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).*\\.(k_proj|v_proj|o_proj|down_proj|up_proj|gate_proj|q_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 sha256:5303727c802af3176a08d9a436d08911279e970ca800d1234558a8c34503adf8
|
| 3 |
+
size 52511776
|
additional_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"lora_dtype": null, "lorap_lr_ratio": null, "lorap_emb_lr": 1e-06}
|
args.json
ADDED
|
@@ -0,0 +1,438 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
|
| 3 |
+
"model_type": "llama3_2_vision",
|
| 4 |
+
"model_revision": null,
|
| 5 |
+
"task_type": "causal_lm",
|
| 6 |
+
"torch_dtype": "bfloat16",
|
| 7 |
+
"attn_impl": null,
|
| 8 |
+
"num_labels": null,
|
| 9 |
+
"problem_type": null,
|
| 10 |
+
"rope_scaling": null,
|
| 11 |
+
"device_map": null,
|
| 12 |
+
"max_memory": {},
|
| 13 |
+
"local_repo_path": null,
|
| 14 |
+
"template": "llama3_2_vision",
|
| 15 |
+
"system": "You are a helpful assistant that truthfully answers user questions about the provided image.",
|
| 16 |
+
"max_length": 4096,
|
| 17 |
+
"truncation_strategy": "delete",
|
| 18 |
+
"max_pixels": null,
|
| 19 |
+
"agent_template": null,
|
| 20 |
+
"norm_bbox": null,
|
| 21 |
+
"response_prefix": null,
|
| 22 |
+
"padding_side": "right",
|
| 23 |
+
"loss_scale": "last_round",
|
| 24 |
+
"sequence_parallel_size": 1,
|
| 25 |
+
"use_chat_template": true,
|
| 26 |
+
"template_backend": "swift",
|
| 27 |
+
"dataset": [
|
| 28 |
+
"task2_dpo_absgt0_taskgt6__pari0.5_no_cheat.json"
|
| 29 |
+
],
|
| 30 |
+
"val_dataset": [],
|
| 31 |
+
"split_dataset_ratio": 0.01,
|
| 32 |
+
"data_seed": 42,
|
| 33 |
+
"dataset_num_proc": 4,
|
| 34 |
+
"dataset_shuffle": true,
|
| 35 |
+
"val_dataset_shuffle": false,
|
| 36 |
+
"streaming": false,
|
| 37 |
+
"interleave_prob": null,
|
| 38 |
+
"stopping_strategy": "first_exhausted",
|
| 39 |
+
"shuffle_buffer_size": 1000,
|
| 40 |
+
"enable_cache": false,
|
| 41 |
+
"download_mode": "reuse_dataset_if_exists",
|
| 42 |
+
"columns": {},
|
| 43 |
+
"strict": false,
|
| 44 |
+
"remove_unused_columns": true,
|
| 45 |
+
"model_name": [
|
| 46 |
+
null,
|
| 47 |
+
null
|
| 48 |
+
],
|
| 49 |
+
"model_author": [
|
| 50 |
+
null,
|
| 51 |
+
null
|
| 52 |
+
],
|
| 53 |
+
"custom_dataset_info": [],
|
| 54 |
+
"quant_method": null,
|
| 55 |
+
"quant_bits": null,
|
| 56 |
+
"hqq_axis": null,
|
| 57 |
+
"bnb_4bit_compute_dtype": "bfloat16",
|
| 58 |
+
"bnb_4bit_quant_type": "nf4",
|
| 59 |
+
"bnb_4bit_use_double_quant": true,
|
| 60 |
+
"bnb_4bit_quant_storage": null,
|
| 61 |
+
"max_new_tokens": 64,
|
| 62 |
+
"temperature": 0.9,
|
| 63 |
+
"top_k": 50,
|
| 64 |
+
"top_p": 0.9,
|
| 65 |
+
"repetition_penalty": 1.0,
|
| 66 |
+
"num_beams": 1,
|
| 67 |
+
"stream": false,
|
| 68 |
+
"stop_words": [],
|
| 69 |
+
"logprobs": false,
|
| 70 |
+
"top_logprobs": null,
|
| 71 |
+
"ckpt_dir": null,
|
| 72 |
+
"load_dataset_config": null,
|
| 73 |
+
"lora_modules": [],
|
| 74 |
+
"tuner_backend": "peft",
|
| 75 |
+
"train_type": "lora",
|
| 76 |
+
"adapters": [],
|
| 77 |
+
"external_plugins": [],
|
| 78 |
+
"seed": 42,
|
| 79 |
+
"model_kwargs": {},
|
| 80 |
+
"load_args": false,
|
| 81 |
+
"load_data_args": false,
|
| 82 |
+
"use_hf": false,
|
| 83 |
+
"hub_token": null,
|
| 84 |
+
"custom_register_path": [],
|
| 85 |
+
"ignore_args_error": false,
|
| 86 |
+
"use_swift_lora": false,
|
| 87 |
+
"output_dir": "/data2/kdd/crag/cjz/output/task2_dpo_absgt0_taskgt6__pari0.5_no_cheat/v0-20250606-013952",
|
| 88 |
+
"overwrite_output_dir": false,
|
| 89 |
+
"do_train": false,
|
| 90 |
+
"do_eval": false,
|
| 91 |
+
"do_predict": false,
|
| 92 |
+
"eval_strategy": "steps",
|
| 93 |
+
"prediction_loss_only": false,
|
| 94 |
+
"per_device_train_batch_size": 1,
|
| 95 |
+
"per_device_eval_batch_size": 1,
|
| 96 |
+
"per_gpu_train_batch_size": null,
|
| 97 |
+
"per_gpu_eval_batch_size": null,
|
| 98 |
+
"gradient_accumulation_steps": 32,
|
| 99 |
+
"eval_accumulation_steps": null,
|
| 100 |
+
"eval_delay": 0,
|
| 101 |
+
"torch_empty_cache_steps": null,
|
| 102 |
+
"learning_rate": 0.0002,
|
| 103 |
+
"weight_decay": 0.1,
|
| 104 |
+
"adam_beta1": 0.9,
|
| 105 |
+
"adam_beta2": 0.95,
|
| 106 |
+
"adam_epsilon": 1e-08,
|
| 107 |
+
"max_grad_norm": 1.0,
|
| 108 |
+
"num_train_epochs": 3.0,
|
| 109 |
+
"max_steps": -1,
|
| 110 |
+
"lr_scheduler_type": "cosine",
|
| 111 |
+
"lr_scheduler_kwargs": null,
|
| 112 |
+
"warmup_ratio": 0.05,
|
| 113 |
+
"warmup_steps": 0,
|
| 114 |
+
"log_level": "passive",
|
| 115 |
+
"log_level_replica": "warning",
|
| 116 |
+
"log_on_each_node": true,
|
| 117 |
+
"logging_dir": "/data2/kdd/crag/cjz/output/task2_dpo_absgt0_taskgt6__pari0.5_no_cheat/v0-20250606-013952/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 0x7b09d341ed40>, model_arch='llama3_2_vision', architectures=['MllamaForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.45'], tags=[])",
|
| 434 |
+
"model_dir": "/data1/chenjiazun/LLM/Llama-3.2-11B-Vision-Instruct",
|
| 435 |
+
"hub": "<class 'swift.hub.hub.MSHub'>",
|
| 436 |
+
"evaluation_strategy": "steps",
|
| 437 |
+
