": -0.13309986889362335,
|
| 152 |
+
"rewards/rejected": 0.08995765447616577,
|
| 153 |
+
"step": 8,
|
| 154 |
+
"train_speed(iter/s)": 0.005502
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.007261541564761352,
|
| 158 |
+
"grad_norm": 3.5513713359832764,
|
| 159 |
+
"learning_rate": 2.9032258064516133e-05,
|
| 160 |
+
"logits/chosen": -0.5994153022766113,
|
| 161 |
+
"logits/rejected": -0.6015004515647888,
|
| 162 |
+
"logps/chosen": -11.679065704345703,
|
| 163 |
+
"logps/rejected": -12.411300659179688,
|
| 164 |
+
"loss": 1.171980857849121,
|
| 165 |
+
"memory(GiB)": 46.83,
|
| 166 |
+
"nll_loss": 0.4982386827468872,
|
| 167 |
+
"rewards/accuracies": 0.5625,
|
| 168 |
+
"rewards/chosen": 0.035934723913669586,
|
| 169 |
+
"rewards/margins": 0.08020534366369247,
|
| 170 |
+
"rewards/rejected": -0.04427062347531319,
|
| 171 |
+
"step": 9,
|
| 172 |
+
"train_speed(iter/s)": 0.005483
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.008068379516401502,
|
| 176 |
+
"grad_norm": 7.633578777313232,
|
| 177 |
+
"learning_rate": 3.2258064516129034e-05,
|
| 178 |
+
"logits/chosen": -0.6122820377349854,
|
| 179 |
+
"logits/rejected": -0.6125556826591492,
|
| 180 |
+
"logps/chosen": -12.382171630859375,
|
| 181 |
+
"logps/rejected": -10.675874710083008,
|
| 182 |
+
"loss": 1.3903008699417114,
|
| 183 |
+
"memory(GiB)": 46.83,
|
| 184 |
+
"nll_loss": 0.6006877422332764,
|
| 185 |
+
"rewards/accuracies": 0.375,
|
| 186 |
+
"rewards/chosen": -0.08313174545764923,
|
| 187 |
+
"rewards/margins": -0.13078533113002777,
|
| 188 |
+
"rewards/rejected": 0.04765357822179794,
|
| 189 |
+
"step": 10,
|
| 190 |
+
"train_speed(iter/s)": 0.005494
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.008875217468041654,
|
| 194 |
+
"grad_norm": 4.818623065948486,
|
| 195 |
+
"learning_rate": 3.548387096774194e-05,
|
| 196 |
+
"logits/chosen": -0.6364253759384155,
|
| 197 |
+
"logits/rejected": -0.6395613551139832,
|
| 198 |
+
"logps/chosen": -7.883171558380127,
|
| 199 |
+
"logps/rejected": -15.457681655883789,
|
| 200 |
+
"loss": 1.1344804763793945,
|
| 201 |
+
"memory(GiB)": 46.83,
|
| 202 |
+
"nll_loss": 0.46768438816070557,
|
| 203 |
+
"rewards/accuracies": 0.53125,
|
| 204 |
+
"rewards/chosen": 0.008294675499200821,
|
| 205 |
+
"rewards/margins": 0.10872506350278854,
|
| 206 |
+
"rewards/rejected": -0.10043041408061981,
|
| 207 |
+
"step": 11,
|
| 208 |
+
"train_speed(iter/s)": 0.00549
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.009682055419681803,
|
| 212 |
+
"grad_norm": 3.8297243118286133,
|
| 213 |
+
"learning_rate": 3.870967741935484e-05,
|
| 214 |
+
"logits/chosen": -0.6082768440246582,
|
| 215 |
+
"logits/rejected": -0.6128782629966736,
|
| 216 |
+
"logps/chosen": -10.993257522583008,
|
| 217 |
+
"logps/rejected": -14.49487018585205,
|
| 218 |
+
"loss": 1.1111013889312744,
|
| 219 |
+
"memory(GiB)": 46.83,
|
| 220 |
+
"nll_loss": 0.42791634798049927,
|
| 221 |
+
"rewards/accuracies": 0.46875,
|
| 222 |
+
"rewards/chosen": -0.013819348067045212,
|
| 223 |
+
"rewards/margins": 0.08211039006710052,
|
| 224 |
+
"rewards/rejected": -0.09592973440885544,
|
| 225 |
+
"step": 12,
|
| 226 |
+
"train_speed(iter/s)": 0.005476
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.010488893371321953,
|
| 230 |
+
"grad_norm": 3.1733834743499756,
|
| 231 |
+
"learning_rate": 4.1935483870967746e-05,
|
| 232 |
+
"logits/chosen": -0.597683846950531,
|
| 233 |
+
"logits/rejected": -0.6003975868225098,
|
| 234 |
+
"logps/chosen": -7.611721038818359,
|
| 235 |
+
"logps/rejected": -13.589990615844727,
|
| 236 |
+
"loss": 0.9966970682144165,
|
| 237 |
+
"memory(GiB)": 46.83,
|
| 238 |
+
"nll_loss": 0.3590569496154785,
|
| 239 |
+
"rewards/accuracies": 0.65625,
|
| 240 |
+
"rewards/chosen": 0.09378139674663544,
|
| 241 |
+
"rewards/margins": 0.16571055352687836,
|
| 242 |
+
"rewards/rejected": -0.07192917168140411,
|
| 243 |
+
"step": 13,
|
| 244 |
+
"train_speed(iter/s)": 0.005488
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.011295731322962103,
|
| 248 |
+
"grad_norm": 5.364157676696777,
|
| 249 |
+
"learning_rate": 4.516129032258064e-05,
|
| 250 |
+
"logits/chosen": -0.5847792029380798,
|
| 251 |
+
"logits/rejected": -0.5856592655181885,
|
| 252 |
+
"logps/chosen": -11.617069244384766,
|
| 253 |
+
"logps/rejected": -11.611367225646973,
|
| 254 |
+
"loss": 1.2788419723510742,
|
| 255 |
+
"memory(GiB)": 46.83,
|
| 256 |
+
"nll_loss": 0.5410674214363098,
|
| 257 |
+
"rewards/accuracies": 0.46875,
|
| 258 |
+
"rewards/chosen": 0.037577953189611435,
|
| 259 |
+
"rewards/margins": -0.04553454369306564,
|
| 260 |
+
"rewards/rejected": 0.08311249315738678,
|
| 261 |
+
"step": 14,
|
| 262 |
+
"train_speed(iter/s)": 0.005494
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.012102569274602255,
|
| 266 |
+
"grad_norm": 3.436879873275757,
|
| 267 |
+
"learning_rate": 4.8387096774193554e-05,
|
| 268 |
+
"logits/chosen": -0.5832479596138,
|
| 269 |
+
"logits/rejected": -0.5832363367080688,
|
| 270 |
+
"logps/chosen": -9.650967597961426,
|
| 271 |
+
"logps/rejected": -11.376352310180664,
|
| 272 |
+
"loss": 1.0634793043136597,
|
| 273 |
+
"memory(GiB)": 46.83,
|
| 274 |
+
"nll_loss": 0.34024038910865784,
|
| 275 |
+
"rewards/accuracies": 0.53125,
|
| 276 |
+
"rewards/chosen": 0.05362201854586601,
|
| 277 |
+
"rewards/margins": -0.029222950339317322,
|
| 278 |
+
"rewards/rejected": 0.08284495770931244,
|
| 279 |
+
"step": 15,
|
| 280 |
+
"train_speed(iter/s)": 0.005497
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.012909407226242404,
|
| 284 |
+
"grad_norm": 3.2199597358703613,
|
| 285 |
+
"learning_rate": 5.161290322580645e-05,
|
| 286 |
+
"logits/chosen": -0.6036816239356995,
|
| 287 |
+
"logits/rejected": -0.606670618057251,
|
| 288 |
+
"logps/chosen": -10.059551239013672,
|
| 289 |
+
"logps/rejected": -16.143247604370117,
|
| 290 |
+
"loss": 1.0622740983963013,
|
| 291 |
+
"memory(GiB)": 46.83,
|
| 292 |
+
"nll_loss": 0.3808459937572479,
|
| 293 |
+
"rewards/accuracies": 0.625,
|
| 294 |
+
"rewards/chosen": 0.11249684542417526,
|
| 295 |
+
"rewards/margins": 0.040876831859350204,
|
| 296 |
+
"rewards/rejected": 0.07162001729011536,
|
| 297 |
+
"step": 16,
|
| 298 |
+
"train_speed(iter/s)": 0.00549
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.013716245177882554,
|
| 302 |
+
"grad_norm": 3.376516819000244,
|
| 303 |
+
"learning_rate": 5.4838709677419355e-05,
|
| 304 |
+
"logits/chosen": -0.5811513662338257,
|
| 305 |
+
"logits/rejected": -0.5822936296463013,
|
| 306 |
+
"logps/chosen": -10.964933395385742,
|
| 307 |
+
"logps/rejected": -13.541594505310059,
|
| 308 |
+
"loss": 1.1284891366958618,
|
| 309 |
+
"memory(GiB)": 46.83,
|
| 310 |
+
"nll_loss": 0.43761852383613586,
|
| 311 |
+
"rewards/accuracies": 0.5,
|
| 312 |
+
"rewards/chosen": 0.11780606210231781,
|
| 313 |
+
"rewards/margins": 0.02046366035938263,
|
| 314 |
+
"rewards/rejected": 0.09734240174293518,
|
| 315 |
+
"step": 17,
|
| 316 |
+
"train_speed(iter/s)": 0.005493
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.014523083129522704,
|
| 320 |
+
"grad_norm": 3.292140245437622,
|
| 321 |
+
"learning_rate": 5.8064516129032266e-05,
|
| 322 |
+
"logits/chosen": -0.5822582840919495,
|
| 323 |
+
"logits/rejected": -0.5867500305175781,
|
| 324 |
+
"logps/chosen": -8.11985969543457,
|
| 325 |
+
"logps/rejected": -14.659163475036621,
|
| 326 |
+
"loss": 1.0471278429031372,
|
| 327 |
+
"memory(GiB)": 46.83,
|
| 328 |
+
"nll_loss": 0.40266352891921997,
|
| 329 |
+
"rewards/accuracies": 0.65625,
|
| 330 |
+
"rewards/chosen": 0.14493001997470856,
|
| 331 |
+
"rewards/margins": 0.1153046265244484,
|
| 332 |
+
"rewards/rejected": 0.029625393450260162,
|
| 333 |
+
"step": 18,
|
| 334 |
+
"train_speed(iter/s)": 0.005497
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.015329921081162856,
|
| 338 |
+
"grad_norm": 3.320802688598633,
|
| 339 |
+
"learning_rate": 6.129032258064517e-05,
|
| 340 |
+
"logits/chosen": -0.5760956406593323,
|
| 341 |
+
"logits/rejected": -0.5776647925376892,
|
| 342 |
+
"logps/chosen": -9.478979110717773,
|
| 343 |
+
"logps/rejected": -11.301219940185547,
|
| 344 |
+
"loss": 1.1243548393249512,
|
| 345 |
+
"memory(GiB)": 46.83,
|
| 346 |
+
"nll_loss": 0.48838871717453003,
|
| 347 |
+
"rewards/accuracies": 0.6875,
|
| 348 |
+
"rewards/chosen": 0.20540697872638702,
|
| 349 |
+
"rewards/margins": 0.14464649558067322,
|
| 350 |
+
"rewards/rejected": 0.06076047942042351,
|
| 351 |
+
"step": 19,
|
| 352 |
+
"train_speed(iter/s)": 0.005493
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.016136759032803004,
|
| 356 |
+
"grad_norm": 2.7910656929016113,
|
| 357 |
+
"learning_rate": 6.451612903225807e-05,
|
| 358 |
+
"logits/chosen": -0.5840404033660889,
|
| 359 |
+
"logits/rejected": -0.582589864730835,
|
| 360 |
+
"logps/chosen": -13.635812759399414,
|
| 361 |
+
"logps/rejected": -10.431557655334473,
|
| 362 |
+
"loss": 1.2548669576644897,
|
| 363 |
+
"memory(GiB)": 46.83,
|
| 364 |
+
"nll_loss": 0.55869460105896,
|
| 365 |
+
"rewards/accuracies": 0.4375,
|
| 366 |
+
"rewards/chosen": 0.13213564455509186,
|
| 367 |
+
"rewards/margins": 0.044300854206085205,
|
| 368 |
+
"rewards/rejected": 0.08783479034900665,
|
| 369 |
+
"step": 20,
|
| 370 |
+
"train_speed(iter/s)": 0.005497
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.016943596984443157,
|
| 374 |
+
"grad_norm": 2.2573604583740234,
|
| 375 |
+
"learning_rate": 6.774193548387096e-05,
|
| 376 |
+
"logits/chosen": -0.5839393734931946,
|
| 377 |
+
"logits/rejected": -0.5850896835327148,
|
| 378 |
+
"logps/chosen": -7.3839006423950195,
|
| 379 |
+
"logps/rejected": -9.574810028076172,
|
| 380 |
+
"loss": 0.9415242075920105,
|
| 381 |
+
"memory(GiB)": 46.83,
|
| 382 |
+
"nll_loss": 0.372197687625885,
|
| 383 |
+
"rewards/accuracies": 0.75,
|
| 384 |
+
"rewards/chosen": 0.2435840666294098,
|
| 385 |
+
"rewards/margins": 0.30735281109809875,
|
| 386 |
+
"rewards/rejected": -0.06376874446868896,
|
| 387 |
+
"step": 21,
|
| 388 |
+
"train_speed(iter/s)": 0.005504
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.017750434936083307,
|
| 392 |
+
"grad_norm": 2.90157413482666,
|
| 393 |
+
"learning_rate": 7.096774193548388e-05,
|
| 394 |
+
"logits/chosen": -0.6114062070846558,
|
| 395 |
+
"logits/rejected": -0.6119735240936279,
|
| 396 |
+
"logps/chosen": -8.486170768737793,
|
| 397 |
+
"logps/rejected": -12.086725234985352,
|
| 398 |
+
"loss": 1.0081067085266113,
|
| 399 |
+
"memory(GiB)": 46.83,
|
| 400 |
+
"nll_loss": 0.34981024265289307,
|
| 401 |
+
"rewards/accuracies": 0.5625,
|
| 402 |
+
"rewards/chosen": 0.10258002579212189,
|
| 403 |
+
"rewards/margins": 0.12462681531906128,
|
| 404 |
+
"rewards/rejected": -0.02204679138958454,
|
| 405 |
+
"step": 22,
|
| 406 |
+
"train_speed(iter/s)": 0.005501
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.018557272887723457,
|
| 410 |
+
"grad_norm": 2.4457335472106934,
|
| 411 |
+
"learning_rate": 7.419354838709677e-05,
|
| 412 |
+
"logits/chosen": -0.5648061037063599,
|
| 413 |
+
"logits/rejected": -0.5674125552177429,
|
| 414 |
+
"logps/chosen": -12.962445259094238,
|
| 415 |
+
"logps/rejected": -15.476969718933105,
|
| 416 |
+
"loss": 1.0199884176254272,
|
| 417 |
+
"memory(GiB)": 46.83,
|
| 418 |
+
"nll_loss": 0.4421077072620392,
|
| 419 |
+
"rewards/accuracies": 0.6875,
|
| 420 |
+
"rewards/chosen": 0.13948869705200195,
|
| 421 |
+
"rewards/margins": 0.31703731417655945,
|
| 422 |
+
"rewards/rejected": -0.17754864692687988,
|
| 423 |
+
"step": 23,
|
| 424 |
+
"train_speed(iter/s)": 0.005504
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.019364110839363607,
|
| 428 |
+
"grad_norm": 2.250957489013672,
|
| 429 |
+
"learning_rate": 7.741935483870968e-05,
|
| 430 |
+
"logits/chosen": -0.5476934313774109,
|
| 431 |
+
"logits/rejected": -0.5497113466262817,
|
| 432 |
+
"logps/chosen": -6.839375972747803,
|
| 433 |
+
"logps/rejected": -12.336576461791992,
|
| 434 |
+
"loss": 0.9127413630485535,
|
| 435 |
+
"memory(GiB)": 46.83,
|
| 436 |
+
"nll_loss": 0.3841688632965088,
|
| 437 |
+
"rewards/accuracies": 0.78125,
|
| 438 |
+
"rewards/chosen": 0.2666690945625305,
|
| 439 |
+
"rewards/margins": 0.42246347665786743,
|
| 440 |
+
"rewards/rejected": -0.1557943969964981,
|
| 441 |
+
"step": 24,
|
| 442 |
+
"train_speed(iter/s)": 0.005505
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.020170948791003757,
|
| 446 |
+
"grad_norm": 3.6742801666259766,
|
| 447 |
+
"learning_rate": 8.064516129032258e-05,
|
| 448 |
+
"logits/chosen": -0.5886849164962769,
|
| 449 |
+
"logits/rejected": -0.5897050499916077,
|
| 450 |
+
"logps/chosen": -8.534127235412598,
|
| 451 |
+
"logps/rejected": -11.815479278564453,
|
| 452 |
+
"loss": 1.0272297859191895,
|
| 453 |
+
"memory(GiB)": 46.83,
|
| 454 |
+
"nll_loss": 0.41876962780952454,
|
| 455 |
+
"rewards/accuracies": 0.625,
|
| 456 |
+
"rewards/chosen": 0.13335689902305603,
|
| 457 |
+
"rewards/margins": 0.27482476830482483,
|
| 458 |
+
"rewards/rejected": -0.1414678543806076,
|
| 459 |
+
"step": 25,
|
| 460 |
+
"train_speed(iter/s)": 0.005513
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.020977786742643906,
|
| 464 |
+
"grad_norm": 2.5902981758117676,
|
| 465 |
+
"learning_rate": 8.387096774193549e-05,
|
| 466 |
+
"logits/chosen": -0.5303969979286194,
|
| 467 |
+
"logits/rejected": -0.5350589752197266,
|
| 468 |
+
"logps/chosen": -9.592369079589844,
|
| 469 |
+
"logps/rejected": -17.837379455566406,
|
| 470 |
+
"loss": 0.9073010087013245,
|
| 471 |
+
"memory(GiB)": 46.83,
|
| 472 |
+
"nll_loss": 0.3275393545627594,
|
| 473 |
+
"rewards/accuracies": 0.75,
|
| 474 |
+
"rewards/chosen": 0.1778128743171692,
|
| 475 |
+
"rewards/margins": 0.31808584928512573,
|