"training_args": "DPOConfig(output_dir='/data2/kdd/crag/cjz/output/task2_dpo_absgt0_taskgt6__pari0.5_no_cheat/v0-20250606-013952', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=32, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0002, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/data2/kdd/crag/cjz/output/task2_dpo_absgt0_taskgt6__pari0.5_no_cheat/v0-20250606-013952/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=50, save_total_limit=None, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=300, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/data2/kdd/crag/cjz/output/task2_dpo_absgt0_taskgt6__pari0.5_no_cheat/v0-20250606-013952', 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 sha256:6d48bb100ad9fb6d06facffb5e87359cdc0ecfe29e26bcbb2d50de33a7e9284f
|
| 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": 0.4287628405538187,
|
| 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": 0.002858418937025458,
|
| 14 |
+
"grad_norm": 10.349462509155273,
|
| 15 |
+
"learning_rate": 3.7735849056603773e-06,
|
| 16 |
+
"logits/chosen": -0.6422490477561951,
|
| 17 |
+
"logits/rejected": -0.6447486877441406,
|
| 18 |
+
"logps/chosen": -6.999429225921631,
|
| 19 |
+
"logps/rejected": -13.433603286743164,
|
| 20 |
+
"loss": 1.3778549432754517,
|
| 21 |
+
"memory(GiB)": 46.73,
|
| 22 |
+
"nll_loss": 0.6847077012062073,
|
| 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)": 0.005273
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.005716837874050916,
|
| 32 |
+
"grad_norm": 11.263213157653809,
|
| 33 |
+
"learning_rate": 7.547169811320755e-06,
|
| 34 |
+
"logits/chosen": -0.6429960131645203,
|
| 35 |
+
"logits/rejected": -0.6454926133155823,
|
| 36 |
+
"logps/chosen": -9.604641914367676,
|
| 37 |
+
"logps/rejected": -13.53369140625,
|
| 38 |
+
"loss": 1.4975696802139282,
|
| 39 |
+
"memory(GiB)": 46.73,
|
| 40 |
+
"nll_loss": 0.8044224381446838,
|
| 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)": 0.005366
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.008575256811076373,
|
| 50 |
+
"grad_norm": 9.09404182434082,
|
| 51 |
+
"learning_rate": 1.1320754716981132e-05,
|
| 52 |
+
"logits/chosen": -0.6486532092094421,
|
| 53 |
+
"logits/rejected": -0.6493248343467712,
|
| 54 |
+
"logps/chosen": -7.69912576675415,
|
| 55 |
+
"logps/rejected": -17.816326141357422,
|
| 56 |
+
"loss": 1.2593896389007568,
|
| 57 |
+
"memory(GiB)": 46.73,
|
| 58 |
+
"nll_loss": 0.5679630637168884,
|
| 59 |
+
"rewards/accuracies": 0.625,
|
| 60 |
+
"rewards/chosen": 0.0031128099653869867,
|
| 61 |
+
"rewards/margins": 0.0035441224463284016,
|
| 62 |
+
"rewards/rejected": -0.00043131227721460164,
|
| 63 |
+
"step": 3,
|
| 64 |
+
"train_speed(iter/s)": 0.005419
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.011433675748101831,
|
| 68 |
+
"grad_norm": 13.314420700073242,
|
| 69 |
+
"learning_rate": 1.509433962264151e-05,
|
| 70 |
+
"logits/chosen": -0.6668453812599182,
|
| 71 |
+
"logits/rejected": -0.6718658804893494,
|
| 72 |
+
"logps/chosen": -6.061755180358887,
|
| 73 |
+
"logps/rejected": -13.643918991088867,
|
| 74 |
+
"loss": 1.4406771659851074,
|
| 75 |
+
"memory(GiB)": 46.73,
|
| 76 |
+
"nll_loss": 0.7477570176124573,
|
| 77 |
+
"rewards/accuracies": 0.53125,
|
| 78 |
+
"rewards/chosen": 0.008774133399128914,
|
| 79 |
+
"rewards/margins": 0.0006008772179484367,
|
| 80 |
+
"rewards/rejected": 0.008173256181180477,
|
| 81 |
+
"step": 4,
|
| 82 |
+
"train_speed(iter/s)": 0.005446
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.014292094685127288,
|
| 86 |
+
"grad_norm": 12.340804100036621,
|
| 87 |
+
"learning_rate": 1.8867924528301888e-05,
|
| 88 |
+
"logits/chosen": -0.6321276426315308,
|
| 89 |
+
"logits/rejected": -0.6321861743927002,
|
| 90 |
+
"logps/chosen": -11.234216690063477,
|
| 91 |
+
"logps/rejected": -11.82645034790039,
|
| 92 |
+
"loss": 1.3690662384033203,
|
| 93 |
+
"memory(GiB)": 46.73,
|
| 94 |
+
"nll_loss": 0.6870464086532593,
|
| 95 |
+
"rewards/accuracies": 0.6875,
|
| 96 |
+
"rewards/chosen": 0.032103873789310455,
|
| 97 |
+
"rewards/margins": 0.022960105910897255,
|
| 98 |
+
"rewards/rejected": 0.0091437678784132,
|
| 99 |
+
"step": 5,
|
| 100 |
+
"train_speed(iter/s)": 0.005444
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.017150513622152745,
|
| 104 |
+
"grad_norm": 13.995262145996094,
|
| 105 |
+
"learning_rate": 2.2641509433962265e-05,
|
| 106 |
+
"logits/chosen": -0.6530164480209351,
|
| 107 |
+
"logits/rejected": -0.6550691723823547,
|
| 108 |
+
"logps/chosen": -8.742744445800781,
|
| 109 |
+
"logps/rejected": -12.622247695922852,
|
| 110 |
+
"loss": 1.3672001361846924,
|
| 111 |
+
"memory(GiB)": 46.73,
|
| 112 |
+
"nll_loss": 0.702350378036499,
|
| 113 |
+
"rewards/accuracies": 0.65625,
|
| 114 |
+
"rewards/chosen": 0.0800265222787857,
|
| 115 |
+
"rewards/margins": 0.06047351285815239,
|
| 116 |
+
"rewards/rejected": 0.019553007557988167,
|
| 117 |
+
"step": 6,
|
| 118 |
+
"train_speed(iter/s)": 0.005427
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.020008932559178204,
|
| 122 |
+
"grad_norm": 10.019857406616211,
|
| 123 |
+
"learning_rate": 2.641509433962264e-05,
|
| 124 |
+
"logits/chosen": -0.6211694478988647,
|
| 125 |
+
"logits/rejected": -0.6214967370033264,