| 476 |
+
"rewards/rejected": -0.14027300477027893,
|
| 477 |
+
"step": 26,
|
| 478 |
+
"train_speed(iter/s)": 0.005509
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.021784624694284056,
|
| 482 |
+
"grad_norm": 3.656003952026367,
|
| 483 |
+
"learning_rate": 8.709677419354839e-05,
|
| 484 |
+
"logits/chosen": -0.5741274356842041,
|
| 485 |
+
"logits/rejected": -0.5759863257408142,
|
| 486 |
+
"logps/chosen": -7.401767253875732,
|
| 487 |
+
"logps/rejected": -9.595519065856934,
|
| 488 |
+
"loss": 0.9837043881416321,
|
| 489 |
+
"memory(GiB)": 46.83,
|
| 490 |
+
"nll_loss": 0.34733644127845764,
|
| 491 |
+
"rewards/accuracies": 0.6875,
|
| 492 |
+
"rewards/chosen": 0.16637372970581055,
|
| 493 |
+
"rewards/margins": 0.22424587607383728,
|
| 494 |
+
"rewards/rejected": -0.05787213519215584,
|
| 495 |
+
"step": 27,
|
| 496 |
+
"train_speed(iter/s)": 0.00551
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.022591462645924206,
|
| 500 |
+
"grad_norm": 3.6425983905792236,
|
| 501 |
+
"learning_rate": 9.032258064516129e-05,
|
| 502 |
+
"logits/chosen": -0.5271302461624146,
|
| 503 |
+
"logits/rejected": -0.5281438231468201,
|
| 504 |
+
"logps/chosen": -10.050779342651367,
|
| 505 |
+
"logps/rejected": -13.247575759887695,
|
| 506 |
+
"loss": 1.0190500020980835,
|
| 507 |
+
"memory(GiB)": 46.83,
|
| 508 |
+
"nll_loss": 0.4030058681964874,
|
| 509 |
+
"rewards/accuracies": 0.625,
|
| 510 |
+
"rewards/chosen": 0.022431861609220505,
|
| 511 |
+
"rewards/margins": 0.22834232449531555,
|
| 512 |
+
"rewards/rejected": -0.20591047406196594,
|
| 513 |
+
"step": 28,
|
| 514 |
+
"train_speed(iter/s)": 0.005522
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.02339830059756436,
|
| 518 |
+
"grad_norm": 4.539236545562744,
|
| 519 |
+
"learning_rate": 9.35483870967742e-05,
|
| 520 |
+
"logits/chosen": -0.574794352054596,
|
| 521 |
+
"logits/rejected": -0.5771020650863647,
|
| 522 |
+
"logps/chosen": -7.9426727294921875,
|
| 523 |
+
"logps/rejected": -11.816813468933105,
|
| 524 |
+
"loss": 1.0435985326766968,
|
| 525 |
+
"memory(GiB)": 46.83,
|
| 526 |
+
"nll_loss": 0.4821586310863495,
|
| 527 |
+
"rewards/accuracies": 0.6875,
|
| 528 |
+
"rewards/chosen": 0.17023718357086182,
|
| 529 |
+
"rewards/margins": 0.36495453119277954,
|
| 530 |
+
"rewards/rejected": -0.19471733272075653,
|
| 531 |
+
"step": 29,
|
| 532 |
+
"train_speed(iter/s)": 0.005525
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.02420513854920451,
|
| 536 |
+
"grad_norm": 3.463172674179077,
|
| 537 |
+
"learning_rate": 9.677419354838711e-05,
|
| 538 |
+
"logits/chosen": -0.5663347244262695,
|
| 539 |
+
"logits/rejected": -0.5671164393424988,
|
| 540 |
+
"logps/chosen": -8.761878967285156,
|
| 541 |
+
"logps/rejected": -16.922607421875,
|
| 542 |
+
"loss": 0.8841907978057861,
|
| 543 |
+
"memory(GiB)": 46.83,
|
| 544 |
+
"nll_loss": 0.37557756900787354,
|
| 545 |
+
"rewards/accuracies": 0.8125,
|
| 546 |
+
"rewards/chosen": 0.23288068175315857,
|
| 547 |
+
"rewards/margins": 0.47043219208717346,
|
| 548 |
+
"rewards/rejected": -0.2375514954328537,
|
| 549 |
+
"step": 30,
|
| 550 |
+
"train_speed(iter/s)": 0.005522
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.02501197650084466,
|
| 554 |
+
"grad_norm": 7.054537296295166,
|
| 555 |
+
"learning_rate": 0.0001,
|
| 556 |
+
"logits/chosen": -0.5485356450080872,
|
| 557 |
+
"logits/rejected": -0.5510388612747192,
|
| 558 |
+
"logps/chosen": -8.1714506149292,
|
| 559 |
+
"logps/rejected": -15.964082717895508,
|
| 560 |
+
"loss": 1.0905145406723022,
|
| 561 |
+
"memory(GiB)": 46.83,
|
| 562 |
+
"nll_loss": 0.48109474778175354,
|
| 563 |
+
"rewards/accuracies": 0.65625,
|
| 564 |
+
"rewards/chosen": 0.14875195920467377,
|
| 565 |
+
"rewards/margins": 0.23288127779960632,
|
| 566 |
+
"rewards/rejected": -0.08412933349609375,
|
| 567 |
+
"step": 31,
|
| 568 |
+
"train_speed(iter/s)": 0.005519
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.02581881445248481,
|
| 572 |
+
"grad_norm": 2.8915627002716064,
|
| 573 |
+
"learning_rate": 0.0001032258064516129,
|
| 574 |
+
"logits/chosen": -0.5415224432945251,
|
| 575 |
+
"logits/rejected": -0.5406837463378906,
|
| 576 |
+
"logps/chosen": -12.312981605529785,
|
| 577 |
+
"logps/rejected": -14.444657325744629,
|
| 578 |
+
"loss": 0.9778124094009399,
|
| 579 |
+
"memory(GiB)": 46.83,
|
| 580 |
+
"nll_loss": 0.4027611017227173,
|
| 581 |
+
"rewards/accuracies": 0.71875,
|
| 582 |
+
"rewards/chosen": 0.11852823942899704,
|
| 583 |
+
"rewards/margins": 0.3213120102882385,
|
| 584 |
+
"rewards/rejected": -0.2027837634086609,
|
| 585 |
+
"step": 32,
|
| 586 |
+
"train_speed(iter/s)": 0.005515
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.02662565240412496,
|
| 590 |
+
"grad_norm": 4.721103191375732,
|
| 591 |
+
"learning_rate": 0.0001064516129032258,
|
| 592 |
+
"logits/chosen": -0.5442328453063965,
|
| 593 |
+
"logits/rejected": -0.5437616109848022,
|
| 594 |
+
"logps/chosen": -9.491182327270508,
|
| 595 |
+
"logps/rejected": -18.31528663635254,
|
| 596 |
+
"loss": 1.1582751274108887,
|
| 597 |
+
"memory(GiB)": 46.83,
|
| 598 |
+
"nll_loss": 0.5736849904060364,
|
| 599 |
+
"rewards/accuracies": 0.75,
|
| 600 |
+
"rewards/chosen": 0.03584306687116623,
|
| 601 |
+
"rewards/margins": 0.35881760716438293,
|
| 602 |
+
"rewards/rejected": -0.3229745626449585,
|
| 603 |
+
"step": 33,
|
| 604 |
+
"train_speed(iter/s)": 0.005515
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.02743249035576511,
|
| 608 |
+
"grad_norm": 4.470694065093994,
|
| 609 |
+
"learning_rate": 0.00010967741935483871,
|
| 610 |
+
"logits/chosen": -0.49901843070983887,
|
| 611 |
+
"logits/rejected": -0.5027262568473816,
|
| 612 |
+
"logps/chosen": -6.029642105102539,
|
| 613 |
+
"logps/rejected": -17.065149307250977,
|
| 614 |
+
"loss": 0.7831717133522034,
|
| 615 |
+
"memory(GiB)": 46.83,
|
| 616 |
+
"nll_loss": 0.2735077142715454,
|
| 617 |
+
"rewards/accuracies": 0.6875,
|
| 618 |
+
"rewards/chosen": 0.172469824552536,
|
| 619 |
+
"rewards/margins": 0.6490951180458069,
|
| 620 |
+
"rewards/rejected": -0.4766252934932709,
|
| 621 |
+
"step": 34,
|
| 622 |
+
"train_speed(iter/s)": 0.005511
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.02823932830740526,
|
| 626 |
+
"grad_norm": 4.086674690246582,
|
| 627 |
+
"learning_rate": 0.00011290322580645163,
|
| 628 |
+
"logits/chosen": -0.5251173377037048,
|
| 629 |
+
"logits/rejected": -0.5295683145523071,
|
| 630 |
+
"logps/chosen": -10.127384185791016,
|
| 631 |
+
"logps/rejected": -19.261646270751953,
|
| 632 |
+
"loss": 0.90782630443573,
|
| 633 |
+
"memory(GiB)": 46.83,
|
| 634 |
+
"nll_loss": 0.37643295526504517,
|
| 635 |
+
"rewards/accuracies": 0.71875,
|
| 636 |
+
"rewards/chosen": 0.059490665793418884,
|
| 637 |
+
"rewards/margins": 0.5788955688476562,
|
| 638 |
+
"rewards/rejected": -0.5194048881530762,
|
| 639 |
+
"step": 35,
|
| 640 |
+
"train_speed(iter/s)": 0.005516
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.02904616625904541,
|
| 644 |
+
"grad_norm": 3.153599977493286,
|
| 645 |
+
"learning_rate": 0.00011612903225806453,
|
| 646 |
+
"logits/chosen": -0.5580551624298096,
|
| 647 |
+
"logits/rejected": -0.561299741268158,
|
| 648 |
+
"logps/chosen": -7.748661518096924,
|
| 649 |
+
"logps/rejected": -18.775835037231445,
|
| 650 |
+
"loss": 0.8323644995689392,
|
| 651 |
+
"memory(GiB)": 46.83,
|
| 652 |
+
"nll_loss": 0.30874526500701904,
|
| 653 |
+
"rewards/accuracies": 0.75,
|
| 654 |
+
"rewards/chosen": 0.17774799466133118,
|
| 655 |
+
"rewards/margins": 0.58444744348526,
|
| 656 |
+
"rewards/rejected": -0.4066994786262512,
|
| 657 |
+
"step": 36,
|
| 658 |
+
"train_speed(iter/s)": 0.005518
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.02985300421068556,
|
| 662 |
+
"grad_norm": 5.963130950927734,
|
| 663 |
+
"learning_rate": 0.00011935483870967743,
|
| 664 |
+
"logits/chosen": -0.5691896080970764,
|
| 665 |
+
"logits/rejected": -0.5704917907714844,
|
| 666 |
+
"logps/chosen": -5.695924282073975,
|
| 667 |
+
"logps/rejected": -15.709531784057617,
|
| 668 |
+
"loss": 0.8581907153129578,
|
| 669 |
+
"memory(GiB)": 46.83,
|
| 670 |
+
"nll_loss": 0.32061102986335754,
|
| 671 |
+
"rewards/accuracies": 0.75,
|
| 672 |
+
"rewards/chosen": 0.13813850283622742,
|
| 673 |
+
"rewards/margins": 0.42923206090927124,
|
| 674 |
+
"rewards/rejected": -0.2910935878753662,
|
| 675 |
+
"step": 37,
|
| 676 |
+
"train_speed(iter/s)": 0.005516
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.03065984216232571,
|
| 680 |
+
"grad_norm": 3.532068967819214,
|
| 681 |
+
"learning_rate": 0.00012258064516129034,
|
| 682 |
+
"logits/chosen": -0.5251184105873108,
|
| 683 |
+
"logits/rejected": -0.5261702537536621,
|
| 684 |
+
"logps/chosen": -8.303949356079102,
|
| 685 |
+
"logps/rejected": -13.649654388427734,
|
| 686 |
+
"loss": 0.9825394153594971,
|
| 687 |
+
"memory(GiB)": 46.83,
|
| 688 |
+
"nll_loss": 0.39365747570991516,
|
| 689 |
+
"rewards/accuracies": 0.65625,
|
| 690 |
+
"rewards/chosen": 0.07552488148212433,
|
| 691 |
+
"rewards/margins": 0.2995148003101349,
|
| 692 |
+
"rewards/rejected": -0.22398996353149414,
|
| 693 |
+
"step": 38,
|
| 694 |
+
"train_speed(iter/s)": 0.005515
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.03146668011396586,
|
| 698 |
+
"grad_norm": 2.9115569591522217,
|
| 699 |
+
"learning_rate": 0.00012580645161290322,
|
| 700 |
+
"logits/chosen": -0.5716837048530579,
|
| 701 |
+
"logits/rejected": -0.5749698281288147,
|
| 702 |
+
"logps/chosen": -7.206816673278809,
|
| 703 |
+
"logps/rejected": -15.77934741973877,
|
| 704 |
+
"loss": 0.7430101633071899,
|
| 705 |
+
"memory(GiB)": 46.83,
|
| 706 |
+
"nll_loss": 0.27266308665275574,
|
| 707 |
+
"rewards/accuracies": 0.90625,
|
| 708 |
+
"rewards/chosen": 0.20229285955429077,
|
| 709 |
+
"rewards/margins": 0.6254266500473022,
|
| 710 |
+
"rewards/rejected": -0.42313385009765625,
|
| 711 |
+
"step": 39,
|
| 712 |
+
"train_speed(iter/s)": 0.005515
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.03227351806560601,
|
| 716 |
+
"grad_norm": 4.647506237030029,
|
| 717 |
+
"learning_rate": 0.00012903225806451613,
|
| 718 |
+
"logits/chosen": -0.5296156406402588,
|
| 719 |
+
"logits/rejected": -0.5295798778533936,
|
| 720 |
+
"logps/chosen": -10.529404640197754,
|
| 721 |
+
"logps/rejected": -16.44548988342285,
|
| 722 |
+
"loss": 0.900071918964386,
|
| 723 |
+
"memory(GiB)": 46.83,
|
| 724 |
+
"nll_loss": 0.3727172613143921,
|
| 725 |
+
"rewards/accuracies": 0.71875,
|
| 726 |
+
"rewards/chosen": 0.08702702820301056,
|
| 727 |
+
"rewards/margins": 0.4926542341709137,
|
| 728 |
+
"rewards/rejected": -0.40562722086906433,
|
| 729 |
+
"step": 40,
|
| 730 |
+
"train_speed(iter/s)": 0.005514
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.03308035601724616,
|
| 734 |
+
"grad_norm": 3.4815561771392822,
|
| 735 |
+
"learning_rate": 0.00013225806451612905,
|
| 736 |
+
"logits/chosen": -0.5586342811584473,
|
| 737 |
+
"logits/rejected": -0.5603877305984497,
|
| 738 |
+
"logps/chosen": -7.803877830505371,
|
| 739 |
+
"logps/rejected": -18.412734985351562,
|
| 740 |
+
"loss": 0.8854644298553467,
|
| 741 |
+
"memory(GiB)": 46.83,
|
| 742 |
+
"nll_loss": 0.37826117873191833,
|
| 743 |
+
"rewards/accuracies": 0.78125,
|
| 744 |
+
"rewards/chosen": 0.08443664014339447,
|
| 745 |
+
"rewards/margins": 0.6363339424133301,
|
| 746 |
+
"rewards/rejected": -0.5518972873687744,
|
| 747 |
+
"step": 41,
|
| 748 |
+
"train_speed(iter/s)": 0.005514
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.033887193968886314,
|
| 752 |
+
"grad_norm": 3.6079483032226562,
|
| 753 |
+
"learning_rate": 0.00013548387096774193,
|
| 754 |
+
"logits/chosen": -0.5233692526817322,
|
| 755 |
+
"logits/rejected": -0.5257354378700256,
|
| 756 |
+
"logps/chosen": -11.989986419677734,
|
| 757 |
+
"logps/rejected": -17.48329734802246,
|
| 758 |
+
"loss": 0.9667509198188782,
|
| 759 |
+
"memory(GiB)": 46.83,
|
| 760 |
+
"nll_loss": 0.4528670608997345,
|
| 761 |
+
"rewards/accuracies": 0.8125,
|
| 762 |
+
"rewards/chosen": 0.14738641679286957,
|
| 763 |
+
"rewards/margins": 0.5997685194015503,
|
| 764 |
+
"rewards/rejected": -0.4523821473121643,
|
| 765 |
+
"step": 42,
|
| 766 |
+
"train_speed(iter/s)": 0.005512
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.03469403192052646,
|
| 770 |
+
"grad_norm": 5.491096019744873,
|
| 771 |
+
"learning_rate": 0.00013870967741935487,
|
| 772 |
+
"logits/chosen": -0.5798231363296509,
|
| 773 |
+
"logits/rejected": -0.5843824148178101,
|
| 774 |
+
"logps/chosen": -8.180625915527344,
|
| 775 |
+
"logps/rejected": -19.614505767822266,
|
| 776 |
+
"loss": 1.0811830759048462,
|
| 777 |
+
"memory(GiB)": 46.83,
|
| 778 |
+
"nll_loss": 0.5651165246963501,
|
| 779 |
+
"rewards/accuracies": 0.75,
|
| 780 |
+
"rewards/chosen": 0.018763888627290726,
|
| 781 |
+
"rewards/margins": 0.5782932043075562,
|
| 782 |
+
"rewards/rejected": -0.5595293641090393,
|
| 783 |
+
"step": 43,
|
| 784 |
+
"train_speed(iter/s)": 0.005507
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.035500869872166614,
|
| 788 |
+
"grad_norm": 5.715328216552734,
|
| 789 |
+
"learning_rate": 0.00014193548387096775,
|
| 790 |
+
"logits/chosen": -0.5562050342559814,
|
| 791 |
+
"logits/rejected": -0.5585284233093262,
|
| 792 |
+
"logps/chosen": -9.799619674682617,
|
| 793 |
+
"logps/rejected": -13.962825775146484,
|
| 794 |
+
"loss": 1.046175479888916,
|
| 795 |
+
"memory(GiB)": 46.83,
|
| 796 |
+
"nll_loss": 0.5247361660003662,
|
| 797 |
+
"rewards/accuracies": 0.6875,
|
| 798 |
+
"rewards/chosen": 0.0660993829369545,
|
| 799 |
+
"rewards/margins": 0.557608425617218,
|
| 800 |
+
"rewards/rejected": -0.4915090799331665,
|
| 801 |
+
"step": 44,
|
| 802 |
+
"train_speed(iter/s)": 0.005509
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.03630770782380676,
|
| 806 |
+
"grad_norm": 4.065413951873779,
|
| 807 |
+
"learning_rate": 0.00014516129032258066,
|
| 808 |
+
"logits/chosen": -0.5717958807945251,
|
| 809 |
+
"logits/rejected": -0.5754877924919128,
|
| 810 |
+
"logps/chosen": -10.432125091552734,
|
| 811 |
+
"logps/rejected": -17.75714111328125,
|
| 812 |
+
"loss": 0.9775079488754272,
|
| 813 |
+
"memory(GiB)": 46.83,
|
| 814 |
+
"nll_loss": 0.4804929792881012,
|
| 815 |
+
"rewards/accuracies": 0.71875,
|
| 816 |
+
"rewards/chosen": 0.16446207463741302,
|
| 817 |
+
"rewards/margins": 0.5844215154647827,
|
| 818 |
+
"rewards/rejected": -0.4199594557285309,
|
| 819 |
+
"step": 45,
|
| 820 |
+
"train_speed(iter/s)": 0.005513
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.037114545775446914,
|
| 824 |
+
"grad_norm": 4.507160663604736,
|
| 825 |
+
"learning_rate": 0.00014838709677419355,
|
| 826 |
+
"logits/chosen": -0.5544827580451965,
|
| 827 |
+
"logits/rejected": -0.5517963171005249,
|
| 828 |
+
"logps/chosen": -10.039278030395508,
|
| 829 |
+
"logps/rejected": -16.44822120666504,
|
| 830 |
+
"loss": 0.9198786020278931,
|
| 831 |
+
"memory(GiB)": 46.83,
|
| 832 |
+
"nll_loss": 0.3649781346321106,
|
| 833 |
+
"rewards/accuracies": 0.6875,
|
| 834 |
+
"rewards/chosen": 0.11493994295597076,
|
| 835 |
+
"rewards/margins": 0.5548471808433533,
|
| 836 |
+
"rewards/rejected": -0.4399072825908661,
|
| 837 |
+
"step": 46,
|
| 838 |
+
"train_speed(iter/s)": 0.005512
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.03792138372708706,
|
| 842 |
+
"grad_norm": 4.040234565734863,
|
| 843 |
+
"learning_rate": 0.00015161290322580646,
|
| 844 |
+
"logits/chosen": -0.5704259276390076,
|
| 845 |
+
"logits/rejected": -0.5716680884361267,
|
| 846 |
+
"logps/chosen": -7.400537967681885,
|
| 847 |
+
"logps/rejected": -17.6054744720459,
|
| 848 |
+
"loss": 0.9134971499443054,
|
| 849 |
+
"memory(GiB)": 46.83,
|
| 850 |
+
"nll_loss": 0.3412809669971466,
|
| 851 |
+
"rewards/accuracies": 0.78125,
|
| 852 |
+
"rewards/chosen": 0.03388240188360214,
|
| 853 |
+
"rewards/margins": 0.4440326392650604,
|
| 854 |
+
"rewards/rejected": -0.4101502001285553,
|
| 855 |
+
"step": 47,
|
| 856 |
+
"train_speed(iter/s)": 0.005512
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.038728221678727213,
|
| 860 |
+
"grad_norm": 2.9642419815063477,
|
| 861 |
+
"learning_rate": 0.00015483870967741937,
|
| 862 |
+
"logits/chosen": -0.5867742300033569,
|
| 863 |
+
"logits/rejected": -0.5912387371063232,
|
| 864 |
+
"logps/chosen": -9.146940231323242,
|
| 865 |
+
"logps/rejected": -16.374706268310547,
|
| 866 |
+
"loss": 0.9059248566627502,
|
| 867 |
+
"memory(GiB)": 46.83,
|
| 868 |
+
"nll_loss": 0.3897300958633423,
|
| 869 |
+
"rewards/accuracies": 0.78125,
|
| 870 |
+
"rewards/chosen": 0.1047089472413063,
|
| 871 |
+
"rewards/margins": 0.5065208673477173,
|
| 872 |
+
"rewards/rejected": -0.4018118679523468,
|
| 873 |
+
"step": 48,
|
| 874 |
+
"train_speed(iter/s)": 0.005514
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.03953505963036737,
|
| 878 |
+
"grad_norm": 3.087217330932617,
|
| 879 |
+
"learning_rate": 0.00015806451612903225,
|
| 880 |
+
"logits/chosen": -0.5874481201171875,
|
| 881 |
+
"logits/rejected": -0.5872229337692261,
|
| 882 |
+
"logps/chosen": -6.086337566375732,
|
| 883 |
+
"logps/rejected": -10.504037857055664,
|
| 884 |
+
"loss": 0.8798851370811462,
|
| 885 |
+
"memory(GiB)": 46.83,
|
| 886 |
+
"nll_loss": 0.33263421058654785,
|
| 887 |
+
"rewards/accuracies": 0.6875,
|
| 888 |
+
"rewards/chosen": 0.20601165294647217,
|
| 889 |
+
"rewards/margins": 0.4658709466457367,
|
| 890 |
+
"rewards/rejected": -0.2598593533039093,
|
| 891 |
+
"step": 49,
|
| 892 |
+
"train_speed(iter/s)": 0.005513
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.04034189758200751,
|
| 896 |
+
"grad_norm": 3.16754150390625,
|
| 897 |
+
"learning_rate": 0.00016129032258064516,
|
| 898 |
+
"logits/chosen": -0.5708128809928894,
|
| 899 |
+
"logits/rejected": -0.5798742175102234,
|
| 900 |
+
"logps/chosen": -6.393083572387695,
|
| 901 |
+
"logps/rejected": -19.668724060058594,
|
| 902 |
+
"loss": 0.8404853343963623,
|
| 903 |
+
"memory(GiB)": 46.83,
|
| 904 |
+
"nll_loss": 0.34996694326400757,
|
| 905 |
+
"rewards/accuracies": 0.84375,
|
| 906 |
+
"rewards/chosen": 0.21206828951835632,
|
| 907 |
+
"rewards/margins": 0.5756057500839233,
|
| 908 |
+
"rewards/rejected": -0.363537460565567,
|
| 909 |
+
"step": 50,
|
| 910 |
+
"train_speed(iter/s)": 0.005516
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.041148735533647667,
|
| 914 |
+
"grad_norm": 2.4009766578674316,
|
| 915 |
+
"learning_rate": 0.00016451612903225807,
|
| 916 |
+
"logits/chosen": -0.5599406361579895,
|
| 917 |
+
"logits/rejected": -0.5625043511390686,
|
| 918 |
+
"logps/chosen": -7.262391090393066,
|
| 919 |
+
"logps/rejected": -15.73720932006836,
|
| 920 |
+
"loss": 0.8113247752189636,
|
| 921 |
+
"memory(GiB)": 46.83,
|
| 922 |
+
"nll_loss": 0.28076890110969543,
|
| 923 |
+
"rewards/accuracies": 0.8125,
|
| 924 |
+
"rewards/chosen": 0.14269490540027618,
|
| 925 |
+
"rewards/margins": 0.47564250230789185,
|
| 926 |
+
"rewards/rejected": -0.33294758200645447,
|
| 927 |
+
"step": 51,
|
| 928 |
+
"train_speed(iter/s)": 0.005506
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.04195557348528781,
|
| 932 |
+
"grad_norm": 3.129634380340576,
|
| 933 |
+
"learning_rate": 0.00016774193548387098,
|
| 934 |
+
"logits/chosen": -0.5937235355377197,
|
| 935 |
+
"logits/rejected": -0.5938248038291931,
|
| 936 |
+
"logps/chosen": -5.393199920654297,
|
| 937 |
+
"logps/rejected": -15.786672592163086,
|
| 938 |
+
"loss": 0.7831435799598694,
|
| 939 |
+
"memory(GiB)": 46.83,
|
| 940 |
+
"nll_loss": 0.2872075140476227,
|
| 941 |
+
"rewards/accuracies": 0.8125,
|
| 942 |
+
"rewards/chosen": 0.19873687624931335,
|
| 943 |
+
"rewards/margins": 0.5569283366203308,
|
| 944 |
+
"rewards/rejected": -0.35819149017333984,
|
| 945 |
+
"step": 52,
|
| 946 |
+
"train_speed(iter/s)": 0.005507
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.042762411436927966,
|
| 950 |
+
"grad_norm": 3.2887256145477295,
|
| 951 |
+
"learning_rate": 0.0001709677419354839,
|
| 952 |
+
"logits/chosen": -0.5820189118385315,
|
| 953 |
+
"logits/rejected": -0.5850119590759277,
|
| 954 |
+
"logps/chosen": -7.764882564544678,
|
| 955 |
+
"logps/rejected": -16.894685745239258,
|
| 956 |
+
"loss": 0.8994213342666626,
|
| 957 |
+
"memory(GiB)": 46.83,
|
| 958 |
+
"nll_loss": 0.3770924210548401,
|
| 959 |
+
"rewards/accuracies": 0.6875,
|
| 960 |
+
"rewards/chosen": 0.20706558227539062,
|
| 961 |
+
"rewards/margins": 0.5472366213798523,
|
| 962 |
+
"rewards/rejected": -0.34017103910446167,
|
| 963 |
+
"step": 53,
|
| 964 |
+
"train_speed(iter/s)": 0.005506
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.04356924938856811,
|
| 968 |
+
"grad_norm": 3.7313735485076904,
|
| 969 |
+
"learning_rate": 0.00017419354838709678,
|
| 970 |
+
"logits/chosen": -0.5833962559700012,
|
| 971 |
+
"logits/rejected": -0.5876143574714661,
|
| 972 |
+
"logps/chosen": -8.130949020385742,
|
| 973 |
+
"logps/rejected": -20.79953384399414,
|
| 974 |
+
"loss": 0.9545141458511353,
|
| 975 |
+
"memory(GiB)": 46.83,
|
| 976 |
+
"nll_loss": 0.4539381265640259,
|
| 977 |
+
"rewards/accuracies": 0.75,
|
| 978 |
+
"rewards/chosen": 0.10515612363815308,
|
| 979 |
+
"rewards/margins": 0.612572968006134,
|
| 980 |
+
"rewards/rejected": -0.5074167847633362,
|
| 981 |
+
"step": 54,
|
| 982 |
+
"train_speed(iter/s)": 0.005507
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.044376087340208266,
|
| 986 |
+
"grad_norm": 5.652868747711182,
|
| 987 |
+
"learning_rate": 0.0001774193548387097,
|
| 988 |
+
"logits/chosen": -0.5960416793823242,
|
| 989 |
+
"logits/rejected": -0.5957224369049072,
|
| 990 |
+
"logps/chosen": -12.674562454223633,
|
| 991 |
+
"logps/rejected": -15.591826438903809,
|
| 992 |
+
"loss": 1.1736092567443848,
|
| 993 |
+
"memory(GiB)": 46.83,
|
| 994 |
+
"nll_loss": 0.5310375690460205,
|
| 995 |
+
"rewards/accuracies": 0.625,
|
| 996 |
+
"rewards/chosen": -0.11839848756790161,
|
| 997 |
+
"rewards/margins": 0.40016603469848633,
|
| 998 |
+
"rewards/rejected": -0.5185645222663879,
|
| 999 |
+
"step": 55,
|
| 1000 |
+
"train_speed(iter/s)": 0.005509
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.04518292529184841,
|
| 1004 |
+
"grad_norm": 4.002370834350586,
|
| 1005 |
+
"learning_rate": 0.00018064516129032257,
|
| 1006 |
+
"logits/chosen": -0.5776931643486023,
|
| 1007 |
+
"logits/rejected": -0.5789185166358948,
|
| 1008 |
+
"logps/chosen": -12.67449951171875,
|
| 1009 |
+
"logps/rejected": -16.919376373291016,
|
| 1010 |
+
"loss": 1.087173342704773,
|
| 1011 |
+
"memory(GiB)": 46.83,
|
| 1012 |
+
"nll_loss": 0.4873568117618561,
|
| 1013 |
+
"rewards/accuracies": 0.65625,
|
| 1014 |
+
"rewards/chosen": -0.15739397704601288,
|
| 1015 |
+
"rewards/margins": 0.39120250940322876,
|
| 1016 |
+
"rewards/rejected": -0.5485965013504028,
|
| 1017 |
+
"step": 56,
|
| 1018 |
+
"train_speed(iter/s)": 0.005507
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.045989763243488566,
|
| 1022 |
+
"grad_norm": 3.5529890060424805,
|
| 1023 |
+
"learning_rate": 0.00018387096774193548,
|
| 1024 |
+
"logits/chosen": -0.5732367634773254,
|
| 1025 |
+
"logits/rejected": -0.5760124921798706,
|
| 1026 |
+
"logps/chosen": -4.22891902923584,
|
| 1027 |
+
"logps/rejected": -18.773733139038086,
|
| 1028 |
+
"loss": 0.6395769119262695,
|
| 1029 |
+
"memory(GiB)": 46.83,
|
| 1030 |
+
"nll_loss": 0.23632453382015228,
|
| 1031 |
+
"rewards/accuracies": 0.9375,
|
| 1032 |
+
"rewards/chosen": 0.12576641142368317,
|
| 1033 |
+
"rewards/margins": 0.908116340637207,
|
| 1034 |
+
"rewards/rejected": -0.7823498845100403,
|
| 1035 |
+
"step": 57,
|
| 1036 |
+
"train_speed(iter/s)": 0.005505
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.04679660119512872,
|
| 1040 |
+
"grad_norm": 2.638253927230835,
|
| 1041 |
+
"learning_rate": 0.0001870967741935484,
|
| 1042 |
+
"logits/chosen": -0.5618959069252014,
|
| 1043 |
+
"logits/rejected": -0.5671476721763611,
|
| 1044 |
+
"logps/chosen": -8.717156410217285,
|
| 1045 |
+
"logps/rejected": -22.3723087310791,
|
| 1046 |
+
"loss": 0.8119103312492371,
|
| 1047 |
+
"memory(GiB)": 46.83,
|
| 1048 |
+
"nll_loss": 0.4192086458206177,
|
| 1049 |
+
"rewards/accuracies": 0.9375,
|
| 1050 |
+
"rewards/chosen": 0.18826748430728912,
|
| 1051 |
+
"rewards/margins": 0.8515651226043701,
|
| 1052 |
+
"rewards/rejected": -0.6632976531982422,
|
| 1053 |
+
"step": 58,
|
| 1054 |
+
"train_speed(iter/s)": 0.005503
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.047603439146768865,
|
| 1058 |
+
"grad_norm": 4.416356086730957,
|
| 1059 |
+
"learning_rate": 0.0001903225806451613,
|
| 1060 |
+
"logits/chosen": -0.5946109294891357,
|
| 1061 |
+
"logits/rejected": -0.5949057340621948,
|
| 1062 |
+
"logps/chosen": -10.825102806091309,
|
| 1063 |
+
"logps/rejected": -15.580466270446777,
|
| 1064 |
+
"loss": 0.9124961495399475,
|
| 1065 |
+
"memory(GiB)": 46.83,
|
| 1066 |
+
"nll_loss": 0.43738603591918945,
|
| 1067 |
+
"rewards/accuracies": 0.78125,
|
| 1068 |
+
"rewards/chosen": 0.1081230640411377,
|
| 1069 |
+
"rewards/margins": 0.6975550651550293,
|
| 1070 |
+
"rewards/rejected": -0.5894321203231812,
|
| 1071 |
+
"step": 59,
|
| 1072 |
+
"train_speed(iter/s)": 0.005503
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 0.04841027709840902,
|
| 1076 |
+
"grad_norm": 3.548480272293091,
|
| 1077 |
+
"learning_rate": 0.00019354838709677422,
|
| 1078 |
+
"logits/chosen": -0.576353907585144,
|
| 1079 |
+
"logits/rejected": -0.5807583928108215,
|
| 1080 |
+
"logps/chosen": -10.603282928466797,
|
| 1081 |
+
"logps/rejected": -20.904390335083008,
|
| 1082 |
+
"loss": 1.033903956413269,
|
| 1083 |
+
"memory(GiB)": 46.83,
|
| 1084 |
+
"nll_loss": 0.5458623170852661,
|
| 1085 |
+
"rewards/accuracies": 0.71875,
|
| 1086 |
+
"rewards/chosen": 0.10141083598136902,
|
| 1087 |
+
"rewards/margins": 0.9133819341659546,
|
| 1088 |
+
"rewards/rejected": -0.8119711875915527,
|
| 1089 |
+
"step": 60,
|
| 1090 |
+
"train_speed(iter/s)": 0.005505
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.049217115050049165,
|
| 1094 |
+
"grad_norm": 4.239978790283203,
|
| 1095 |
+
"learning_rate": 0.0001967741935483871,
|
| 1096 |
+
"logits/chosen": -0.6159774661064148,
|
| 1097 |
+
"logits/rejected": -0.6157845258712769,
|
| 1098 |
+
"logps/chosen": -13.034204483032227,
|
| 1099 |
+
"logps/rejected": -21.466520309448242,
|
| 1100 |
+
"loss": 0.9954748153686523,
|
| 1101 |
+
"memory(GiB)": 46.83,
|
| 1102 |
+
"nll_loss": 0.5168237090110779,
|
| 1103 |
+
"rewards/accuracies": 0.8125,
|
| 1104 |
+
"rewards/chosen": -0.15371613204479218,
|
| 1105 |
+
"rewards/margins": 0.9087806940078735,
|
| 1106 |
+
"rewards/rejected": -1.0624967813491821,
|
| 1107 |
+
"step": 61,
|
| 1108 |
+
"train_speed(iter/s)": 0.005508
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.05002395300168932,
|
| 1112 |
+
"grad_norm": 4.418153285980225,
|
| 1113 |
+
"learning_rate": 0.0002,
|
| 1114 |
+
"logits/chosen": -0.586484968662262,
|
| 1115 |
+
"logits/rejected": -0.5885915756225586,
|
| 1116 |
+
"logps/chosen": -6.359810829162598,
|
| 1117 |
+
"logps/rejected": -18.686431884765625,
|
| 1118 |
+
"loss": 0.7993831634521484,
|
| 1119 |
+
"memory(GiB)": 46.83,
|
| 1120 |
+
"nll_loss": 0.3223803639411926,
|
| 1121 |
+
"rewards/accuracies": 0.84375,
|
| 1122 |
+
"rewards/chosen": 0.1565569043159485,
|
| 1123 |
+
"rewards/margins": 0.8297788500785828,
|
| 1124 |
+
"rewards/rejected": -0.6732219457626343,
|