|
| 126 |
+
"logps/chosen": -6.935382843017578,
|
| 127 |
+
"logps/rejected": -13.613279342651367,
|
| 128 |
+
"loss": 1.2412631511688232,
|
| 129 |
+
"memory(GiB)": 46.73,
|
| 130 |
+
"nll_loss": 0.5651571154594421,
|
| 131 |
+
"rewards/accuracies": 0.59375,
|
| 132 |
+
"rewards/chosen": 0.13799861073493958,
|
| 133 |
+
"rewards/margins": 0.053141627460718155,
|
| 134 |
+
"rewards/rejected": 0.08485697209835052,
|
| 135 |
+
"step": 7,
|
| 136 |
+
"train_speed(iter/s)": 0.005441
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.022867351496203663,
|
| 140 |
+
"grad_norm": 6.375326633453369,
|
| 141 |
+
"learning_rate": 3.018867924528302e-05,
|
| 142 |
+
"logits/chosen": -0.6666778326034546,
|
| 143 |
+
"logits/rejected": -0.6678147315979004,
|
| 144 |
+
"logps/chosen": -8.114872932434082,
|
| 145 |
+
"logps/rejected": -13.392560958862305,
|
| 146 |
+
"loss": 1.1162137985229492,
|
| 147 |
+
"memory(GiB)": 46.73,
|
| 148 |
+
"nll_loss": 0.4446101784706116,
|
| 149 |
+
"rewards/accuracies": 0.59375,
|
| 150 |
+
"rewards/chosen": 0.20063960552215576,
|
| 151 |
+
"rewards/margins": 0.09723334014415741,
|
| 152 |
+
"rewards/rejected": 0.10340625792741776,
|
| 153 |
+
"step": 8,
|
| 154 |
+
"train_speed(iter/s)": 0.005426
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.02572577043322912,
|
| 158 |
+
"grad_norm": 4.046472549438477,
|
| 159 |
+
"learning_rate": 3.39622641509434e-05,
|
| 160 |
+
"logits/chosen": -0.5842044949531555,
|
| 161 |
+
"logits/rejected": -0.5892153382301331,
|
| 162 |
+
"logps/chosen": -13.877073287963867,
|
| 163 |
+
"logps/rejected": -14.116576194763184,
|
| 164 |
+
"loss": 1.259205937385559,
|
| 165 |
+
"memory(GiB)": 46.73,
|
| 166 |
+
"nll_loss": 0.5605449676513672,
|
| 167 |
+
"rewards/accuracies": 0.53125,
|
| 168 |
+
"rewards/chosen": 0.1392103135585785,
|
| 169 |
+
"rewards/margins": 0.09627523273229599,
|
| 170 |
+
"rewards/rejected": 0.0429350808262825,
|
| 171 |
+
"step": 9,
|
| 172 |
+
"train_speed(iter/s)": 0.005422
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.028584189370254576,
|
| 176 |
+
"grad_norm": 7.625533103942871,
|
| 177 |
+
"learning_rate": 3.7735849056603776e-05,
|
| 178 |
+
"logits/chosen": -0.6210353374481201,
|
| 179 |
+
"logits/rejected": -0.6224341988563538,
|
| 180 |
+
"logps/chosen": -8.32575798034668,
|
| 181 |
+
"logps/rejected": -11.715011596679688,
|
| 182 |
+
"loss": 1.2772942781448364,
|
| 183 |
+
"memory(GiB)": 46.73,
|
| 184 |
+
"nll_loss": 0.48597079515457153,
|
| 185 |
+
"rewards/accuracies": 0.4375,
|
| 186 |
+
"rewards/chosen": -0.043191827833652496,
|
| 187 |
+
"rewards/margins": 0.029505327343940735,
|
| 188 |
+
"rewards/rejected": -0.07269717007875443,
|
| 189 |
+
"step": 10,
|
| 190 |
+
"train_speed(iter/s)": 0.005438
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.03144260830728004,
|
| 194 |
+
"grad_norm": 11.961934089660645,
|
| 195 |
+
"learning_rate": 4.150943396226415e-05,
|
| 196 |
+
"logits/chosen": -0.6265634298324585,
|
| 197 |
+
"logits/rejected": -0.6289808750152588,
|
| 198 |
+
"logps/chosen": -7.129470348358154,
|
| 199 |
+
"logps/rejected": -11.932583808898926,
|
| 200 |
+
"loss": 1.447466254234314,
|
| 201 |
+
"memory(GiB)": 46.73,
|
| 202 |
+
"nll_loss": 0.6552329659461975,
|
| 203 |
+
"rewards/accuracies": 0.46875,
|
| 204 |
+
"rewards/chosen": -0.00290237320587039,
|
| 205 |
+
"rewards/margins": 0.06162097677588463,
|
| 206 |
+
"rewards/rejected": -0.0645233541727066,
|
| 207 |
+
"step": 11,
|
| 208 |
+
"train_speed(iter/s)": 0.005473
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.03430102724430549,
|
| 212 |
+
"grad_norm": 10.154479026794434,
|
| 213 |
+
"learning_rate": 4.528301886792453e-05,
|
| 214 |
+
"logits/chosen": -0.6388624906539917,
|
| 215 |
+
"logits/rejected": -0.639370858669281,
|
| 216 |
+
"logps/chosen": -9.490407943725586,
|
| 217 |
+
"logps/rejected": -12.88676643371582,
|
| 218 |
+
"loss": 1.4875458478927612,
|
| 219 |
+
"memory(GiB)": 46.73,
|
| 220 |
+
"nll_loss": 0.6468713283538818,
|
| 221 |
+
"rewards/accuracies": 0.4375,
|
| 222 |
+
"rewards/chosen": -0.10463112592697144,
|
| 223 |
+
"rewards/margins": -0.059477299451828,
|
| 224 |
+
"rewards/rejected": -0.04515381157398224,
|
| 225 |
+
"step": 12,
|
| 226 |
+
"train_speed(iter/s)": 0.005465
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.03715944618133095,
|
| 230 |
+
"grad_norm": 4.091860771179199,
|
| 231 |
+
"learning_rate": 4.9056603773584906e-05,
|
| 232 |
+
"logits/chosen": -0.6680133938789368,
|
| 233 |
+
"logits/rejected": -0.6711105108261108,
|
| 234 |
+
"logps/chosen": -5.56148099899292,
|
| 235 |
+
"logps/rejected": -13.865943908691406,
|
| 236 |
+
"loss": 1.0154482126235962,
|
| 237 |
+
"memory(GiB)": 46.73,
|
| 238 |
+
"nll_loss": 0.35444653034210205,
|
| 239 |
+
"rewards/accuracies": 0.65625,
|
| 240 |
+
"rewards/chosen": 0.12190410494804382,
|
| 241 |
+
"rewards/margins": 0.3293726444244385,
|
| 242 |
+
"rewards/rejected": -0.20746850967407227,
|
| 243 |
+
"step": 13,
|
| 244 |
+
"train_speed(iter/s)": 0.00546
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.04001786511835641,
|
| 248 |
+
"grad_norm": 3.039910316467285,