| 1125 |
+
"step": 62,
|
| 1126 |
+
"train_speed(iter/s)": 0.005508
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 0.050830790953329465,
|
| 1130 |
+
"grad_norm": 3.898704767227173,
|
| 1131 |
+
"learning_rate": 0.00019999964378142156,
|
| 1132 |
+
"logits/chosen": -0.5732013583183289,
|
| 1133 |
+
"logits/rejected": -0.5774887800216675,
|
| 1134 |
+
"logps/chosen": -12.98630142211914,
|
| 1135 |
+
"logps/rejected": -19.34124755859375,
|
| 1136 |
+
"loss": 0.9075384140014648,
|
| 1137 |
+
"memory(GiB)": 46.83,
|
| 1138 |
+
"nll_loss": 0.39192017912864685,
|
| 1139 |
+
"rewards/accuracies": 0.75,
|
| 1140 |
+
"rewards/chosen": 0.19289077818393707,
|
| 1141 |
+
"rewards/margins": 0.6732712388038635,
|
| 1142 |
+
"rewards/rejected": -0.48038047552108765,
|
| 1143 |
+
"step": 63,
|
| 1144 |
+
"train_speed(iter/s)": 0.005507
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.05163762890496962,
|
| 1148 |
+
"grad_norm": 6.278528213500977,
|
| 1149 |
+
"learning_rate": 0.00019999857512822402,
|
| 1150 |
+
"logits/chosen": -0.6075332760810852,
|
| 1151 |
+
"logits/rejected": -0.6084203124046326,
|
| 1152 |
+
"logps/chosen": -14.246075630187988,
|
| 1153 |
+
"logps/rejected": -20.948959350585938,
|
| 1154 |
+
"loss": 1.0685752630233765,
|
| 1155 |
+
"memory(GiB)": 46.83,
|
| 1156 |
+
"nll_loss": 0.5815572142601013,
|
| 1157 |
+
"rewards/accuracies": 0.71875,
|
| 1158 |
+
"rewards/chosen": -0.03303433582186699,
|
| 1159 |
+
"rewards/margins": 0.8266364336013794,
|
| 1160 |
+
"rewards/rejected": -0.8596707582473755,
|
| 1161 |
+
"step": 64,
|
| 1162 |
+
"train_speed(iter/s)": 0.005508
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.052444466856609764,
|
| 1166 |
+
"grad_norm": 4.589393615722656,
|
| 1167 |
+
"learning_rate": 0.00019999679404802089,
|
| 1168 |
+
"logits/chosen": -0.596097469329834,
|
| 1169 |
+
"logits/rejected": -0.5986034870147705,
|
| 1170 |
+
"logps/chosen": -8.450502395629883,
|
| 1171 |
+
"logps/rejected": -21.070066452026367,
|
| 1172 |
+
"loss": 0.7867013812065125,
|
| 1173 |
+
"memory(GiB)": 46.83,
|
| 1174 |
+
"nll_loss": 0.31530463695526123,
|
| 1175 |
+
"rewards/accuracies": 0.84375,
|
| 1176 |
+
"rewards/chosen": 0.13920006155967712,
|
| 1177 |
+
"rewards/margins": 0.9423296451568604,
|
| 1178 |
+
"rewards/rejected": -0.8031294941902161,
|
| 1179 |
+
"step": 65,
|
| 1180 |
+
"train_speed(iter/s)": 0.005508
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.05325130480824992,
|
| 1184 |
+
"grad_norm": 4.376217365264893,
|
| 1185 |
+
"learning_rate": 0.00019999430055350122,
|
| 1186 |
+
"logits/chosen": -0.5758836269378662,
|
| 1187 |
+
"logits/rejected": -0.5734565854072571,
|
| 1188 |
+
"logps/chosen": -10.774827003479004,
|
| 1189 |
+
"logps/rejected": -19.021907806396484,
|
| 1190 |
+
"loss": 0.7909982800483704,
|
| 1191 |
+
"memory(GiB)": 46.83,
|
| 1192 |
+
"nll_loss": 0.3997192680835724,
|
| 1193 |
+
"rewards/accuracies": 0.9375,
|
| 1194 |
+
"rewards/chosen": 0.23606787621974945,
|
| 1195 |
+
"rewards/margins": 1.0521141290664673,
|
| 1196 |
+
"rewards/rejected": -0.8160461783409119,
|
| 1197 |
+
"step": 66,
|
| 1198 |
+
"train_speed(iter/s)": 0.005507
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.05405814275989007,
|
| 1202 |
+
"grad_norm": 6.438971042633057,
|
| 1203 |
+
"learning_rate": 0.00019999109466242962,
|
| 1204 |
+
"logits/chosen": -0.5718088150024414,
|
| 1205 |
+
"logits/rejected": -0.5745964050292969,
|
| 1206 |
+
"logps/chosen": -6.671931266784668,
|
| 1207 |
+
"logps/rejected": -20.429231643676758,
|
| 1208 |
+
"loss": 0.8954067230224609,
|
| 1209 |
+
"memory(GiB)": 46.83,
|
| 1210 |
+
"nll_loss": 0.45071539282798767,
|
| 1211 |
+
"rewards/accuracies": 0.8125,
|
| 1212 |
+
"rewards/chosen": 0.10935470461845398,
|
| 1213 |
+
"rewards/margins": 0.8583554029464722,
|
| 1214 |
+
"rewards/rejected": -0.7490006685256958,
|
| 1215 |
+
"step": 67,
|
| 1216 |
+
"train_speed(iter/s)": 0.005509
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.05486498071153022,
|
| 1220 |
+
"grad_norm": 11.16286849975586,
|
| 1221 |
+
"learning_rate": 0.00019998717639764602,
|
| 1222 |
+
"logits/chosen": -0.5820534229278564,
|
| 1223 |
+
"logits/rejected": -0.5849663019180298,
|
| 1224 |
+
"logps/chosen": -9.799468040466309,
|
| 1225 |
+
"logps/rejected": -18.977890014648438,
|
| 1226 |
+
"loss": 1.29269278049469,
|
| 1227 |
+
"memory(GiB)": 46.83,
|
| 1228 |
+
"nll_loss": 0.7049438953399658,
|
| 1229 |
+
"rewards/accuracies": 0.6875,
|
| 1230 |
+
"rewards/chosen": -0.04825415089726448,
|
| 1231 |
+
"rewards/margins": 0.5465254783630371,
|
| 1232 |
+
"rewards/rejected": -0.5947796702384949,
|
| 1233 |
+
"step": 68,
|
| 1234 |
+
"train_speed(iter/s)": 0.005506
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.05567181866317037,
|
| 1238 |
+
"grad_norm": 9.413622856140137,
|
| 1239 |
+
"learning_rate": 0.00019998254578706563,
|
| 1240 |
+
"logits/chosen": -0.5850585699081421,
|
| 1241 |
+
"logits/rejected": -0.5874402523040771,
|
| 1242 |
+
"logps/chosen": -8.903966903686523,
|
| 1243 |
+
"logps/rejected": -14.843938827514648,
|
| 1244 |
+
"loss": 1.3027069568634033,
|
| 1245 |
+
"memory(GiB)": 46.83,
|
| 1246 |
+
"nll_loss": 0.744307279586792,
|
| 1247 |
+
"rewards/accuracies": 0.78125,
|
| 1248 |
+
"rewards/chosen": -0.0512365885078907,
|
| 1249 |
+
"rewards/margins": 0.5854336619377136,
|
| 1250 |
+
"rewards/rejected": -0.6366702318191528,
|
| 1251 |
+
"step": 69,
|
| 1252 |
+
"train_speed(iter/s)": 0.005505
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.05647865661481052,
|
| 1256 |
+
"grad_norm": 7.333382606506348,
|
| 1257 |
+
"learning_rate": 0.00019997720286367863,
|
| 1258 |
+
"logits/chosen": -0.545200765132904,
|
| 1259 |
+
"logits/rejected": -0.5484716296195984,
|
| 1260 |
+
"logps/chosen": -7.334113121032715,
|
| 1261 |
+
"logps/rejected": -22.56316375732422,
|
| 1262 |
+
"loss": 1.0189334154129028,
|
| 1263 |
+
"memory(GiB)": 46.83,
|
| 1264 |
+
"nll_loss": 0.5383400321006775,
|
| 1265 |
+
"rewards/accuracies": 0.84375,
|
| 1266 |
+
"rewards/chosen": 0.061847541481256485,
|
| 1267 |
+
"rewards/margins": 0.6559779644012451,
|
| 1268 |
+
"rewards/rejected": -0.5941305160522461,
|
| 1269 |
+
"step": 70,
|
| 1270 |
+
"train_speed(iter/s)": 0.005506
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.05728549456645067,
|
| 1274 |
+
"grad_norm": 2.8849799633026123,
|
| 1275 |
+
"learning_rate": 0.00019997114766554997,
|
| 1276 |
+
"logits/chosen": -0.5629354119300842,
|
| 1277 |
+
"logits/rejected": -0.5669021606445312,
|
| 1278 |
+
"logps/chosen": -8.698752403259277,
|
| 1279 |
+
"logps/rejected": -21.85867691040039,
|
| 1280 |
+
"loss": 0.8375651836395264,
|
| 1281 |
+
"memory(GiB)": 46.83,
|
| 1282 |
+
"nll_loss": 0.3549819886684418,
|
| 1283 |
+
"rewards/accuracies": 0.78125,
|
| 1284 |
+
"rewards/chosen": 0.09854352474212646,
|
| 1285 |
+
"rewards/margins": 0.7027413845062256,
|
| 1286 |
+
"rewards/rejected": -0.6041979193687439,
|
| 1287 |
+
"step": 71,
|
| 1288 |
+
"train_speed(iter/s)": 0.005507
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 0.05809233251809082,
|
| 1292 |
+
"grad_norm": 2.6733391284942627,
|
| 1293 |
+
"learning_rate": 0.00019996438023581913,
|
| 1294 |
+
"logits/chosen": -0.5319232940673828,
|
| 1295 |
+
"logits/rejected": -0.5331096649169922,
|
| 1296 |
+
"logps/chosen": -10.13166332244873,
|
| 1297 |
+
"logps/rejected": -13.18635368347168,
|
| 1298 |
+
"loss": 1.014970302581787,
|
| 1299 |
+
"memory(GiB)": 46.83,
|
| 1300 |
+
"nll_loss": 0.4217461347579956,
|
| 1301 |
+
"rewards/accuracies": 0.71875,
|
| 1302 |
+
"rewards/chosen": 0.14793364703655243,
|
| 1303 |
+
"rewards/margins": 0.3788456618785858,
|
| 1304 |
+
"rewards/rejected": -0.2309119999408722,
|
| 1305 |
+
"step": 72,
|
| 1306 |
+
"train_speed(iter/s)": 0.005505
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.05889917046973097,
|
| 1310 |
+
"grad_norm": 2.497257947921753,
|
| 1311 |
+
"learning_rate": 0.00019995690062269984,
|
| 1312 |
+
"logits/chosen": -0.5789859294891357,
|
| 1313 |
+
"logits/rejected": -0.5805359482765198,
|
| 1314 |
+
"logps/chosen": -8.139568328857422,
|
| 1315 |
+
"logps/rejected": -10.85029411315918,
|
| 1316 |
+
"loss": 0.9546566605567932,
|
| 1317 |
+
"memory(GiB)": 46.83,
|
| 1318 |
+
"nll_loss": 0.4216194450855255,
|
| 1319 |
+
"rewards/accuracies": 0.65625,
|
| 1320 |
+
"rewards/chosen": 0.24104151129722595,
|
| 1321 |
+
"rewards/margins": 0.5254760980606079,
|
| 1322 |
+
"rewards/rejected": -0.28443461656570435,
|
| 1323 |
+
"step": 73,
|
| 1324 |
+
"train_speed(iter/s)": 0.005506
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.05970600842137112,
|
| 1328 |
+
"grad_norm": 2.8535268306732178,
|
| 1329 |
+
"learning_rate": 0.00019994870887947962,
|
| 1330 |
+
"logits/chosen": -0.5778383612632751,
|
| 1331 |
+
"logits/rejected": -0.5768096446990967,
|
| 1332 |
+
"logps/chosen": -8.49053955078125,
|
| 1333 |
+
"logps/rejected": -12.854878425598145,
|
| 1334 |
+
"loss": 1.0000758171081543,
|
| 1335 |
+
"memory(GiB)": 46.83,
|
| 1336 |
+
"nll_loss": 0.37848562002182007,
|
| 1337 |
+
"rewards/accuracies": 0.625,
|
| 1338 |
+
"rewards/chosen": 0.1054978221654892,
|
| 1339 |
+
"rewards/margins": 0.3462926149368286,
|
| 1340 |
+
"rewards/rejected": -0.24079479277133942,
|
| 1341 |
+
"step": 74,
|
| 1342 |
+
"train_speed(iter/s)": 0.005506
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.06051284637301127,
|
| 1346 |
+
"grad_norm": 2.198607921600342,
|
| 1347 |
+
"learning_rate": 0.00019993980506451947,
|
| 1348 |
+
"logits/chosen": -0.530400276184082,
|
| 1349 |
+
"logits/rejected": -0.530858039855957,
|
| 1350 |
+
"logps/chosen": -9.152666091918945,
|
| 1351 |
+
"logps/rejected": -11.464343070983887,
|
| 1352 |
+
"loss": 0.9745745658874512,
|
| 1353 |
+
"memory(GiB)": 46.83,
|
| 1354 |
+
"nll_loss": 0.38881367444992065,
|
| 1355 |
+
"rewards/accuracies": 0.75,
|
| 1356 |
+
"rewards/chosen": 0.1783803254365921,
|
| 1357 |
+
"rewards/margins": 0.3315035402774811,
|
| 1358 |
+
"rewards/rejected": -0.1531231850385666,
|
| 1359 |
+
"step": 75,
|
| 1360 |
+
"train_speed(iter/s)": 0.005504
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.06131968432465142,
|
| 1364 |
+
"grad_norm": 2.461601972579956,
|
| 1365 |
+
"learning_rate": 0.0001999301892412535,
|
| 1366 |
+
"logits/chosen": -0.5405219197273254,
|
| 1367 |
+
"logits/rejected": -0.5390282869338989,
|
| 1368 |
+
"logps/chosen": -10.537997245788574,
|
| 1369 |
+
"logps/rejected": -15.066913604736328,
|
| 1370 |
+
"loss": 0.966386079788208,
|
| 1371 |
+
"memory(GiB)": 46.83,
|
| 1372 |
+
"nll_loss": 0.38977545499801636,
|
| 1373 |
+
"rewards/accuracies": 0.71875,
|
| 1374 |
+
"rewards/chosen": 0.09291838109493256,
|
| 1375 |
+
"rewards/margins": 0.3533533811569214,
|
| 1376 |
+
"rewards/rejected": -0.26043495535850525,
|
| 1377 |
+
"step": 76,
|
| 1378 |
+
"train_speed(iter/s)": 0.005503
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.06212652227629157,
|
| 1382 |
+
"grad_norm": 2.0999796390533447,
|
| 1383 |
+
"learning_rate": 0.00019991986147818838,
|
| 1384 |
+
"logits/chosen": -0.5443933010101318,
|
| 1385 |
+
"logits/rejected": -0.5435047149658203,
|
| 1386 |
+
"logps/chosen": -12.143755912780762,
|
| 1387 |
+
"logps/rejected": -15.766691207885742,
|
| 1388 |
+
"loss": 0.8939065337181091,
|
| 1389 |
+
"memory(GiB)": 46.83,
|
| 1390 |
+
"nll_loss": 0.415660560131073,
|
| 1391 |
+
"rewards/accuracies": 0.90625,
|
| 1392 |
+
"rewards/chosen": 0.23248028755187988,
|
| 1393 |
+
"rewards/margins": 0.5352811217308044,
|
| 1394 |
+
"rewards/rejected": -0.30280083417892456,
|
| 1395 |
+
"step": 77,
|
| 1396 |
+
"train_speed(iter/s)": 0.005502
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.06293336022793172,
|
| 1400 |
+
"grad_norm": 2.9244306087493896,
|
| 1401 |
+
"learning_rate": 0.00019990882184890295,
|
| 1402 |
+
"logits/chosen": -0.5794655680656433,
|
| 1403 |
+
"logits/rejected": -0.5853484869003296,
|
| 1404 |
+
"logps/chosen": -7.884565353393555,
|
| 1405 |
+
"logps/rejected": -20.136926651000977,
|
| 1406 |
+
"loss": 0.8564450740814209,
|
| 1407 |
+
"memory(GiB)": 46.83,
|
| 1408 |
+
"nll_loss": 0.34705159068107605,
|
| 1409 |
+
"rewards/accuracies": 0.875,
|
| 1410 |
+
"rewards/chosen": 0.14532405138015747,
|
| 1411 |
+
"rewards/margins": 0.5069437623023987,
|
| 1412 |
+
"rewards/rejected": -0.3616198003292084,
|
| 1413 |
+
"step": 78,
|
| 1414 |
+
"train_speed(iter/s)": 0.005502
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.06374019817957187,
|
| 1418 |
+
"grad_norm": 3.461670160293579,
|
| 1419 |
+
"learning_rate": 0.00019989707043204763,
|
| 1420 |
+
"logits/chosen": -0.5491142868995667,
|
| 1421 |
+
"logits/rejected": -0.553898811340332,
|
| 1422 |
+
"logps/chosen": -7.547114372253418,
|
| 1423 |
+
"logps/rejected": -18.740121841430664,
|
| 1424 |
+
"loss": 0.9405392408370972,
|
| 1425 |
+
"memory(GiB)": 46.83,
|
| 1426 |
+
"nll_loss": 0.48508715629577637,
|
| 1427 |
+
"rewards/accuracies": 0.84375,
|
| 1428 |
+
"rewards/chosen": 0.128400981426239,
|
| 1429 |
+
"rewards/margins": 0.6999501585960388,
|
| 1430 |
+
"rewards/rejected": -0.5715491771697998,
|
| 1431 |
+
"step": 79,
|
| 1432 |
+
"train_speed(iter/s)": 0.005502
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 0.06454703613121202,
|
| 1436 |
+
"grad_norm": 4.165892601013184,
|
| 1437 |
+
"learning_rate": 0.0001998846073113439,
|
| 1438 |
+
"logits/chosen": -0.6084208488464355,
|
| 1439 |
+
"logits/rejected": -0.6092680096626282,
|
| 1440 |
+
"logps/chosen": -12.750446319580078,
|
| 1441 |
+
"logps/rejected": -19.53608512878418,
|
| 1442 |
+
"loss": 1.0978946685791016,
|
| 1443 |
+
"memory(GiB)": 46.83,
|
| 1444 |
+
"nll_loss": 0.5877176523208618,
|
| 1445 |
+
"rewards/accuracies": 0.75,
|
| 1446 |
+
"rewards/chosen": 0.1369042694568634,
|
| 1447 |
+
"rewards/margins": 0.6248282194137573,
|
| 1448 |
+
"rewards/rejected": -0.48792392015457153,
|
| 1449 |
+
"step": 80,
|
| 1450 |
+
"train_speed(iter/s)": 0.005503
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.06535387408285218,
|
| 1454 |
+
"grad_norm": 2.287846565246582,
|
| 1455 |
+
"learning_rate": 0.00019987143257558365,
|
| 1456 |
+
"logits/chosen": -0.59499192237854,
|
| 1457 |
+
"logits/rejected": -0.5972949862480164,
|
| 1458 |
+
"logps/chosen": -4.519835948944092,
|
| 1459 |
+
"logps/rejected": -19.02901840209961,
|
| 1460 |
+
"loss": 0.58124840259552,
|
| 1461 |
+
"memory(GiB)": 46.83,
|
| 1462 |
+
"nll_loss": 0.15051494538784027,
|
| 1463 |
+
"rewards/accuracies": 0.84375,
|
| 1464 |
+
"rewards/chosen": 0.19695766270160675,
|
| 1465 |
+
"rewards/margins": 0.9730147123336792,
|
| 1466 |
+
"rewards/rejected": -0.7760570049285889,
|
| 1467 |
+
"step": 81,
|
| 1468 |
+
"train_speed(iter/s)": 0.005506
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 0.06616071203449232,
|
| 1472 |
+
"grad_norm": 7.103982448577881,
|
| 1473 |
+
"learning_rate": 0.00019985754631862855,
|
| 1474 |
+
"logits/chosen": -0.5857492089271545,
|
| 1475 |
+
"logits/rejected": -0.5895102024078369,
|
| 1476 |
+
"logps/chosen": -8.903070449829102,
|
| 1477 |
+
"logps/rejected": -26.571044921875,
|
| 1478 |
+
"loss": 0.7599916458129883,
|
| 1479 |
+
"memory(GiB)": 46.83,
|
| 1480 |
+
"nll_loss": 0.3342111110687256,
|
| 1481 |
+
"rewards/accuracies": 0.84375,
|
| 1482 |
+
"rewards/chosen": 0.14196240901947021,
|
| 1483 |
+
"rewards/margins": 0.8281382322311401,
|
| 1484 |
+
"rewards/rejected": -0.6861758232116699,
|
| 1485 |
+
"step": 82,
|
| 1486 |
+
"train_speed(iter/s)": 0.005504
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.06696754998613247,
|
| 1490 |
+
"grad_norm": 3.764488697052002,
|
| 1491 |
+
"learning_rate": 0.00019984294863940955,
|
| 1492 |
+
"logits/chosen": -0.6067830324172974,
|
| 1493 |
+
"logits/rejected": -0.6086009740829468,
|
| 1494 |
+
"logps/chosen": -9.9881591796875,
|
| 1495 |
+
"logps/rejected": -19.09541893005371,
|
| 1496 |
+
"loss": 0.9752427339553833,
|
| 1497 |
+
"memory(GiB)": 46.83,
|
| 1498 |
+
"nll_loss": 0.5045859217643738,
|
| 1499 |
+
"rewards/accuracies": 0.71875,
|
| 1500 |
+
"rewards/chosen": 0.14939813315868378,
|
| 1501 |
+
"rewards/margins": 0.8939868211746216,
|
| 1502 |
+
"rewards/rejected": -0.7445887327194214,
|
| 1503 |
+
"step": 83,
|
| 1504 |
+
"train_speed(iter/s)": 0.005503
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 0.06777438793777263,
|
| 1508 |
+
"grad_norm": 3.3293344974517822,
|
| 1509 |
+
"learning_rate": 0.00019982763964192585,
|
| 1510 |
+
"logits/chosen": -0.6330868601799011,
|
| 1511 |
+
"logits/rejected": -0.6393448710441589,
|
| 1512 |
+
"logps/chosen": -4.902453899383545,
|
| 1513 |
+
"logps/rejected": -24.94113540649414,
|
| 1514 |
+
"loss": 0.7244248390197754,
|
| 1515 |
+
"memory(GiB)": 46.83,
|
| 1516 |
+
"nll_loss": 0.3735058009624481,
|
| 1517 |
+
"rewards/accuracies": 0.9375,
|
| 1518 |
+
"rewards/chosen": 0.17604736983776093,
|
| 1519 |
+
"rewards/margins": 1.2982165813446045,
|
| 1520 |
+
"rewards/rejected": -1.1221693754196167,
|
| 1521 |
+
"step": 84,
|
| 1522 |
+
"train_speed(iter/s)": 0.005503
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.06858122588941278,
|
| 1526 |
+
"grad_norm": 5.892284393310547,
|
| 1527 |
+
"learning_rate": 0.00019981161943524443,
|
| 1528 |
+
"logits/chosen": -0.6240598559379578,
|
| 1529 |
+
"logits/rejected": -0.620481550693512,
|
| 1530 |
+
"logps/chosen": -14.60393238067627,
|
| 1531 |
+
"logps/rejected": -13.365829467773438,
|
| 1532 |
+
"loss": 1.1541650295257568,
|
| 1533 |
+
"memory(GiB)": 46.83,
|
| 1534 |
+
"nll_loss": 0.48222166299819946,
|
| 1535 |
+
"rewards/accuracies": 0.59375,
|
| 1536 |
+
"rewards/chosen": -0.09302416443824768,
|
| 1537 |
+
"rewards/margins": 0.41544342041015625,
|
| 1538 |
+
"rewards/rejected": -0.5084675550460815,
|
| 1539 |
+
"step": 85,
|
| 1540 |
+
"train_speed(iter/s)": 0.0055
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.06938806384105292,
|
| 1544 |
+
"grad_norm": 4.056751251220703,
|
| 1545 |
+
"learning_rate": 0.00019979488813349933,
|
| 1546 |
+
"logits/chosen": -0.6613398194313049,
|
| 1547 |
+
"logits/rejected": -0.6640070080757141,
|
| 1548 |
+
"logps/chosen": -9.50975227355957,
|
| 1549 |
+
"logps/rejected": -21.119707107543945,
|
| 1550 |
+
"loss": 0.8987840414047241,
|
| 1551 |
+
"memory(GiB)": 46.83,
|
| 1552 |
+
"nll_loss": 0.52927565574646,
|
| 1553 |
+
"rewards/accuracies": 0.84375,
|
| 1554 |
+
"rewards/chosen": 0.1999521553516388,
|
| 1555 |
+
"rewards/margins": 1.1560636758804321,
|
| 1556 |
+
"rewards/rejected": -0.956111490726471,
|
| 1557 |
+
"step": 86,
|
| 1558 |
+
"train_speed(iter/s)": 0.005499
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.07019490179269307,
|
| 1562 |
+
"grad_norm": 4.351650238037109,
|
| 1563 |
+
"learning_rate": 0.0001997774458558904,
|
| 1564 |
+
"logits/chosen": -0.6547818779945374,
|
| 1565 |
+
"logits/rejected": -0.6571143269538879,
|
| 1566 |
+
"logps/chosen": -8.668737411499023,
|
| 1567 |
+
"logps/rejected": -20.79534149169922,
|
| 1568 |
+
"loss": 0.9775034785270691,
|
| 1569 |
+
"memory(GiB)": 46.83,
|
| 1570 |
+
"nll_loss": 0.4770965874195099,
|
| 1571 |
+
"rewards/accuracies": 0.75,
|
| 1572 |
+
"rewards/chosen": 0.02099284902215004,
|
| 1573 |
+
"rewards/margins": 0.8633284568786621,
|
| 1574 |
+
"rewards/rejected": -0.84233558177948,
|
| 1575 |
+
"step": 87,
|
| 1576 |
+
"train_speed(iter/s)": 0.005498
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 0.07100173974433323,
|
| 1580 |
+
"grad_norm": 2.8774800300598145,
|
| 1581 |
+
"learning_rate": 0.00019975929272668296,
|
| 1582 |
+
"logits/chosen": -0.6759516000747681,
|
| 1583 |
+
"logits/rejected": -0.6824960708618164,
|
| 1584 |
+
"logps/chosen": -4.7266716957092285,
|
| 1585 |
+
"logps/rejected": -26.10918426513672,
|
| 1586 |
+
"loss": 0.5549903512001038,
|
| 1587 |
+
"memory(GiB)": 46.83,
|
| 1588 |
+
"nll_loss": 0.271104097366333,
|
| 1589 |
+
"rewards/accuracies": 0.9375,
|
| 1590 |
+
"rewards/chosen": 0.25939324498176575,
|
| 1591 |
+
"rewards/margins": 1.6003111600875854,
|
| 1592 |
+
"rewards/rejected": -1.340917944908142,
|
| 1593 |
+
"step": 88,
|
| 1594 |
+
"train_speed(iter/s)": 0.005501
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.07180857769597337,
|
| 1598 |
+
"grad_norm": 2.8334577083587646,
|
| 1599 |
+
"learning_rate": 0.00019974042887520668,
|
| 1600 |
+
"logits/chosen": -0.6535760164260864,
|
| 1601 |
+
"logits/rejected": -0.6572569608688354,
|
| 1602 |
+
"logps/chosen": -13.957544326782227,
|
| 1603 |
+
"logps/rejected": -18.16533851623535,
|
| 1604 |
+
"loss": 0.9486314654350281,
|
| 1605 |
+
"memory(GiB)": 46.83,
|
| 1606 |
+
"nll_loss": 0.5478147268295288,
|
| 1607 |
+
"rewards/accuracies": 0.90625,
|
| 1608 |
+
"rewards/chosen": 0.2579638957977295,
|
| 1609 |
+
"rewards/margins": 0.9477146863937378,
|
| 1610 |
+
"rewards/rejected": -0.6897507905960083,
|
| 1611 |
+
"step": 89,
|
| 1612 |
+
"train_speed(iter/s)": 0.005501
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.07261541564761352,
|
| 1616 |
+
"grad_norm": 5.422155380249023,
|
| 1617 |
+
"learning_rate": 0.00019972085443585463,
|
| 1618 |
+
"logits/chosen": -0.6767461895942688,
|
| 1619 |
+
"logits/rejected": -0.6799700260162354,
|
| 1620 |
+
"logps/chosen": -7.507997512817383,
|
| 1621 |
+
"logps/rejected": -20.19240951538086,
|
| 1622 |
+
"loss": 0.8118898868560791,
|
| 1623 |
+
"memory(GiB)": 46.83,
|
| 1624 |
+
"nll_loss": 0.4057508111000061,
|
| 1625 |
+
"rewards/accuracies": 0.84375,
|
| 1626 |
+
"rewards/chosen": 0.11586537212133408,
|
| 1627 |
+
"rewards/margins": 1.0757473707199097,
|
| 1628 |
+
"rewards/rejected": -0.9598820805549622,
|
| 1629 |
+
"step": 90,
|
| 1630 |
+
"train_speed(iter/s)": 0.005503
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 0.07342225359925368,
|
| 1634 |
+
"grad_norm": 5.541498184204102,
|
| 1635 |
+
"learning_rate": 0.0001997005695480824,
|
| 1636 |
+
"logits/chosen": -0.6398351788520813,
|
| 1637 |
+
"logits/rejected": -0.642162024974823,
|
| 1638 |
+
"logps/chosen": -10.117485046386719,
|
| 1639 |
+
"logps/rejected": -21.747928619384766,
|
| 1640 |
+
"loss": 0.8823437094688416,
|
| 1641 |
+
"memory(GiB)": 46.83,
|
| 1642 |
+
"nll_loss": 0.47223997116088867,
|
| 1643 |
+
"rewards/accuracies": 0.75,
|
| 1644 |
+
"rewards/chosen": -0.04489222913980484,
|
| 1645 |
+
"rewards/margins": 1.1613836288452148,
|
| 1646 |
+
"rewards/rejected": -1.2062758207321167,
|
| 1647 |
+
"step": 91,
|
| 1648 |
+
"train_speed(iter/s)": 0.005504
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.07422909155089383,
|
| 1652 |
+
"grad_norm": 2.0107967853546143,
|
| 1653 |
+
"learning_rate": 0.000199679574356407,
|
| 1654 |
+
"logits/chosen": -0.6938130855560303,
|
| 1655 |
+
"logits/rejected": -0.7009856104850769,
|
| 1656 |
+
"logps/chosen": -6.915675163269043,
|
| 1657 |
+
"logps/rejected": -28.84608268737793,
|
| 1658 |
+
"loss": 0.5776476860046387,
|
| 1659 |
+
"memory(GiB)": 46.83,
|
| 1660 |
+
"nll_loss": 0.3677330017089844,
|
| 1661 |
+
"rewards/accuracies": 1.0,
|
| 1662 |
+
"rewards/chosen": 0.3187498152256012,
|
| 1663 |
+
"rewards/margins": 1.8727707862854004,
|
| 1664 |
+
"rewards/rejected": -1.554020881652832,
|
| 1665 |
+
"step": 92,
|
| 1666 |
+
"train_speed(iter/s)": 0.005502
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.07503592950253397,
|
| 1670 |
+
"grad_norm": 3.5870676040649414,
|
| 1671 |
+
"learning_rate": 0.0001996578690104061,
|
| 1672 |
+
"logits/chosen": -0.673453688621521,
|
| 1673 |
+
"logits/rejected": -0.676167905330658,
|
| 1674 |
+
"logps/chosen": -10.7222318649292,
|
| 1675 |
+
"logps/rejected": -28.77056884765625,
|
| 1676 |
+
"loss": 0.950826108455658,
|
| 1677 |
+
"memory(GiB)": 46.83,
|
| 1678 |
+
"nll_loss": 0.5682229399681091,
|
| 1679 |
+
"rewards/accuracies": 0.8125,
|
| 1680 |
+
"rewards/chosen": 0.09744307398796082,
|
| 1681 |
+
"rewards/margins": 1.4789444208145142,
|
| 1682 |
+
"rewards/rejected": -1.381501317024231,
|
| 1683 |
+
"step": 93,
|
| 1684 |
+
"train_speed(iter/s)": 0.005502
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.07584276745417412,
|
| 1688 |
+
"grad_norm": 4.2534661293029785,
|
| 1689 |
+
"learning_rate": 0.00019963545366471653,
|
| 1690 |
+
"logits/chosen": -0.6709739565849304,
|
| 1691 |
+
"logits/rejected": -0.6747978925704956,
|
| 1692 |
+
"logps/chosen": -8.219854354858398,
|
| 1693 |
+
"logps/rejected": -29.13772201538086,
|
| 1694 |
+
"loss": 0.6954976320266724,
|
| 1695 |
+
"memory(GiB)": 46.83,
|
| 1696 |
+
"nll_loss": 0.35414204001426697,
|
| 1697 |
+
"rewards/accuracies": 0.8125,
|
| 1698 |
+
"rewards/chosen": 0.10512235760688782,
|
| 1699 |
+
"rewards/margins": 1.6458101272583008,
|
| 1700 |
+
"rewards/rejected": -1.5406877994537354,
|
| 1701 |
+
"step": 94,
|
| 1702 |
+
"train_speed(iter/s)": 0.005503
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.07664960540581428,
|
| 1706 |
+
"grad_norm": 7.663514614105225,
|
| 1707 |
+
"learning_rate": 0.0001996123284790336,
|
| 1708 |
+
"logits/chosen": -0.6834441423416138,
|
| 1709 |
+
"logits/rejected": -0.6848198771476746,
|
| 1710 |
+
"logps/chosen": -17.666513442993164,
|
| 1711 |
+
"logps/rejected": -19.222387313842773,
|
| 1712 |
+
"loss": 1.197723388671875,
|
| 1713 |
+
"memory(GiB)": 46.83,
|
| 1714 |
+
"nll_loss": 0.6848238110542297,
|
| 1715 |
+
"rewards/accuracies": 0.84375,
|
| 1716 |
+
"rewards/chosen": -0.06468406319618225,
|
| 1717 |
+
"rewards/margins": 0.8658308982849121,
|
| 1718 |
+
"rewards/rejected": -0.9305148720741272,
|
| 1719 |
+
"step": 95,
|
| 1720 |
+
"train_speed(iter/s)": 0.005502
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 0.07745644335745443,
|
| 1724 |
+
"grad_norm": 16.187767028808594,
|
| 1725 |
+
"learning_rate": 0.0001995884936181097,
|
| 1726 |
+
"logits/chosen": -0.7171504497528076,
|
| 1727 |
+
"logits/rejected": -0.7210450172424316,
|
| 1728 |
+
"logps/chosen": -4.8772172927856445,
|
| 1729 |
+
"logps/rejected": -27.164329528808594,
|
| 1730 |
+
"loss": 0.7958053946495056,
|
| 1731 |
+
"memory(GiB)": 46.83,
|
| 1732 |
+
"nll_loss": 0.437197208404541,
|
| 1733 |
+
"rewards/accuracies": 0.875,
|
| 1734 |
+
"rewards/chosen": 0.04156189411878586,
|
| 1735 |
+
"rewards/margins": 1.4321556091308594,
|
| 1736 |
+
"rewards/rejected": -1.3905936479568481,
|
| 1737 |
+
"step": 96,
|
| 1738 |
+
"train_speed(iter/s)": 0.005502
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.07826328130909457,
|
| 1742 |
+
"grad_norm": 4.373692989349365,
|
| 1743 |
+
"learning_rate": 0.00019956394925175328,
|
| 1744 |
+
"logits/chosen": -0.720712423324585,
|
| 1745 |
+
"logits/rejected": -0.7263769507408142,
|
| 1746 |
+
"logps/chosen": -8.969646453857422,
|
| 1747 |
+
"logps/rejected": -27.35487174987793,
|
| 1748 |
+
"loss": 0.8684384822845459,
|
| 1749 |
+
"memory(GiB)": 46.83,
|
| 1750 |
+
"nll_loss": 0.4593147337436676,
|
| 1751 |
+
"rewards/accuracies": 0.78125,
|
| 1752 |
+
"rewards/chosen": 0.039580605924129486,
|
| 1753 |
+
"rewards/margins": 1.404889702796936,
|
| 1754 |
+
"rewards/rejected": -1.3653091192245483,
|
| 1755 |
+
"step": 97,
|
| 1756 |
+
"train_speed(iter/s)": 0.005502
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.07907011926073473,
|
| 1760 |
+
"grad_norm": 3.1712353229522705,
|
| 1761 |
+
"learning_rate": 0.00019953869555482752,
|
| 1762 |
+
"logits/chosen": -0.6926038265228271,
|
| 1763 |
+
"logits/rejected": -0.6906510591506958,
|
| 1764 |
+
"logps/chosen": -10.361411094665527,