|
| 249 |
+
"learning_rate": 5.283018867924528e-05,
|
| 250 |
+
"logits/chosen": -0.6245446801185608,
|
| 251 |
+
"logits/rejected": -0.6280595660209656,
|
| 252 |
+
"logps/chosen": -7.042474269866943,
|
| 253 |
+
"logps/rejected": -14.88280963897705,
|
| 254 |
+
"loss": 0.9986085295677185,
|
| 255 |
+
"memory(GiB)": 46.73,
|
| 256 |
+
"nll_loss": 0.32772910594940186,
|
| 257 |
+
"rewards/accuracies": 0.59375,
|
| 258 |
+
"rewards/chosen": 0.08946945518255234,
|
| 259 |
+
"rewards/margins": 0.2548743188381195,
|
| 260 |
+
"rewards/rejected": -0.16540484130382538,
|
| 261 |
+
"step": 14,
|
| 262 |
+
"train_speed(iter/s)": 0.005463
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.042876284055381866,
|
| 266 |
+
"grad_norm": 2.8388617038726807,
|
| 267 |
+
"learning_rate": 5.660377358490566e-05,
|
| 268 |
+
"logits/chosen": -0.613956868648529,
|
| 269 |
+
"logits/rejected": -0.6150951385498047,
|
| 270 |
+
"logps/chosen": -5.398593902587891,
|
| 271 |
+
"logps/rejected": -10.984697341918945,
|
| 272 |
+
"loss": 0.9217166900634766,
|
| 273 |
+
"memory(GiB)": 46.73,
|
| 274 |
+
"nll_loss": 0.21436992287635803,
|
| 275 |
+
"rewards/accuracies": 0.65625,
|
| 276 |
+
"rewards/chosen": 0.2490297257900238,
|
| 277 |
+
"rewards/margins": 0.12685683369636536,
|
| 278 |
+
"rewards/rejected": 0.12217291444540024,
|
| 279 |
+
"step": 15,
|
| 280 |
+
"train_speed(iter/s)": 0.005464
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.045734702992407325,
|
| 284 |
+
"grad_norm": 3.623399257659912,
|
| 285 |
+
"learning_rate": 6.037735849056604e-05,
|
| 286 |
+
"logits/chosen": -0.5890456438064575,
|
| 287 |
+
"logits/rejected": -0.5907201766967773,
|
| 288 |
+
"logps/chosen": -7.65773868560791,
|
| 289 |
+
"logps/rejected": -15.24480152130127,
|
| 290 |
+
"loss": 1.130638837814331,
|
| 291 |
+
"memory(GiB)": 46.73,
|
| 292 |
+
"nll_loss": 0.33376482129096985,
|
| 293 |
+
"rewards/accuracies": 0.46875,
|
| 294 |
+
"rewards/chosen": 0.14319562911987305,
|
| 295 |
+
"rewards/margins": 0.01443251222372055,
|
| 296 |
+
"rewards/rejected": 0.1287631392478943,
|
| 297 |
+
"step": 16,
|
| 298 |
+
"train_speed(iter/s)": 0.00546
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.048593121929432784,
|
| 302 |
+
"grad_norm": 2.9850075244903564,
|
| 303 |
+
"learning_rate": 6.415094339622641e-05,
|
| 304 |
+
"logits/chosen": -0.5228931903839111,
|
| 305 |
+
"logits/rejected": -0.524716854095459,
|
| 306 |
+
"logps/chosen": -10.135262489318848,
|
| 307 |
+
"logps/rejected": -14.072588920593262,
|
| 308 |
+
"loss": 1.2770482301712036,
|
| 309 |
+
"memory(GiB)": 46.73,
|
| 310 |
+
"nll_loss": 0.48515522480010986,
|
| 311 |
+
"rewards/accuracies": 0.40625,
|
| 312 |
+
"rewards/chosen": 0.13618236780166626,
|
| 313 |
+
"rewards/margins": -0.09462341666221619,
|
| 314 |
+
"rewards/rejected": 0.23080575466156006,
|
| 315 |
+
"step": 17,
|
| 316 |
+
"train_speed(iter/s)": 0.005473
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.05145154086645824,
|
| 320 |
+
"grad_norm": 2.2746193408966064,
|
| 321 |
+
"learning_rate": 6.79245283018868e-05,
|
| 322 |
+
"logits/chosen": -0.5764032006263733,
|
| 323 |
+
"logits/rejected": -0.5762047171592712,
|
| 324 |
+
"logps/chosen": -11.675098419189453,
|
| 325 |
+
"logps/rejected": -10.994297981262207,
|
| 326 |
+
"loss": 1.2022521495819092,
|
| 327 |
+
"memory(GiB)": 46.73,
|
| 328 |
+
"nll_loss": 0.4671342372894287,
|
| 329 |
+
"rewards/accuracies": 0.46875,
|
| 330 |
+
"rewards/chosen": 0.1593710482120514,
|
| 331 |
+
"rewards/margins": -0.00844080001115799,
|
| 332 |
+
"rewards/rejected": 0.1678118258714676,
|
| 333 |
+
"step": 18,
|
| 334 |
+
"train_speed(iter/s)": 0.005468
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.0543099598034837,
|
| 338 |
+
"grad_norm": 2.324984312057495,
|
| 339 |
+
"learning_rate": 7.169811320754717e-05,
|
| 340 |
+
"logits/chosen": -0.567690908908844,
|
| 341 |
+
"logits/rejected": -0.569066047668457,
|
| 342 |
+
"logps/chosen": -9.0728120803833,
|
| 343 |
+
"logps/rejected": -12.268661499023438,
|
| 344 |
+
"loss": 1.1060843467712402,
|
| 345 |
+
"memory(GiB)": 46.73,
|
| 346 |
+
"nll_loss": 0.41300466656684875,
|
| 347 |
+
"rewards/accuracies": 0.53125,
|
| 348 |
+
"rewards/chosen": 0.23081830143928528,
|
| 349 |
+
"rewards/margins": 0.0710553377866745,
|
| 350 |
+
"rewards/rejected": 0.15976294875144958,
|
| 351 |
+
"step": 19,
|
| 352 |
+
"train_speed(iter/s)": 0.005463
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.05716837874050915,
|
| 356 |
+
"grad_norm": 2.0991458892822266,
|
| 357 |
+
"learning_rate": 7.547169811320755e-05,
|
| 358 |
+
"logits/chosen": -0.6091416478157043,
|
| 359 |
+
"logits/rejected": -0.6100034117698669,
|
| 360 |
+
"logps/chosen": -4.26762056350708,
|
| 361 |
+
"logps/rejected": -11.32187557220459,
|
| 362 |
+
"loss": 1.0021699666976929,
|
| 363 |
+
"memory(GiB)": 46.73,
|
| 364 |
+
"nll_loss": 0.2933175265789032,
|
| 365 |
+
"rewards/accuracies": 0.4375,
|
| 366 |
+
"rewards/chosen": 0.15655732154846191,
|
| 367 |
+
"rewards/margins": 0.008277412503957748,
|
| 368 |
+
"rewards/rejected": 0.14827993512153625,
|
| 369 |
+
"step": 20,
|
| 370 |
+
"train_speed(iter/s)": 0.005467
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.06002679767753461,
|
| 374 |
+
"grad_norm": 3.4958455562591553,
|
| 375 |
+
"learning_rate": 7.924528301886794e-05,
|
| 376 |
+
"logits/chosen": -0.571921706199646,
|
| 377 |
+
"logits/rejected": -0.572878360748291,
|
| 378 |
+
"logps/chosen": -9.612298965454102,
|
| 379 |
+
"logps/rejected": -12.484085083007812,
|
| 380 |
+
"loss": 1.1795949935913086,
|
| 381 |
+
"memory(GiB)": 46.73,
|
| 382 |
+
"nll_loss": 0.46760252118110657,
|
| 383 |
+
"rewards/accuracies": 0.53125,
|
| 384 |
+
"rewards/chosen": 0.14643912017345428,
|
| 385 |
+
"rewards/margins": 0.0200117826461792,
|
| 386 |
+
"rewards/rejected": 0.12642733752727509,
|
| 387 |
+
"step": 21,
|
| 388 |
+
"train_speed(iter/s)": 0.005467
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.06288521661456008,
|
| 392 |
+
"grad_norm": 3.175546169281006,
|
| 393 |
+
"learning_rate": 8.30188679245283e-05,
|
| 394 |
+
"logits/chosen": -0.6108006834983826,
|
| 395 |
+
"logits/rejected": -0.6111243367195129,
|
| 396 |
+
"logps/chosen": -6.207424163818359,
|
| 397 |
+
"logps/rejected": -7.818144798278809,
|
| 398 |
+
"loss": 1.1549935340881348,
|
| 399 |
+
"memory(GiB)": 46.73,
|
| 400 |
+
"nll_loss": 0.4683927893638611,
|
| 401 |
+
"rewards/accuracies": 0.5625,
|
| 402 |
+
"rewards/chosen": 0.23285990953445435,
|
| 403 |
+
"rewards/margins": 0.06634702533483505,
|
| 404 |
+
"rewards/rejected": 0.16651292145252228,
|
| 405 |
+
"step": 22,
|
| 406 |
+
"train_speed(iter/s)": 0.00547
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.06574363555158554,
|
| 410 |
+
"grad_norm": 2.4531869888305664,
|
| 411 |
+
"learning_rate": 8.679245283018869e-05,
|
| 412 |
+
"logits/chosen": -0.6012008190155029,
|
| 413 |
+
"logits/rejected": -0.6021928191184998,
|
| 414 |
+
"logps/chosen": -5.598664283752441,
|
| 415 |
+
"logps/rejected": -13.262667655944824,
|
| 416 |
+
"loss": 0.9260164499282837,
|
| 417 |
+
"memory(GiB)": 46.73,
|
| 418 |
+
"nll_loss": 0.33532270789146423,
|
| 419 |
+
"rewards/accuracies": 0.625,
|
| 420 |
+
"rewards/chosen": 0.3285560607910156,
|
| 421 |
+
"rewards/margins": 0.3157665431499481,
|
| 422 |
+
"rewards/rejected": 0.012789532542228699,
|
| 423 |
+
"step": 23,
|
| 424 |
+
"train_speed(iter/s)": 0.005467
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.06860205448861098,
|
| 428 |
+
"grad_norm": 2.328031301498413,
|
| 429 |
+
"learning_rate": 9.056603773584906e-05,
|
| 430 |
+
"logits/chosen": -0.567432165145874,
|
| 431 |
+
"logits/rejected": -0.5704118609428406,
|
| 432 |
+
"logps/chosen": -4.026985168457031,
|
| 433 |
+
"logps/rejected": -14.904813766479492,
|
| 434 |
+
"loss": 0.8984246253967285,
|
| 435 |
+
"memory(GiB)": 46.73,
|
| 436 |
+
"nll_loss": 0.26223480701446533,
|
| 437 |
+
"rewards/accuracies": 0.59375,
|
| 438 |
+
"rewards/chosen": 0.26638248562812805,
|
| 439 |
+
"rewards/margins": 0.22430934011936188,
|
| 440 |
+
"rewards/rejected": 0.042073119431734085,
|
| 441 |
+
"step": 24,
|
| 442 |
+
"train_speed(iter/s)": 0.005466
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.07146047342563644,
|
| 446 |
+
"grad_norm": 3.5119378566741943,
|
| 447 |
+
"learning_rate": 9.433962264150944e-05,
|
| 448 |
+
"logits/chosen": -0.5709348320960999,
|
| 449 |
+
"logits/rejected": -0.5688645243644714,
|
| 450 |
+
"logps/chosen": -7.716845989227295,
|
| 451 |
+
"logps/rejected": -7.868927955627441,
|
| 452 |
+
"loss": 1.113297462463379,
|
| 453 |
+
"memory(GiB)": 46.73,
|
| 454 |
+
"nll_loss": 0.3660878837108612,
|
| 455 |
+
"rewards/accuracies": 0.5,
|
| 456 |
+
"rewards/chosen": 0.2273256480693817,
|
| 457 |
+
"rewards/margins": -0.0022776294499635696,
|
| 458 |
+
"rewards/rejected": 0.22960326075553894,
|
| 459 |
+
"step": 25,
|
| 460 |
+
"train_speed(iter/s)": 0.005475
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.0743188923626619,
|
| 464 |
+
"grad_norm": 3.2336673736572266,
|
| 465 |
+
"learning_rate": 9.811320754716981e-05,
|
| 466 |
+
"logits/chosen": -0.589956521987915,
|
| 467 |
+
"logits/rejected": -0.587726354598999,
|
| 468 |
+
"logps/chosen": -10.072855949401855,
|
| 469 |
+
"logps/rejected": -12.857810974121094,
|
| 470 |
+
"loss": 1.0991942882537842,
|
| 471 |
+
"memory(GiB)": 46.73,
|
| 472 |
+
"nll_loss": 0.40110355615615845,
|
| 473 |
+
"rewards/accuracies": 0.5,
|
| 474 |
+
"rewards/chosen": 0.17576897144317627,
|
| 475 |
+
"rewards/margins": 0.11343260109424591,
|
| 476 |
+
"rewards/rejected": 0.062336355447769165,
|
| 477 |
+
"step": 26,
|
| 478 |
+
"train_speed(iter/s)": 0.005466
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.07717731129968736,
|
| 482 |
+
"grad_norm": 2.6813321113586426,
|
| 483 |
+
"learning_rate": 0.0001018867924528302,
|
| 484 |
+
"logits/chosen": -0.5832359194755554,
|
| 485 |
+
"logits/rejected": -0.5860565304756165,
|
| 486 |
+
"logps/chosen": -8.197118759155273,
|
| 487 |
+
"logps/rejected": -17.45244598388672,
|
| 488 |
+
"loss": 0.8366767764091492,
|
| 489 |
+
"memory(GiB)": 46.73,
|
| 490 |
+
"nll_loss": 0.2886262536048889,
|
| 491 |
+
"rewards/accuracies": 0.71875,
|
| 492 |
+
"rewards/chosen": 0.3297504186630249,
|
| 493 |
+