|
| 1765 |
+
"logps/rejected": -19.763336181640625,
|
| 1766 |
+
"loss": 1.0741034746170044,
|
| 1767 |
+
"memory(GiB)": 46.83,
|
| 1768 |
+
"nll_loss": 0.5404018759727478,
|
| 1769 |
+
"rewards/accuracies": 0.75,
|
| 1770 |
+
"rewards/chosen": 0.0025314167141914368,
|
| 1771 |
+
"rewards/margins": 0.9160726070404053,
|
| 1772 |
+
"rewards/rejected": -0.9135411381721497,
|
| 1773 |
+
"step": 98,
|
| 1774 |
+
"train_speed(iter/s)": 0.005502
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.07987695721237488,
|
| 1778 |
+
"grad_norm": 5.8308796882629395,
|
| 1779 |
+
"learning_rate": 0.00019951273270724906,
|
| 1780 |
+
"logits/chosen": -0.6417246460914612,
|
| 1781 |
+
"logits/rejected": -0.6460192203521729,
|
| 1782 |
+
"logps/chosen": -10.923144340515137,
|
| 1783 |
+
"logps/rejected": -23.332698822021484,
|
| 1784 |
+
"loss": 1.129643201828003,
|
| 1785 |
+
"memory(GiB)": 46.83,
|
| 1786 |
+
"nll_loss": 0.5929185748100281,
|
| 1787 |
+
"rewards/accuracies": 0.78125,
|
| 1788 |
+
"rewards/chosen": -0.004452264867722988,
|
| 1789 |
+
"rewards/margins": 0.6838414669036865,
|
| 1790 |
+
"rewards/rejected": -0.6882937550544739,
|
| 1791 |
+
"step": 99,
|
| 1792 |
+
"train_speed(iter/s)": 0.005503
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 0.08068379516401503,
|
| 1796 |
+
"grad_norm": 4.135215759277344,
|
| 1797 |
+
"learning_rate": 0.00019948606089398698,
|
| 1798 |
+
"logits/chosen": -0.6685535907745361,
|
| 1799 |
+
"logits/rejected": -0.6693455576896667,
|
| 1800 |
+
"logps/chosen": -11.88547134399414,
|
| 1801 |
+
"logps/rejected": -19.97165298461914,
|
| 1802 |
+
"loss": 0.9190454483032227,
|
| 1803 |
+
"memory(GiB)": 46.83,
|
| 1804 |
+
"nll_loss": 0.39345046877861023,
|
| 1805 |
+
"rewards/accuracies": 0.59375,
|
| 1806 |
+
"rewards/chosen": 0.05392542481422424,
|
| 1807 |
+
"rewards/margins": 0.8620406985282898,
|
| 1808 |
+
"rewards/rejected": -0.8081153035163879,
|
| 1809 |
+
"step": 100,
|
| 1810 |
+
"train_speed(iter/s)": 0.005503
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.08149063311565517,
|
| 1814 |
+
"grad_norm": 3.957230806350708,
|
| 1815 |
+
"learning_rate": 0.0001994586803050611,
|
| 1816 |
+
"logits/chosen": -0.6953858137130737,
|
| 1817 |
+
"logits/rejected": -0.6939084529876709,
|
| 1818 |
+
"logps/chosen": -11.101150512695312,
|
| 1819 |
+
"logps/rejected": -19.7403621673584,
|
| 1820 |
+
"loss": 1.0100737810134888,
|
| 1821 |
+
"memory(GiB)": 46.83,
|
| 1822 |
+
"nll_loss": 0.45983725786209106,
|
| 1823 |
+
"rewards/accuracies": 0.65625,
|
| 1824 |
+
"rewards/chosen": 0.061330053955316544,
|
| 1825 |
+
"rewards/margins": 0.847295880317688,
|
| 1826 |
+
"rewards/rejected": -0.7859657406806946,
|
| 1827 |
+
"step": 101,
|
| 1828 |
+
"train_speed(iter/s)": 0.005498
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 0.08229747106729533,
|
| 1832 |
+
"grad_norm": 2.8322761058807373,
|
| 1833 |
+
"learning_rate": 0.00019943059113554097,
|
| 1834 |
+
"logits/chosen": -0.6899937987327576,
|
| 1835 |
+
"logits/rejected": -0.6942430734634399,
|
| 1836 |
+
"logps/chosen": -8.74088191986084,
|
| 1837 |
+
"logps/rejected": -20.522974014282227,
|
| 1838 |
+
"loss": 0.8782400488853455,
|
| 1839 |
+
"memory(GiB)": 46.83,
|
| 1840 |
+
"nll_loss": 0.3525587320327759,
|
| 1841 |
+
"rewards/accuracies": 0.8125,
|
| 1842 |
+
"rewards/chosen": 0.05973733961582184,
|
| 1843 |
+
"rewards/margins": 0.882931113243103,
|
| 1844 |
+
"rewards/rejected": -0.8231937289237976,
|
| 1845 |
+
"step": 102,
|
| 1846 |
+
"train_speed(iter/s)": 0.005498
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.08310430901893548,
|
| 1850 |
+
"grad_norm": 2.8728346824645996,
|
| 1851 |
+
"learning_rate": 0.00019940179358554424,
|
| 1852 |
+
"logits/chosen": -0.6615633964538574,
|
| 1853 |
+
"logits/rejected": -0.6650048494338989,
|
| 1854 |
+
"logps/chosen": -8.035341262817383,
|
| 1855 |
+
"logps/rejected": -13.233177185058594,
|
| 1856 |
+
"loss": 0.95281982421875,
|
| 1857 |
+
"memory(GiB)": 46.83,
|
| 1858 |
+
"nll_loss": 0.39267757534980774,
|
| 1859 |
+
"rewards/accuracies": 0.71875,
|
| 1860 |
+
"rewards/chosen": 0.12954209744930267,
|
| 1861 |
+
"rewards/margins": 0.4833866357803345,
|
| 1862 |
+
"rewards/rejected": -0.3538445234298706,
|
| 1863 |
+
"step": 103,
|
| 1864 |
+
"train_speed(iter/s)": 0.005496
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.08391114697057563,
|
| 1868 |
+
"grad_norm": 2.8015527725219727,
|
| 1869 |
+
"learning_rate": 0.00019937228786023542,
|
| 1870 |
+
"logits/chosen": -0.6321871280670166,
|
| 1871 |
+
"logits/rejected": -0.6367396116256714,
|
| 1872 |
+
"logps/chosen": -9.781261444091797,
|
| 1873 |
+
"logps/rejected": -18.408084869384766,
|
| 1874 |
+
"loss": 0.9027690887451172,
|
| 1875 |
+
"memory(GiB)": 46.83,
|
| 1876 |
+
"nll_loss": 0.4244646430015564,
|
| 1877 |
+
"rewards/accuracies": 0.84375,
|
| 1878 |
+
"rewards/chosen": 0.18980520963668823,
|
| 1879 |
+
"rewards/margins": 0.7434633374214172,
|
| 1880 |
+
"rewards/rejected": -0.5536580681800842,
|
| 1881 |
+
"step": 104,
|
| 1882 |
+
"train_speed(iter/s)": 0.005498
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.08471798492221577,
|
| 1886 |
+
"grad_norm": 3.581929922103882,
|
| 1887 |
+
"learning_rate": 0.00019934207416982415,
|
| 1888 |
+
"logits/chosen": -0.6718648076057434,
|
| 1889 |
+
"logits/rejected": -0.6737468242645264,
|
| 1890 |
+
"logps/chosen": -10.535398483276367,
|
| 1891 |
+
"logps/rejected": -15.517889022827148,
|
| 1892 |
+
"loss": 0.9467408657073975,
|
| 1893 |
+
"memory(GiB)": 46.83,
|
| 1894 |
+
"nll_loss": 0.4431319832801819,
|
| 1895 |
+
"rewards/accuracies": 0.84375,
|
| 1896 |
+
"rewards/chosen": 0.24101108312606812,
|
| 1897 |
+
"rewards/margins": 0.5848699808120728,
|
| 1898 |
+
"rewards/rejected": -0.34385889768600464,
|
| 1899 |
+
"step": 105,
|
| 1900 |
+
"train_speed(iter/s)": 0.005496
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.08552482287385593,
|
| 1904 |
+
"grad_norm": 2.7180511951446533,
|
| 1905 |
+
"learning_rate": 0.00019931115272956405,
|
| 1906 |
+
"logits/chosen": -0.6360980272293091,
|
| 1907 |
+
"logits/rejected": -0.6395881175994873,
|
| 1908 |
+
"logps/chosen": -5.562260627746582,
|
| 1909 |
+
"logps/rejected": -16.978702545166016,
|
| 1910 |
+
"loss": 0.9048807621002197,
|
| 1911 |
+
"memory(GiB)": 46.83,
|
| 1912 |
+
"nll_loss": 0.3790769577026367,
|
| 1913 |
+
"rewards/accuracies": 0.71875,
|
| 1914 |
+
"rewards/chosen": 0.1927444040775299,
|
| 1915 |
+
"rewards/margins": 0.6174324750900269,
|
| 1916 |
+
"rewards/rejected": -0.42468804121017456,
|
| 1917 |
+
"step": 106,
|
| 1918 |
+
"train_speed(iter/s)": 0.005495
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.08633166082549608,
|
| 1922 |
+
"grad_norm": 2.7166309356689453,
|
| 1923 |
+
"learning_rate": 0.00019927952375975094,
|
| 1924 |
+
"logits/chosen": -0.6151095628738403,
|
| 1925 |
+
"logits/rejected": -0.6214062571525574,
|
| 1926 |
+
"logps/chosen": -8.704106330871582,
|
| 1927 |
+
"logps/rejected": -15.980009078979492,
|
| 1928 |
+
"loss": 0.9614640474319458,
|
| 1929 |
+
"memory(GiB)": 46.83,
|
| 1930 |
+
"nll_loss": 0.41800737380981445,
|
| 1931 |
+
"rewards/accuracies": 0.71875,
|
| 1932 |
+
"rewards/chosen": 0.22499363124370575,
|
| 1933 |
+
"rewards/margins": 0.5540027618408203,
|
| 1934 |
+
"rewards/rejected": -0.32900911569595337,
|
| 1935 |
+
"step": 107,
|
| 1936 |
+
"train_speed(iter/s)": 0.005495
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 0.08713849877713623,
|
| 1940 |
+
"grad_norm": 3.3143904209136963,
|
| 1941 |
+
"learning_rate": 0.00019924718748572136,
|
| 1942 |
+
"logits/chosen": -0.6602830290794373,
|
| 1943 |
+
"logits/rejected": -0.6590928435325623,
|
| 1944 |
+
"logps/chosen": -14.48482608795166,
|
| 1945 |
+
"logps/rejected": -13.489812850952148,
|
| 1946 |
+
"loss": 1.1007349491119385,
|
| 1947 |
+
"memory(GiB)": 46.83,
|
| 1948 |
+
"nll_loss": 0.38938888907432556,
|
| 1949 |
+
"rewards/accuracies": 0.5625,
|
| 1950 |
+
"rewards/chosen": -0.005462644621729851,
|
| 1951 |
+
"rewards/margins": 0.18356555700302124,
|
| 1952 |
+
"rewards/rejected": -0.18902820348739624,
|
| 1953 |
+
"step": 108,
|
| 1954 |
+
"train_speed(iter/s)": 0.005493
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.08794533672877639,
|
| 1958 |
+
"grad_norm": 3.3967649936676025,
|
| 1959 |
+
"learning_rate": 0.00019921414413785095,
|
| 1960 |
+
"logits/chosen": -0.6139819025993347,
|
| 1961 |
+
"logits/rejected": -0.6160064935684204,
|
| 1962 |
+
"logps/chosen": -10.462146759033203,
|
| 1963 |
+
"logps/rejected": -13.9368896484375,
|
| 1964 |
+
"loss": 1.1163440942764282,
|
| 1965 |
+
"memory(GiB)": 46.83,
|
| 1966 |
+
"nll_loss": 0.5643093585968018,
|
| 1967 |
+
"rewards/accuracies": 0.625,
|
| 1968 |
+
"rewards/chosen": 0.13518843054771423,
|
| 1969 |
+
"rewards/margins": 0.5490553379058838,
|
| 1970 |
+
"rewards/rejected": -0.41386690735816956,
|
| 1971 |
+
"step": 109,
|
| 1972 |
+
"train_speed(iter/s)": 0.005493
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 0.08875217468041653,
|
| 1976 |
+
"grad_norm": 2.125051498413086,
|
| 1977 |
+
"learning_rate": 0.00019918039395155278,
|
| 1978 |
+
"logits/chosen": -0.5581845641136169,
|
| 1979 |
+
"logits/rejected": -0.5609419345855713,
|
| 1980 |
+
"logps/chosen": -8.664159774780273,
|
| 1981 |
+
"logps/rejected": -16.96472930908203,
|
| 1982 |
+
"loss": 0.8410020470619202,
|
| 1983 |
+
"memory(GiB)": 46.83,
|
| 1984 |
+
"nll_loss": 0.3336779773235321,
|
| 1985 |
+
"rewards/accuracies": 0.75,
|
| 1986 |
+
"rewards/chosen": 0.24867044389247894,
|
| 1987 |
+
"rewards/margins": 0.6479009985923767,
|
| 1988 |
+
"rewards/rejected": -0.3992305099964142,
|
| 1989 |
+
"step": 110,
|
| 1990 |
+
"train_speed(iter/s)": 0.005494
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.08955901263205668,
|
| 1994 |
+
"grad_norm": 2.7272698879241943,
|
| 1995 |
+
"learning_rate": 0.00019914593716727574,
|
| 1996 |
+
"logits/chosen": -0.640306830406189,
|
| 1997 |
+
"logits/rejected": -0.6392278075218201,
|
| 1998 |
+
"logps/chosen": -10.320524215698242,
|
| 1999 |
+
"logps/rejected": -15.357645034790039,
|
| 2000 |
+
"loss": 0.9050614237785339,
|
| 2001 |
+
"memory(GiB)": 46.83,
|
| 2002 |
+
"nll_loss": 0.3744095265865326,
|
| 2003 |
+
"rewards/accuracies": 0.78125,
|
| 2004 |
+
"rewards/chosen": 0.228946715593338,
|
| 2005 |
+
"rewards/margins": 0.46916335821151733,
|
| 2006 |
+
"rewards/rejected": -0.24021662771701813,
|
| 2007 |
+
"step": 111,
|
| 2008 |
+
"train_speed(iter/s)": 0.005492
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.09036585058369682,
|
| 2012 |
+
"grad_norm": 2.552009344100952,
|
| 2013 |
+
"learning_rate": 0.0001991107740305027,
|
| 2014 |
+
"logits/chosen": -0.6064003109931946,
|
| 2015 |
+
"logits/rejected": -0.607029914855957,
|
| 2016 |
+
"logps/chosen": -10.730403900146484,
|
| 2017 |
+
"logps/rejected": -20.569438934326172,
|
| 2018 |
+
"loss": 0.9093858003616333,
|
| 2019 |
+
"memory(GiB)": 46.83,
|
| 2020 |
+
"nll_loss": 0.4844813346862793,
|
| 2021 |
+
"rewards/accuracies": 0.84375,
|
| 2022 |
+
"rewards/chosen": 0.320293128490448,
|
| 2023 |
+
"rewards/margins": 0.8029977083206177,
|
| 2024 |
+
"rewards/rejected": -0.48270460963249207,
|
| 2025 |
+
"step": 112,
|
| 2026 |
+
"train_speed(iter/s)": 0.005492
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.09117268853533698,
|
| 2030 |
+
"grad_norm": 2.875828742980957,
|
| 2031 |
+
"learning_rate": 0.00019907490479174902,
|
| 2032 |
+
"logits/chosen": -0.6465886831283569,
|
| 2033 |
+
"logits/rejected": -0.6483381986618042,
|
| 2034 |
+
"logps/chosen": -6.758854389190674,
|
| 2035 |
+
"logps/rejected": -20.76754379272461,
|
| 2036 |
+
"loss": 0.7010505199432373,
|
| 2037 |
+
"memory(GiB)": 46.83,
|
| 2038 |
+
"nll_loss": 0.29428812861442566,
|
| 2039 |
+
"rewards/accuracies": 0.875,
|
| 2040 |
+
"rewards/chosen": 0.3071742653846741,
|
| 2041 |
+
"rewards/margins": 1.1479661464691162,
|
| 2042 |
+
"rewards/rejected": -0.8407919406890869,
|
| 2043 |
+
"step": 113,
|
| 2044 |
+
"train_speed(iter/s)": 0.005491
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.09197952648697713,
|
| 2048 |
+
"grad_norm": 2.3188178539276123,
|
| 2049 |
+
"learning_rate": 0.00019903832970656038,
|
| 2050 |
+
"logits/chosen": -0.6288142800331116,
|
| 2051 |
+
"logits/rejected": -0.6315409541130066,
|
| 2052 |
+
"logps/chosen": -8.855584144592285,
|
| 2053 |
+
"logps/rejected": -20.421592712402344,
|
| 2054 |
+
"loss": 0.841801106929779,
|
| 2055 |
+
"memory(GiB)": 46.83,
|
| 2056 |
+
"nll_loss": 0.38135236501693726,
|
| 2057 |
+
"rewards/accuracies": 0.8125,
|
| 2058 |
+
"rewards/chosen": 0.19283419847488403,
|
| 2059 |
+
"rewards/margins": 0.8724038004875183,
|
| 2060 |
+
"rewards/rejected": -0.6795696020126343,
|
| 2061 |
+
"step": 114,
|
| 2062 |
+
"train_speed(iter/s)": 0.005492
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.09278636443861728,
|
| 2066 |
+
"grad_norm": 3.3676164150238037,
|
| 2067 |
+
"learning_rate": 0.00019900104903551133,
|
| 2068 |
+
"logits/chosen": -0.6541582345962524,
|
| 2069 |
+
"logits/rejected": -0.6580868363380432,
|
| 2070 |
+
"logps/chosen": -8.595276832580566,
|
| 2071 |
+
"logps/rejected": -24.14297866821289,
|
| 2072 |
+
"loss": 0.7960852980613708,
|
| 2073 |
+
"memory(GiB)": 46.83,
|
| 2074 |
+
"nll_loss": 0.4363185465335846,
|
| 2075 |
+
"rewards/accuracies": 0.875,
|
| 2076 |
+
"rewards/chosen": 0.10038560628890991,
|
| 2077 |
+
"rewards/margins": 1.4183127880096436,
|
| 2078 |
+
"rewards/rejected": -1.3179271221160889,
|
| 2079 |
+
"step": 115,
|
| 2080 |
+
"train_speed(iter/s)": 0.005493
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 0.09359320239025744,
|