"rewards/margins": 0.48295390605926514,
|
| 494 |
+
"rewards/rejected": -0.15320347249507904,
|
| 495 |
+
"step": 27,
|
| 496 |
+
"train_speed(iter/s)": 0.00547
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.08003573023671282,
|
| 500 |
+
"grad_norm": 3.3800203800201416,
|
| 501 |
+
"learning_rate": 0.00010566037735849057,
|
| 502 |
+
"logits/chosen": -0.6005962491035461,
|
| 503 |
+
"logits/rejected": -0.6035608053207397,
|
| 504 |
+
"logps/chosen": -9.117904663085938,
|
| 505 |
+
"logps/rejected": -18.400161743164062,
|
| 506 |
+
"loss": 1.0143071413040161,
|
| 507 |
+
"memory(GiB)": 46.73,
|
| 508 |
+
"nll_loss": 0.4276316463947296,
|
| 509 |
+
"rewards/accuracies": 0.65625,
|
| 510 |
+
"rewards/chosen": 0.271901398897171,
|
| 511 |
+
"rewards/margins": 0.4319708049297333,
|
| 512 |
+
"rewards/rejected": -0.16006940603256226,
|
| 513 |
+
"step": 28,
|
| 514 |
+
"train_speed(iter/s)": 0.005468
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.08289414917373827,
|
| 518 |
+
"grad_norm": 5.461134433746338,
|
| 519 |
+
"learning_rate": 0.00010943396226415095,
|
| 520 |
+
"logits/chosen": -0.5596082210540771,
|
| 521 |
+
"logits/rejected": -0.5635167360305786,
|
| 522 |
+
"logps/chosen": -7.607837200164795,
|
| 523 |
+
"logps/rejected": -23.174358367919922,
|
| 524 |
+
"loss": 0.803045928478241,
|
| 525 |
+
"memory(GiB)": 46.73,
|
| 526 |
+
"nll_loss": 0.32165294885635376,
|
| 527 |
+
"rewards/accuracies": 0.78125,
|
| 528 |
+
"rewards/chosen": 0.3834715783596039,
|
| 529 |
+
"rewards/margins": 0.7214637994766235,
|
| 530 |
+
"rewards/rejected": -0.33799219131469727,
|
| 531 |
+
"step": 29,
|
| 532 |
+
"train_speed(iter/s)": 0.005464
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.08575256811076373,
|
| 536 |
+
"grad_norm": 3.3377506732940674,
|
| 537 |
+
"learning_rate": 0.00011320754716981132,
|
| 538 |
+
"logits/chosen": -0.6003558039665222,
|
| 539 |
+
"logits/rejected": -0.6016712188720703,
|
| 540 |
+
"logps/chosen": -7.941157817840576,
|
| 541 |
+
"logps/rejected": -10.564404487609863,
|
| 542 |
+
"loss": 1.0737202167510986,
|
| 543 |
+
"memory(GiB)": 46.73,
|
| 544 |
+
"nll_loss": 0.3314566910266876,
|
| 545 |
+
"rewards/accuracies": 0.59375,
|
| 546 |
+
"rewards/chosen": 0.1379927098751068,
|
| 547 |
+
"rewards/margins": 0.08670535683631897,
|
| 548 |
+
"rewards/rejected": 0.05128733813762665,
|
| 549 |
+
"step": 30,
|
| 550 |
+
"train_speed(iter/s)": 0.005461
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.08861098704778919,
|
| 554 |
+
"grad_norm": 3.4214563369750977,
|
| 555 |
+
"learning_rate": 0.0001169811320754717,
|
| 556 |
+
"logits/chosen": -0.5186684727668762,
|
| 557 |
+
"logits/rejected": -0.5189406275749207,
|
| 558 |
+
"logps/chosen": -6.143683910369873,
|
| 559 |
+
"logps/rejected": -25.88001251220703,
|
| 560 |
+
"loss": 0.7792238593101501,
|
| 561 |
+
"memory(GiB)": 46.73,
|
| 562 |
+
"nll_loss": 0.2514691948890686,
|
| 563 |
+
"rewards/accuracies": 0.71875,
|
| 564 |
+
"rewards/chosen": 0.2757991552352905,
|
| 565 |
+
"rewards/margins": 0.5156725645065308,
|
| 566 |
+
"rewards/rejected": -0.23987337946891785,
|
| 567 |
+
"step": 31,
|
| 568 |
+
"train_speed(iter/s)": 0.005459
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.09146940598481465,
|
| 572 |
+
"grad_norm": 4.645272731781006,
|
| 573 |
+
"learning_rate": 0.00012075471698113207,
|
| 574 |
+
"logits/chosen": -0.5988867282867432,
|
| 575 |
+
"logits/rejected": -0.6016397476196289,
|
| 576 |
+
"logps/chosen": -5.776233673095703,
|
| 577 |
+
"logps/rejected": -22.404544830322266,
|
| 578 |
+
"loss": 0.8892383575439453,
|
| 579 |
+
"memory(GiB)": 46.73,
|
| 580 |
+
"nll_loss": 0.4011368155479431,
|
| 581 |
+
"rewards/accuracies": 0.75,
|
| 582 |
+
"rewards/chosen": 0.22715790569782257,
|
| 583 |
+
"rewards/margins": 0.7724226713180542,
|
| 584 |
+
"rewards/rejected": -0.545264720916748,
|
| 585 |
+
"step": 32,
|
| 586 |
+
"train_speed(iter/s)": 0.005454
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.09432782492184011,
|
| 590 |
+
"grad_norm": 4.586422443389893,
|
| 591 |
+
"learning_rate": 0.00012452830188679244,
|
| 592 |
+
"logits/chosen": -0.5782912969589233,
|
| 593 |
+
"logits/rejected": -0.5768243670463562,
|
| 594 |
+
"logps/chosen": -10.939208984375,
|
| 595 |
+
"logps/rejected": -13.486640930175781,
|
| 596 |
+
"loss": 1.1607584953308105,
|
| 597 |
+
"memory(GiB)": 46.73,
|
| 598 |
+
"nll_loss": 0.5877886414527893,
|
| 599 |
+
"rewards/accuracies": 0.65625,
|
| 600 |
+
"rewards/chosen": 0.21317529678344727,
|
| 601 |
+
"rewards/margins": 0.4141784906387329,
|
| 602 |
+
"rewards/rejected": -0.20100319385528564,
|
| 603 |
+
"step": 33,
|
| 604 |
+
"train_speed(iter/s)": 0.005457
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.09718624385886557,
|
| 608 |
+
"grad_norm": 4.306042671203613,
|
| 609 |
+
"learning_rate": 0.00012830188679245283,
|
| 610 |
+
"logits/chosen": -0.6019578576087952,
|
| 611 |
+
"logits/rejected": -0.6082974076271057,
|
| 612 |
+
"logps/chosen": -7.5170578956604,
|
| 613 |
+
"logps/rejected": -24.889406204223633,
|
| 614 |
+
"loss": 0.9509016275405884,
|
| 615 |
+
"memory(GiB)": 46.73,
|
| 616 |
+
"nll_loss": 0.4247076213359833,
|
| 617 |
+
"rewards/accuracies": 0.75,
|
| 618 |
+
"rewards/chosen": 0.10510760545730591,
|
| 619 |
+
"rewards/margins": 0.8037927150726318,
|
| 620 |
+
"rewards/rejected": -0.6986850500106812,
|
| 621 |
+
"step": 34,
|
| 622 |
+
"train_speed(iter/s)": 0.005456
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.10004466279589103,
|
| 626 |
+
"grad_norm": 3.002274751663208,
|
| 627 |
+
"learning_rate": 0.0001320754716981132,
|
| 628 |
+
"logits/chosen": -0.6133092641830444,
|
| 629 |
+
"logits/rejected": -0.6127456426620483,
|
| 630 |
+
"logps/chosen": -8.859593391418457,
|
| 631 |
+
"logps/rejected": -17.02096939086914,
|
| 632 |
+
"loss": 0.8848217129707336,
|
| 633 |
+
"memory(GiB)": 46.73,
|
| 634 |
+
"nll_loss": 0.3480006456375122,
|
| 635 |
+
"rewards/accuracies": 0.71875,
|
| 636 |
+
"rewards/chosen": 0.25863486528396606,
|
| 637 |
+
"rewards/margins": 0.5611037015914917,
|
| 638 |
+
"rewards/rejected": -0.302468866109848,
|
| 639 |
+
"step": 35,
|
| 640 |
+
"train_speed(iter/s)": 0.005457
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.10290308173291648,
|
| 644 |
+
"grad_norm": 4.299287796020508,
|
| 645 |
+
"learning_rate": 0.0001358490566037736,
|
| 646 |
+
"logits/chosen": -0.5725032091140747,
|
| 647 |
+
"logits/rejected": -0.5725142359733582,
|
| 648 |
+
"logps/chosen": -8.371686935424805,
|
| 649 |
+
"logps/rejected": -17.657852172851562,
|
| 650 |
+
"loss": 0.8825864791870117,
|
| 651 |
+
"memory(GiB)": 46.73,
|
| 652 |
+
"nll_loss": 0.3603641390800476,
|
| 653 |
+
"rewards/accuracies": 0.78125,
|
| 654 |
+
"rewards/chosen": 0.23154468834400177,
|
| 655 |
+
"rewards/margins": 0.6240193843841553,
|
| 656 |
+
"rewards/rejected": -0.3924746811389923,
|
| 657 |
+
"step": 36,
|
| 658 |
+
"train_speed(iter/s)": 0.005457
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.10576150066994194,
|
| 662 |
+
"grad_norm": 4.354315757751465,
|
| 663 |
+
"learning_rate": 0.00013962264150943395,
|
| 664 |
+
"logits/chosen": -0.5943324565887451,
|
| 665 |
+
"logits/rejected": -0.5940991044044495,
|
| 666 |
+
"logps/chosen": -7.657552719116211,
|
| 667 |
+
"logps/rejected": -16.183917999267578,
|
| 668 |
+
"loss": 0.8160735964775085,
|
| 669 |
+
"memory(GiB)": 46.73,
|
| 670 |
+
"nll_loss": 0.3588046133518219,
|
| 671 |
+
"rewards/accuracies": 0.84375,
|
| 672 |
+
"rewards/chosen": 0.27544379234313965,
|
| 673 |
+
"rewards/margins": 0.7471473217010498,
|
| 674 |
+
"rewards/rejected": -0.4717034101486206,
|
| 675 |
+
"step": 37,
|
| 676 |
+
"train_speed(iter/s)": 0.005452
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.1086199196069674,
|
| 680 |
+
"grad_norm": 7.916697978973389,
|
| 681 |
+
"learning_rate": 0.00014339622641509434,
|
| 682 |
+
"logits/chosen": -0.5676888227462769,
|
| 683 |
+
"logits/rejected": -0.5714080929756165,
|
| 684 |
+
"logps/chosen": -6.698662757873535,
|
| 685 |
+
"logps/rejected": -17.954235076904297,
|
| 686 |
+
"loss": 1.052018642425537,
|
| 687 |
+
"memory(GiB)": 46.73,
|
| 688 |
+
"nll_loss": 0.49374452233314514,
|
| 689 |
+
"rewards/accuracies": 0.78125,
|
| 690 |
+
"rewards/chosen": 0.16338011622428894,
|
| 691 |
+
"rewards/margins": 0.42942774295806885,
|
| 692 |
+
"rewards/rejected": -0.2660475969314575,
|
| 693 |
+
"step": 38,
|
| 694 |
+
"train_speed(iter/s)": 0.005452
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.11147833854399285,
|
| 698 |
+
"grad_norm": 3.7271363735198975,
|
| 699 |
+
"learning_rate": 0.00014716981132075472,
|
| 700 |
+
"logits/chosen": -0.6097766160964966,
|
| 701 |
+
"logits/rejected": -0.6135396957397461,
|
| 702 |
+
"logps/chosen": -7.30006217956543,
|
| 703 |
+
"logps/rejected": -22.168495178222656,
|
| 704 |
+
"loss": 0.7729511857032776,
|
| 705 |
+
"memory(GiB)": 46.73,
|
| 706 |
+
"nll_loss": 0.3125499188899994,
|
| 707 |
+
"rewards/accuracies": 0.78125,
|
| 708 |
+
"rewards/chosen": 0.28066012263298035,
|
| 709 |
+
"rewards/margins": 0.8108583092689514,
|
| 710 |
+
"rewards/rejected": -0.5301981568336487,
|
| 711 |
+
"step": 39,
|
| 712 |
+
"train_speed(iter/s)": 0.00545
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.1143367574810183,
|
| 716 |
+
"grad_norm": 2.960145950317383,
|
| 717 |
+
"learning_rate": 0.0001509433962264151,
|
| 718 |
+
"logits/chosen": -0.6132208108901978,
|
| 719 |
+
"logits/rejected": -0.6163071990013123,
|
| 720 |
+
"logps/chosen": -7.024638652801514,
|
| 721 |
+
"logps/rejected": -21.298852920532227,
|
| 722 |
+
"loss": 0.8106734752655029,
|
| 723 |
+
"memory(GiB)": 46.73,
|
| 724 |
+
"nll_loss": 0.37238579988479614,
|
| 725 |
+
"rewards/accuracies": 0.78125,
|
| 726 |
+
"rewards/chosen": 0.39458420872688293,
|
| 727 |
+
"rewards/margins": 0.8476477265357971,
|
| 728 |
+
"rewards/rejected": -0.45306357741355896,
|
| 729 |
+
"step": 40,
|
| 730 |
+
"train_speed(iter/s)": 0.005445
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.11719517641804376,
|
| 734 |
+
"grad_norm": 4.3203630447387695,
|
| 735 |
+
"learning_rate": 0.0001547169811320755,
|
| 736 |
+
"logits/chosen": -0.5976248979568481,
|
| 737 |
+
"logits/rejected": -0.6002449989318848,
|
| 738 |
+
"logps/chosen": -5.656615257263184,
|