| 2084 |
+
"grad_norm": 3.721660614013672,
|
| 2085 |
+
"learning_rate": 0.00019896306304420324,
|
| 2086 |
+
"logits/chosen": -0.6533216834068298,
|
| 2087 |
+
"logits/rejected": -0.6518760919570923,
|
| 2088 |
+
"logps/chosen": -9.710081100463867,
|
| 2089 |
+
"logps/rejected": -20.333324432373047,
|
| 2090 |
+
"loss": 0.842086911201477,
|
| 2091 |
+
"memory(GiB)": 46.83,
|
| 2092 |
+
"nll_loss": 0.38858363032341003,
|
| 2093 |
+
"rewards/accuracies": 0.75,
|
| 2094 |
+
"rewards/chosen": 0.1398862600326538,
|
| 2095 |
+
"rewards/margins": 1.0419015884399414,
|
| 2096 |
+
"rewards/rejected": -0.9020152688026428,
|
| 2097 |
+
"step": 116,
|
| 2098 |
+
"train_speed(iter/s)": 0.005494
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.09440004034189758,
|
| 2102 |
+
"grad_norm": 4.8206281661987305,
|
| 2103 |
+
"learning_rate": 0.0001989243720032624,
|
| 2104 |
+
"logits/chosen": -0.7102867364883423,
|
| 2105 |
+
"logits/rejected": -0.712573230266571,
|
| 2106 |
+
"logps/chosen": -14.843620300292969,
|
| 2107 |
+
"logps/rejected": -23.658788681030273,
|
| 2108 |
+
"loss": 1.1279860734939575,
|
| 2109 |
+
"memory(GiB)": 46.83,
|
| 2110 |
+
"nll_loss": 0.5651426315307617,
|
| 2111 |
+
"rewards/accuracies": 0.78125,
|
| 2112 |
+
"rewards/chosen": -0.09467095881700516,
|
| 2113 |
+
"rewards/margins": 0.9923723340034485,
|
| 2114 |
+
"rewards/rejected": -1.087043285369873,
|
| 2115 |
+
"step": 117,
|
| 2116 |
+
"train_speed(iter/s)": 0.005494
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.09520687829353773,
|
| 2120 |
+
"grad_norm": 3.194643020629883,
|
| 2121 |
+
"learning_rate": 0.00019888497618833815,
|
| 2122 |
+
"logits/chosen": -0.6824992895126343,
|
| 2123 |
+
"logits/rejected": -0.6856632232666016,
|
| 2124 |
+
"logps/chosen": -7.8865065574646,
|
| 2125 |
+
"logps/rejected": -22.186447143554688,
|
| 2126 |
+
"loss": 0.8166104555130005,
|
| 2127 |
+
"memory(GiB)": 46.83,
|
| 2128 |
+
"nll_loss": 0.3872619867324829,
|
| 2129 |
+
"rewards/accuracies": 0.8125,
|
| 2130 |
+
"rewards/chosen": 0.1935729682445526,
|
| 2131 |
+
"rewards/margins": 1.2844107151031494,
|
| 2132 |
+
"rewards/rejected": -1.090837836265564,
|
| 2133 |
+
"step": 118,
|
| 2134 |
+
"train_speed(iter/s)": 0.005494
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.09601371624517788,
|
| 2138 |
+
"grad_norm": 5.332725524902344,
|
| 2139 |
+
"learning_rate": 0.00019884487588010092,
|
| 2140 |
+
"logits/chosen": -0.6963211894035339,
|
| 2141 |
+
"logits/rejected": -0.695375382900238,
|
| 2142 |
+
"logps/chosen": -12.522785186767578,
|
| 2143 |
+
"logps/rejected": -21.899364471435547,
|
| 2144 |
+
"loss": 1.200064778327942,
|
| 2145 |
+
"memory(GiB)": 46.83,
|
| 2146 |
+
"nll_loss": 0.5619924068450928,
|
| 2147 |
+
"rewards/accuracies": 0.625,
|
| 2148 |
+
"rewards/chosen": -0.2353927493095398,
|
| 2149 |
+
"rewards/margins": 0.7032212018966675,
|
| 2150 |
+
"rewards/rejected": -0.9386140704154968,
|
| 2151 |
+
"step": 119,
|
| 2152 |
+
"train_speed(iter/s)": 0.005493
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.09682055419681804,
|
| 2156 |
+
"grad_norm": 3.189750909805298,
|
| 2157 |
+
"learning_rate": 0.0001988040713642402,
|
| 2158 |
+
"logits/chosen": -0.7008529305458069,
|
| 2159 |
+
"logits/rejected": -0.7020596265792847,
|
| 2160 |
+
"logps/chosen": -7.352071285247803,
|
| 2161 |
+
"logps/rejected": -26.472599029541016,
|
| 2162 |
+
"loss": 0.6796355843544006,
|
| 2163 |
+
"memory(GiB)": 46.83,
|
| 2164 |
+
"nll_loss": 0.30471816658973694,
|
| 2165 |
+
"rewards/accuracies": 0.84375,
|
| 2166 |
+
"rewards/chosen": 0.0883897989988327,
|
| 2167 |
+
"rewards/margins": 1.43672513961792,
|
| 2168 |
+
"rewards/rejected": -1.3483351469039917,
|
| 2169 |
+
"step": 120,
|
| 2170 |
+
"train_speed(iter/s)": 0.005494
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.09762739214845818,
|
| 2174 |
+
"grad_norm": 3.5756847858428955,
|
| 2175 |
+
"learning_rate": 0.00019876256293146257,
|
| 2176 |
+
"logits/chosen": -0.6855241060256958,
|
| 2177 |
+
"logits/rejected": -0.6870216131210327,
|
| 2178 |
+
"logps/chosen": -9.13197135925293,
|
| 2179 |
+
"logps/rejected": -25.256893157958984,
|
| 2180 |
+
"loss": 0.9965449571609497,
|
| 2181 |
+
"memory(GiB)": 46.83,
|
| 2182 |
+
"nll_loss": 0.5001046061515808,
|
| 2183 |
+
"rewards/accuracies": 0.78125,
|
| 2184 |
+
"rewards/chosen": 0.036015622317790985,
|
| 2185 |
+
"rewards/margins": 1.1710847616195679,
|
| 2186 |
+
"rewards/rejected": -1.1350692510604858,
|
| 2187 |
+
"step": 121,
|
| 2188 |
+
"train_speed(iter/s)": 0.005494
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 0.09843423010009833,
|
| 2192 |
+
"grad_norm": 2.60425066947937,
|
| 2193 |
+
"learning_rate": 0.0001987203508774895,
|
| 2194 |
+
"logits/chosen": -0.7040928602218628,
|
| 2195 |
+
"logits/rejected": -0.7042312026023865,
|
| 2196 |
+
"logps/chosen": -8.471722602844238,
|
| 2197 |
+
"logps/rejected": -26.018938064575195,
|
| 2198 |
+
"loss": 0.8459987640380859,
|
| 2199 |
+
"memory(GiB)": 46.83,
|
| 2200 |
+
"nll_loss": 0.4736276865005493,
|
| 2201 |
+
"rewards/accuracies": 0.875,
|
| 2202 |
+
"rewards/chosen": 0.24268555641174316,
|
| 2203 |
+
"rewards/margins": 1.318163514137268,
|
| 2204 |
+
"rewards/rejected": -1.075477957725525,
|
| 2205 |
+
"step": 122,
|
| 2206 |
+
"train_speed(iter/s)": 0.005493
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.09924106805173849,
|
| 2210 |
+
"grad_norm": 4.30864143371582,
|
| 2211 |
+
"learning_rate": 0.0001986774355030553,
|
| 2212 |
+
"logits/chosen": -0.6747640371322632,
|
| 2213 |
+
"logits/rejected": -0.677899956703186,
|
| 2214 |
+
"logps/chosen": -14.706789016723633,
|
| 2215 |
+
"logps/rejected": -16.56774139404297,
|
| 2216 |
+
"loss": 1.2364513874053955,
|
| 2217 |
+
"memory(GiB)": 46.83,
|
| 2218 |
+
"nll_loss": 0.6511675119400024,
|
| 2219 |
+
"rewards/accuracies": 0.71875,
|
| 2220 |
+
"rewards/chosen": 0.08759675920009613,
|
| 2221 |
+
"rewards/margins": 0.6008309721946716,
|
| 2222 |
+
"rewards/rejected": -0.5132341384887695,
|
| 2223 |
+
"step": 123,
|
| 2224 |
+
"train_speed(iter/s)": 0.005492
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.10004790600337864,
|
| 2228 |
+
"grad_norm": 6.867242813110352,
|
| 2229 |
+
"learning_rate": 0.00019863381711390508,
|
| 2230 |
+
"logits/chosen": -0.6869674324989319,
|
| 2231 |
+
"logits/rejected": -0.6902846693992615,
|
| 2232 |
+
"logps/chosen": -7.829675197601318,
|
| 2233 |
+
"logps/rejected": -20.095809936523438,
|
| 2234 |
+
"loss": 1.043180227279663,
|
| 2235 |
+
"memory(GiB)": 46.83,
|
| 2236 |
+
"nll_loss": 0.530769944190979,
|
| 2237 |
+
"rewards/accuracies": 0.71875,
|
| 2238 |
+
"rewards/chosen": 0.11925126612186432,
|
| 2239 |
+
"rewards/margins": 0.697571337223053,
|
| 2240 |
+
"rewards/rejected": -0.5783200860023499,
|
| 2241 |
+
"step": 124,
|
| 2242 |
+
"train_speed(iter/s)": 0.005491
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.10085474395501878,
|
| 2246 |
+
"grad_norm": 3.041438579559326,
|
| 2247 |
+
"learning_rate": 0.0001985894960207925,
|
| 2248 |
+
"logits/chosen": -0.6505750417709351,
|
| 2249 |
+
"logits/rejected": -0.6511614918708801,
|
| 2250 |
+
"logps/chosen": -9.314833641052246,
|
| 2251 |
+
"logps/rejected": -27.15659523010254,
|
| 2252 |
+
"loss": 0.7252060174942017,
|
| 2253 |
+
"memory(GiB)": 46.83,
|
| 2254 |
+
"nll_loss": 0.39295029640197754,
|
| 2255 |
+
"rewards/accuracies": 0.96875,
|
| 2256 |
+
"rewards/chosen": 0.24510648846626282,
|
| 2257 |
+
"rewards/margins": 1.2805438041687012,
|
| 2258 |
+
"rewards/rejected": -1.0354374647140503,
|
| 2259 |
+
"step": 125,
|
| 2260 |
+
"train_speed(iter/s)": 0.005491
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.10166158190665893,
|
| 2264 |
+
"grad_norm": 2.509321689605713,
|
| 2265 |
+
"learning_rate": 0.00019854447253947743,
|
| 2266 |
+
"logits/chosen": -0.6497823596000671,
|
| 2267 |
+
"logits/rejected": -0.6492441892623901,
|
| 2268 |
+
"logps/chosen": -8.02572250366211,
|
| 2269 |
+
"logps/rejected": -20.3783016204834,
|
| 2270 |
+
"loss": 0.7982934713363647,
|
| 2271 |
+
"memory(GiB)": 46.83,
|
| 2272 |
+
"nll_loss": 0.33212146162986755,
|
| 2273 |
+
"rewards/accuracies": 0.78125,
|
| 2274 |
+
"rewards/chosen": 0.17764505743980408,
|
| 2275 |
+
"rewards/margins": 0.8625535368919373,
|
| 2276 |
+
"rewards/rejected": -0.6849085092544556,
|
| 2277 |
+
"step": 126,
|
| 2278 |
+
"train_speed(iter/s)": 0.005489
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.10246841985829909,
|
| 2282 |
+
"grad_norm": 3.6529700756073,
|
| 2283 |
+
"learning_rate": 0.0001984987469907239,
|
| 2284 |
+
"logits/chosen": -0.6466098427772522,
|
| 2285 |
+
"logits/rejected": -0.6480077505111694,
|
| 2286 |
+
"logps/chosen": -9.548334121704102,
|
| 2287 |
+
"logps/rejected": -16.65062141418457,
|
| 2288 |
+
"loss": 0.914322555065155,
|
| 2289 |
+
"memory(GiB)": 46.83,
|
| 2290 |
+
"nll_loss": 0.4001502990722656,
|
| 2291 |
+
"rewards/accuracies": 0.75,
|
| 2292 |
+
"rewards/chosen": 0.1684979647397995,
|
| 2293 |
+
"rewards/margins": 0.6629449129104614,
|
| 2294 |
+
"rewards/rejected": -0.49444690346717834,
|
| 2295 |
+
"step": 127,
|
| 2296 |
+
"train_speed(iter/s)": 0.005491
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 0.10327525780993924,
|
| 2300 |
+
"grad_norm": 3.3696913719177246,
|
| 2301 |
+
"learning_rate": 0.00019845231970029773,
|
| 2302 |
+
"logits/chosen": -0.679688572883606,
|
| 2303 |
+
"logits/rejected": -0.6820145845413208,
|
| 2304 |
+
"logps/chosen": -10.874924659729004,
|
| 2305 |
+
"logps/rejected": -24.826372146606445,
|
| 2306 |
+
"loss": 0.9800686240196228,
|
| 2307 |
+
"memory(GiB)": 46.83,
|
| 2308 |
+
"nll_loss": 0.4543137848377228,
|
| 2309 |
+
"rewards/accuracies": 0.8125,
|
| 2310 |
+
"rewards/chosen": -0.048351842910051346,
|
| 2311 |
+
"rewards/margins": 0.8677961230278015,
|
| 2312 |
+
"rewards/rejected": -0.9161479473114014,
|
| 2313 |
+
"step": 128,
|
| 2314 |
+
"train_speed(iter/s)": 0.00549
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.10408209576157938,
|
| 2318 |
+
"grad_norm": 2.5419607162475586,
|
| 2319 |
+
"learning_rate": 0.00019840519099896414,
|
| 2320 |
+
"logits/chosen": -0.6618849039077759,
|
| 2321 |
+
"logits/rejected": -0.6671789288520813,
|
| 2322 |
+
"logps/chosen": -8.426337242126465,
|
| 2323 |
+
"logps/rejected": -25.797517776489258,
|
| 2324 |
+
"loss": 0.6974806189537048,
|
| 2325 |
+
"memory(GiB)": 46.83,
|
| 2326 |
+
"nll_loss": 0.2836017608642578,
|
| 2327 |
+
"rewards/accuracies": 0.84375,
|
| 2328 |
+
"rewards/chosen": 0.18303632736206055,
|
| 2329 |
+
"rewards/margins": 1.1003562211990356,
|
| 2330 |
+
"rewards/rejected": -0.9173198938369751,
|
| 2331 |
+
"step": 129,
|
| 2332 |
+
"train_speed(iter/s)": 0.005491
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.10488893371321953,
|
| 2336 |
+
"grad_norm": 3.4203388690948486,
|
| 2337 |
+
"learning_rate": 0.00019835736122248557,
|
| 2338 |
+
"logits/chosen": -0.6606203317642212,
|
| 2339 |
+
"logits/rejected": -0.6571654677391052,
|
| 2340 |
+
"logps/chosen": -14.509614944458008,
|
| 2341 |
+
"logps/rejected": -15.079629898071289,
|
| 2342 |
+
"loss": 1.101165771484375,
|
| 2343 |
+
"memory(GiB)": 46.83,
|
| 2344 |
+
"nll_loss": 0.5219094157218933,
|
| 2345 |
+
"rewards/accuracies": 0.75,
|
| 2346 |
+
"rewards/chosen": 0.07395613193511963,
|
| 2347 |
+
"rewards/margins": 0.5427431464195251,
|
| 2348 |
+
"rewards/rejected": -0.46878698468208313,
|
| 2349 |
+
"step": 130,
|
| 2350 |
+
"train_speed(iter/s)": 0.005492
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 0.10569577166485969,
|
| 2354 |
+
"grad_norm": 2.7750000953674316,
|
| 2355 |
+
"learning_rate": 0.00019830883071161908,
|
| 2356 |
+
"logits/chosen": -0.7152451276779175,
|
| 2357 |
+
"logits/rejected": -0.7131951451301575,
|
| 2358 |
+
"logps/chosen": -9.166630744934082,
|
| 2359 |
+
"logps/rejected": -19.46819305419922,
|
| 2360 |
+
"loss": 0.7933439016342163,
|
| 2361 |
+
"memory(GiB)": 46.83,
|
| 2362 |
+
"nll_loss": 0.3753884732723236,
|
| 2363 |
+
"rewards/accuracies": 0.78125,
|
| 2364 |
+
"rewards/chosen": 0.26774072647094727,
|
| 2365 |
+
"rewards/margins": 1.0882644653320312,
|
| 2366 |
+
"rewards/rejected": -0.8205236196517944,
|
| 2367 |
+
"step": 131,
|
| 2368 |
+
"train_speed(iter/s)": 0.005492
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.10650260961649984,
|
| 2372 |
+
"grad_norm": 3.1810927391052246,
|
| 2373 |
+
"learning_rate": 0.00019825959981211407,
|
| 2374 |
+
"logits/chosen": -0.714146614074707,
|
| 2375 |
+
"logits/rejected": -0.7095440626144409,
|
| 2376 |
+
"logps/chosen": -11.571531295776367,
|
| 2377 |
+
"logps/rejected": -14.685908317565918,
|
| 2378 |
+
"loss": 0.8970522880554199,
|
| 2379 |
+
"memory(GiB)": 46.83,
|
| 2380 |
+
"nll_loss": 0.38512641191482544,
|
| 2381 |
+
"rewards/accuracies": 0.75,
|
| 2382 |
+
"rewards/chosen": 0.09674864262342453,
|
| 2383 |
+
"rewards/margins": 0.5926439762115479,
|
| 2384 |
+
"rewards/rejected": -0.4958953261375427,
|
| 2385 |
+
"step": 132,
|
| 2386 |
+
"train_speed(iter/s)": 0.005491
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.10730944756813998,
|
| 2390 |
+
"grad_norm": 2.839996099472046,
|
| 2391 |
+
"learning_rate": 0.00019820966887470974,
|
| 2392 |
+
"logits/chosen": -0.7237828373908997,
|
| 2393 |
+
"logits/rejected": -0.7276051640510559,
|
| 2394 |
+
"logps/chosen": -7.241865158081055,
|
| 2395 |
+
"logps/rejected": -21.9840145111084,
|
| 2396 |
+
"loss": 0.7037227153778076,
|
| 2397 |
+
"memory(GiB)": 46.83,
|
| 2398 |
+
"nll_loss": 0.29994940757751465,
|
| 2399 |
+
"rewards/accuracies": 0.84375,
|
| 2400 |
+
"rewards/chosen": 0.18205267190933228,
|
| 2401 |
+
"rewards/margins": 1.1148442029953003,
|
| 2402 |
+
"rewards/rejected": -0.9327914714813232,
|
| 2403 |
+
"step": 133,
|
| 2404 |
+
"train_speed(iter/s)": 0.005492
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.10811628551978014,
|
| 2408 |
+
"grad_norm": 2.8896915912628174,
|
| 2409 |
+
"learning_rate": 0.0001981590382551327,
|
| 2410 |
+
"logits/chosen": -0.796456515789032,
|
| 2411 |
+
"logits/rejected": -0.7955406308174133,
|
| 2412 |
+
"logps/chosen": -11.131935119628906,
|
| 2413 |
+
"logps/rejected": -19.525293350219727,
|
| 2414 |
+
"loss": 0.8993673920631409,
|
| 2415 |
+
"memory(GiB)": 46.83,
|
| 2416 |
+
"nll_loss": 0.4546453356742859,
|
| 2417 |
+
"rewards/accuracies": 0.8125,
|
| 2418 |
+
"rewards/chosen": 0.21719151735305786,
|
| 2419 |
+
"rewards/margins": 1.081808090209961,
|
| 2420 |
+
"rewards/rejected": -0.8646165728569031,
|
| 2421 |
+
"step": 134,
|
| 2422 |
+
"train_speed(iter/s)": 0.00549
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.10892312347142029,
|
| 2426 |
+
"grad_norm": 3.5537540912628174,
|
| 2427 |
+
"learning_rate": 0.00019810770831409425,
|
| 2428 |
+
"logits/chosen": -0.7755017876625061,
|
| 2429 |
+
"logits/rejected": -0.7809926271438599,
|
| 2430 |
+
"logps/chosen": -7.181600093841553,
|
| 2431 |
+
"logps/rejected": -19.379724502563477,
|
| 2432 |
+
"loss": 0.7135071754455566,
|
| 2433 |
+
"memory(GiB)": 46.83,
|
| 2434 |
+
"nll_loss": 0.32976531982421875,
|
| 2435 |
+
"rewards/accuracies": 0.78125,
|
| 2436 |
+
"rewards/chosen": 0.26019349694252014,
|
| 2437 |
+
"rewards/margins": 1.110938310623169,
|
| 2438 |
+
"rewards/rejected": -0.850744903087616,
|
| 2439 |
+
"step": 135,
|
| 2440 |
+
"train_speed(iter/s)": 0.005491
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 0.10972996142306043,
|
| 2444 |
+
"grad_norm": 3.5955300331115723,
|
| 2445 |
+
"learning_rate": 0.00019805567941728796,
|
| 2446 |
+
"logits/chosen": -0.799392819404602,
|
| 2447 |
+
"logits/rejected": -0.8029282689094543,
|
| 2448 |
+
"logps/chosen": -9.87437629699707,
|
| 2449 |
+
"logps/rejected": -23.02663230895996,
|
| 2450 |
+
"loss": 0.9438092708587646,
|
| 2451 |
+
"memory(GiB)": 46.83,
|
| 2452 |
+
"nll_loss": 0.5196709036827087,
|
| 2453 |
+
"rewards/accuracies": 0.78125,
|
| 2454 |
+
"rewards/chosen": 0.07922995835542679,
|
| 2455 |
+
"rewards/margins": 1.070724606513977,
|
| 2456 |
+
"rewards/rejected": -0.9914946556091309,
|
| 2457 |
+
"step": 136,
|
| 2458 |
+
"train_speed(iter/s)": 0.005492
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.11053679937470058,
|
| 2462 |
+
"grad_norm": 3.7304859161376953,
|
| 2463 |
+
"learning_rate": 0.00019800295193538704,
|
| 2464 |
+
"logits/chosen": -0.8160181641578674,
|
| 2465 |
+
"logits/rejected": -0.8219972848892212,
|
| 2466 |
+
"logps/chosen": -7.596125602722168,
|
| 2467 |
+
"logps/rejected": -20.225994110107422,
|
| 2468 |
+
"loss": 0.8485299348831177,
|
| 2469 |
+
"memory(GiB)": 46.83,
|
| 2470 |
+
"nll_loss": 0.43994730710983276,
|
| 2471 |
+
"rewards/accuracies": 0.90625,
|
| 2472 |
+
"rewards/chosen": 0.26360729336738586,
|
| 2473 |
+
"rewards/margins": 0.9658780694007874,
|
| 2474 |
+
"rewards/rejected": -0.7022708058357239,
|
| 2475 |
+
"step": 137,
|
| 2476 |
+
"train_speed(iter/s)": 0.005491
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.11134363732634074,
|
| 2480 |
+
"grad_norm": 4.919328689575195,
|
| 2481 |
+
"learning_rate": 0.00019794952624404167,
|
| 2482 |
+
"logits/chosen": -0.7996560335159302,
|
| 2483 |
+
"logits/rejected": -0.801249623298645,
|
| 2484 |
+
"logps/chosen": -9.867749214172363,
|
| 2485 |
+
"logps/rejected": -22.316190719604492,
|
| 2486 |
+
"loss": 0.8283427357673645,
|
| 2487 |
+
"memory(GiB)": 46.83,
|
| 2488 |
+
"nll_loss": 0.4471050798892975,
|
| 2489 |
+
"rewards/accuracies": 0.84375,
|
| 2490 |
+
"rewards/chosen": 0.13889475166797638,
|
| 2491 |
+
"rewards/margins": 1.224007248878479,
|
| 2492 |
+
"rewards/rejected": -1.085112452507019,
|
| 2493 |
+
"step": 138,
|
| 2494 |
+
"train_speed(iter/s)": 0.005491
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.11215047527798089,
|
| 2498 |
+
"grad_norm": 2.821730136871338,
|
| 2499 |
+
"learning_rate": 0.0001978954027238763,
|
| 2500 |
+
"logits/chosen": -0.818955659866333,
|
| 2501 |
+
"logits/rejected": -0.8183013796806335,
|
| 2502 |
+
"logps/chosen": -6.504258632659912,
|
| 2503 |
+
"logps/rejected": -14.602479934692383,
|
| 2504 |
+
"loss": 0.7571634650230408,
|
| 2505 |
+
"memory(GiB)": 46.83,
|
| 2506 |
+
"nll_loss": 0.29162290692329407,
|
| 2507 |
+
"rewards/accuracies": 0.71875,
|
| 2508 |
+
"rewards/chosen": 0.1784059852361679,
|
| 2509 |
+
"rewards/margins": 0.8354285955429077,
|
| 2510 |
+
"rewards/rejected": -0.6570225954055786,
|
| 2511 |
+
"step": 139,
|
| 2512 |
+
"train_speed(iter/s)": 0.005491
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.11295731322962103,
|
| 2516 |
+
"grad_norm": 2.481306552886963,
|
| 2517 |
+
"learning_rate": 0.000197840581760487,
|
| 2518 |
+
"logits/chosen": -0.8251004219055176,
|
| 2519 |
+
"logits/rejected": -0.8243045210838318,
|
| 2520 |
+
"logps/chosen": -13.736132621765137,
|
| 2521 |
+
"logps/rejected": -17.749977111816406,
|
| 2522 |
+
"loss": 0.8052379488945007,
|
| 2523 |
+
"memory(GiB)": 46.83,
|
| 2524 |
+
"nll_loss": 0.4064967930316925,
|
| 2525 |
+
"rewards/accuracies": 0.8125,
|
| 2526 |
+
"rewards/chosen": 0.3237007260322571,
|
| 2527 |
+
"rewards/margins": 1.295701026916504,
|
| 2528 |
+
"rewards/rejected": -0.9720003008842468,
|
| 2529 |
+
"step": 140,
|
| 2530 |
+
"train_speed(iter/s)": 0.005492
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.1137641511812612,
|
| 2534 |
+
"grad_norm": 2.695176601409912,
|
| 2535 |
+
"learning_rate": 0.00019778506374443873,
|
| 2536 |
+
"logits/chosen": -0.7545077800750732,
|
| 2537 |
+
"logits/rejected": -0.7596877813339233,
|
| 2538 |
+
"logps/chosen": -5.598147869110107,
|
| 2539 |
+
"logps/rejected": -26.760784149169922,
|
| 2540 |
+
"loss": 0.6858201026916504,
|
| 2541 |
+
"memory(GiB)": 46.83,
|
| 2542 |
+
"nll_loss": 0.3311798572540283,
|
| 2543 |
+
"rewards/accuracies": 0.84375,
|
| 2544 |
+
"rewards/chosen": 0.19224056601524353,
|
| 2545 |
+
"rewards/margins": 1.6802706718444824,
|
| 2546 |
+
"rewards/rejected": -1.488029956817627,
|
| 2547 |
+
"step": 141,
|
| 2548 |
+
"train_speed(iter/s)": 0.005493
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.11457098913290134,
|
| 2552 |
+
"grad_norm": 2.5371341705322266,
|
| 2553 |
+
"learning_rate": 0.0001977288490712624,
|
| 2554 |
+
"logits/chosen": -0.8187113404273987,
|
| 2555 |
+
"logits/rejected": -0.8256463408470154,
|
| 2556 |
+
"logps/chosen": -9.480502128601074,
|
| 2557 |
+
"logps/rejected": -26.96666717529297,
|
| 2558 |
+
"loss": 0.6809464693069458,
|
| 2559 |
+
"memory(GiB)": 46.83,
|
| 2560 |
+
"nll_loss": 0.3261586129665375,
|
| 2561 |
+
"rewards/accuracies": 0.90625,
|
| 2562 |
+
"rewards/chosen": 0.2214924842119217,
|
| 2563 |
+
"rewards/margins": 1.2070807218551636,
|
| 2564 |
+
"rewards/rejected": -0.9855883121490479,
|
| 2565 |
+
"step": 142,
|
| 2566 |
+
"train_speed(iter/s)": 0.005493
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 0.11537782708454149,
|
| 2570 |
+
"grad_norm": 3.33156681060791,
|
| 2571 |
+
"learning_rate": 0.00019767193814145226,
|
| 2572 |
+
"logits/chosen": -0.8148999214172363,
|
| 2573 |
+
"logits/rejected": -0.8154037594795227,
|
| 2574 |
+
"logps/chosen": -16.6685848236084,
|
| 2575 |
+
"logps/rejected": -16.219863891601562,
|
| 2576 |
+
"loss": 1.1351004838943481,
|
| 2577 |
+
"memory(GiB)": 46.83,
|
| 2578 |
+
"nll_loss": 0.6261444091796875,
|
| 2579 |
+
"rewards/accuracies": 0.78125,
|
| 2580 |
+
"rewards/chosen": 0.06986582279205322,
|
| 2581 |
+
"rewards/margins": 0.7546349763870239,
|
| 2582 |
+
"rewards/rejected": -0.6847692728042603,
|
| 2583 |
+
"step": 143,
|
| 2584 |
+
"train_speed(iter/s)": 0.005493
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.11618466503618163,
|
| 2588 |
+
"grad_norm": 2.587169647216797,
|
| 2589 |
+
"learning_rate": 0.00019761433136046295,
|
| 2590 |
+
"logits/chosen": -0.7679372429847717,
|
| 2591 |
+
"logits/rejected": -0.7711334824562073,
|
| 2592 |
+
"logps/chosen": -6.84797477722168,
|
| 2593 |
+
"logps/rejected": -25.882503509521484,
|
| 2594 |
+
"loss": 0.6460772752761841,
|
| 2595 |
+
"memory(GiB)": 46.83,
|
| 2596 |
+
"nll_loss": 0.2581600844860077,
|
| 2597 |
+
"rewards/accuracies": 0.78125,
|
| 2598 |
+
"rewards/chosen": 0.25230705738067627,
|
| 2599 |
+
"rewards/margins": 1.5044541358947754,
|
| 2600 |
+
"rewards/rejected": -1.2521469593048096,
|
| 2601 |
+
"step": 144,
|
| 2602 |
+
"train_speed(iter/s)": 0.005494
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.1169915029878218,
|
| 2606 |
+
"grad_norm": 3.816577911376953,
|
| 2607 |
+
"learning_rate": 0.00019755602913870655,
|
| 2608 |
+
"logits/chosen": -0.7486031651496887,
|
| 2609 |
+
"logits/rejected": -0.7545804977416992,
|
| 2610 |
+
"logps/chosen": -11.378320693969727,
|
| 2611 |
+
"logps/rejected": -23.90165901184082,
|
| 2612 |
+
"loss": 1.0280418395996094,
|
| 2613 |
+
"memory(GiB)": 46.83,
|
| 2614 |
+
"nll_loss": 0.5588870644569397,
|
| 2615 |
+
"rewards/accuracies": 0.84375,
|
| 2616 |
+
"rewards/chosen": 0.10084515064954758,
|
| 2617 |
+
"rewards/margins": 1.0778629779815674,
|
| 2618 |
+
"rewards/rejected": -0.9770178198814392,
|
| 2619 |
+
"step": 145,
|
| 2620 |
+
"train_speed(iter/s)": 0.005494
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.11779834093946194,
|
| 2624 |
+
"grad_norm": 3.1183438301086426,
|
| 2625 |
+
"learning_rate": 0.00019749703189154971,
|
| 2626 |
+
"logits/chosen": -0.7663331031799316,
|
| 2627 |
+
"logits/rejected": -0.7778096199035645,
|
| 2628 |
+
"logps/chosen": -7.461820602416992,
|
| 2629 |
+
"logps/rejected": -26.446916580200195,
|
| 2630 |
+
"loss": 0.6639739274978638,
|
| 2631 |
+
"memory(GiB)": 46.83,
|
| 2632 |
+
"nll_loss": 0.28250494599342346,
|
| 2633 |
+
"rewards/accuracies": 0.78125,
|
| 2634 |
+
"rewards/chosen": 0.04964678734540939,
|
| 2635 |
+
"rewards/margins": 1.1681914329528809,
|
| 2636 |
+
"rewards/rejected": -1.1185446977615356,
|
| 2637 |
+
"step": 146,
|
| 2638 |
+
"train_speed(iter/s)": 0.005493
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.11860517889110209,
|
| 2642 |
+
"grad_norm": 2.613725423812866,
|
| 2643 |
+
"learning_rate": 0.00019743734003931082,
|
| 2644 |
+
"logits/chosen": -0.808351993560791,
|
| 2645 |
+
"logits/rejected": -0.8136962652206421,
|
| 2646 |
+
"logps/chosen": -7.89247989654541,
|
| 2647 |
+
"logps/rejected": -23.955371856689453,
|
| 2648 |
+
"loss": 0.8511161804199219,
|
| 2649 |
+
"memory(GiB)": 46.83,
|
| 2650 |
+
"nll_loss": 0.39823463559150696,
|
| 2651 |
+
"rewards/accuracies": 0.78125,
|
| 2652 |
+
"rewards/chosen": 0.12871581315994263,
|
| 2653 |
+
"rewards/margins": 1.055700659751892,
|
| 2654 |
+
"rewards/rejected": -0.9269847869873047,
|
| 2655 |
+
"step": 147,
|
| 2656 |
+
"train_speed(iter/s)": 0.005493
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.11941201684274225,
|
| 2660 |
+
"grad_norm": 3.0128729343414307,
|
| 2661 |
+
"learning_rate": 0.00019737695400725677,
|
| 2662 |
+
"logits/chosen": -0.7927097678184509,
|
| 2663 |
+
"logits/rejected": -0.7919667959213257,
|
| 2664 |
+
"logps/chosen": -9.877785682678223,
|
| 2665 |
+
"logps/rejected": -23.477703094482422,
|
| 2666 |
+
"loss": 0.7262363433837891,
|
| 2667 |
+
"memory(GiB)": 46.83,
|
| 2668 |
+
"nll_loss": 0.3177878260612488,
|
| 2669 |
+
"rewards/accuracies": 0.78125,
|
| 2670 |
+
"rewards/chosen": 0.12437373399734497,
|
| 2671 |
+
"rewards/margins": 1.1551289558410645,
|
| 2672 |
+
"rewards/rejected": -1.0307551622390747,
|
| 2673 |
+
"step": 148,
|
| 2674 |
+
"train_speed(iter/s)": 0.005495
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.1202188547943824,
|
| 2678 |
+
"grad_norm": 2.5007247924804688,
|
| 2679 |
+
"learning_rate": 0.0001973158742256001,
|
| 2680 |
+
"logits/chosen": -0.815912663936615,
|
| 2681 |
+
"logits/rejected": -0.8240575194358826,
|
| 2682 |
+
"logps/chosen": -7.251335620880127,
|
| 2683 |
+
"logps/rejected": -23.904563903808594,
|
| 2684 |
+
"loss": 0.7742622494697571,
|
| 2685 |
+
"memory(GiB)": 46.83,
|
| 2686 |
+
"nll_loss": 0.3948196470737457,
|
| 2687 |
+
"rewards/accuracies": 0.8125,
|
| 2688 |
+
"rewards/chosen": 0.16877779364585876,
|
| 2689 |
+
"rewards/margins": 1.3302054405212402,
|
| 2690 |
+
"rewards/rejected": -1.1614277362823486,
|
| 2691 |
+
"step": 149,
|
| 2692 |
+
"train_speed(iter/s)": 0.005495
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.12102569274602254,
|
| 2696 |
+
"grad_norm": 2.996128559112549,
|
| 2697 |
+
"learning_rate": 0.0001972541011294959,
|
| 2698 |
+
"logits/chosen": -0.8276554346084595,
|
| 2699 |
+
"logits/rejected": -0.832274317741394,
|
| 2700 |
+
"logps/chosen": -9.810141563415527,
|
| 2701 |
+
"logps/rejected": -24.295639038085938,
|
| 2702 |
+
"loss": 0.733322262763977,
|
| 2703 |
+
"memory(GiB)": 46.83,
|
| 2704 |
+
"nll_loss": 0.34396567940711975,
|
| 2705 |
+
"rewards/accuracies": 0.90625,
|
| 2706 |
+
"rewards/chosen": 0.03486320376396179,
|
| 2707 |
+
"rewards/margins": 1.2224383354187012,
|
| 2708 |
+
"rewards/rejected": -1.1875752210617065,
|
| 2709 |
+
"step": 150,
|
| 2710 |
+
"train_speed(iter/s)": 0.005495
|
| 2711 |
+
}
|
| 2712 |
+
],
|
| 2713 |
+
"logging_steps": 1,
|
| 2714 |
+
"max_steps": 1239,
|
| 2715 |
+
"num_input_tokens_seen": 0,
|
| 2716 |
+
"num_train_epochs": 1,
|
| 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": 1.0977195623354204e+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 sha256:2cfa6616c6897406a916597ab742df18763c5a008e9494f8a7a1a9aee1e84f8d
|
| 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)
|