| 739 |
+
"logps/rejected": -16.032827377319336,
|
| 740 |
+
"loss": 0.9440045952796936,
|
| 741 |
+
"memory(GiB)": 46.73,
|
| 742 |
+
"nll_loss": 0.33253738284111023,
|
| 743 |
+
"rewards/accuracies": 0.59375,
|
| 744 |
+
"rewards/chosen": 0.14414586126804352,
|
| 745 |
+
"rewards/margins": 0.5006409287452698,
|
| 746 |
+
"rewards/rejected": -0.35649505257606506,
|
| 747 |
+
"step": 41,
|
| 748 |
+
"train_speed(iter/s)": 0.005445
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.12005359535506922,
|
| 752 |
+
"grad_norm": 4.646491527557373,
|
| 753 |
+
"learning_rate": 0.00015849056603773587,
|
| 754 |
+
"logits/chosen": -0.5956608653068542,
|
| 755 |
+
"logits/rejected": -0.5956603288650513,
|
| 756 |
+
"logps/chosen": -11.719463348388672,
|
| 757 |
+
"logps/rejected": -27.070579528808594,
|
| 758 |
+
"loss": 1.0076483488082886,
|
| 759 |
+
"memory(GiB)": 46.73,
|
| 760 |
+
"nll_loss": 0.5070124268531799,
|
| 761 |
+
"rewards/accuracies": 0.71875,
|
| 762 |
+
"rewards/chosen": 0.15291725099086761,
|
| 763 |
+
"rewards/margins": 0.7710469961166382,
|
| 764 |
+
"rewards/rejected": -0.6181297898292542,
|
| 765 |
+
"step": 42,
|
| 766 |
+
"train_speed(iter/s)": 0.005446
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.12291201429209468,
|
| 770 |
+
"grad_norm": 4.248917102813721,
|
| 771 |
+
"learning_rate": 0.00016226415094339625,
|
| 772 |
+
"logits/chosen": -0.6100150942802429,
|
| 773 |
+
"logits/rejected": -0.6142972111701965,
|
| 774 |
+
"logps/chosen": -6.201865196228027,
|
| 775 |
+
"logps/rejected": -19.82131004333496,
|
| 776 |
+
"loss": 0.7969534993171692,
|
| 777 |
+
"memory(GiB)": 46.73,
|
| 778 |
+
"nll_loss": 0.31225427985191345,
|
| 779 |
+
"rewards/accuracies": 0.6875,
|
| 780 |
+
"rewards/chosen": 0.20218679308891296,
|
| 781 |
+
"rewards/margins": 0.705444872379303,
|
| 782 |
+
"rewards/rejected": -0.5032580494880676,
|
| 783 |
+
"step": 43,
|
| 784 |
+
"train_speed(iter/s)": 0.005448
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.12577043322912015,
|
| 788 |
+
"grad_norm": 3.1864066123962402,
|
| 789 |
+
"learning_rate": 0.0001660377358490566,
|
| 790 |
+
"logits/chosen": -0.6078452467918396,
|
| 791 |
+
"logits/rejected": -0.6100336909294128,
|
| 792 |
+
"logps/chosen": -8.217609405517578,
|
| 793 |
+
"logps/rejected": -18.432086944580078,
|
| 794 |
+
"loss": 0.945781409740448,
|
| 795 |
+
"memory(GiB)": 46.73,
|
| 796 |
+
"nll_loss": 0.4447736442089081,
|
| 797 |
+
"rewards/accuracies": 0.8125,
|
| 798 |
+
"rewards/chosen": 0.16747131943702698,
|
| 799 |
+
"rewards/margins": 0.6352076530456543,
|
| 800 |
+
"rewards/rejected": -0.46773630380630493,
|
| 801 |
+
"step": 44,
|
| 802 |
+
"train_speed(iter/s)": 0.005451
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.1286288521661456,
|
| 806 |
+
"grad_norm": 2.8247904777526855,
|
| 807 |
+
"learning_rate": 0.000169811320754717,
|
| 808 |
+
"logits/chosen": -0.6424505710601807,
|
| 809 |
+
"logits/rejected": -0.6455634236335754,
|
| 810 |
+
"logps/chosen": -4.626930236816406,
|
| 811 |
+
"logps/rejected": -18.966289520263672,
|
| 812 |
+
"loss": 0.6525614261627197,
|
| 813 |
+
"memory(GiB)": 46.73,
|
| 814 |
+
"nll_loss": 0.20949943363666534,
|
| 815 |
+
"rewards/accuracies": 0.90625,
|
| 816 |
+
"rewards/chosen": 0.1935967355966568,
|
| 817 |
+
"rewards/margins": 0.7181073427200317,
|
| 818 |
+
"rewards/rejected": -0.5245105624198914,
|
| 819 |
+
"step": 45,
|
| 820 |
+
"train_speed(iter/s)": 0.005449
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.13148727110317107,
|
| 824 |
+
"grad_norm": 5.167090892791748,
|
| 825 |
+
"learning_rate": 0.00017358490566037738,
|
| 826 |
+
"logits/chosen": -0.6390554308891296,
|
| 827 |
+
"logits/rejected": -0.6455981731414795,
|
| 828 |
+
"logps/chosen": -7.457905292510986,
|
| 829 |
+
"logps/rejected": -22.238759994506836,
|
| 830 |
+
"loss": 0.7052859663963318,
|
| 831 |
+
"memory(GiB)": 46.73,
|
| 832 |
+
"nll_loss": 0.3145278990268707,
|
| 833 |
+
"rewards/accuracies": 0.90625,
|
| 834 |
+
"rewards/chosen": 0.3406547009944916,
|
| 835 |
+
"rewards/margins": 0.9184964895248413,
|
| 836 |
+
"rewards/rejected": -0.5778417587280273,
|
| 837 |
+
"step": 46,
|
| 838 |
+
"train_speed(iter/s)": 0.005446
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.1343456900401965,
|
| 842 |
+
"grad_norm": 3.110797166824341,
|
| 843 |
+
"learning_rate": 0.00017735849056603776,
|
| 844 |
+
"logits/chosen": -0.6957284212112427,
|
| 845 |
+
"logits/rejected": -0.6984925866127014,
|
| 846 |
+
"logps/chosen": -6.335500717163086,
|
| 847 |
+
"logps/rejected": -26.946069717407227,
|
| 848 |
+
"loss": 0.6684616804122925,
|
| 849 |
+
"memory(GiB)": 46.73,
|
| 850 |
+
"nll_loss": 0.27696001529693604,
|
| 851 |
+
"rewards/accuracies": 0.78125,
|
| 852 |
+
"rewards/chosen": 0.36519789695739746,
|
| 853 |
+
"rewards/margins": 1.261197805404663,
|
| 854 |
+
"rewards/rejected": -0.8960000276565552,
|
| 855 |
+
"step": 47,
|
| 856 |
+
"train_speed(iter/s)": 0.005445
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.13720410897722196,
|
| 860 |
+
"grad_norm": 5.592209815979004,
|
| 861 |
+
"learning_rate": 0.00018113207547169812,
|
| 862 |
+
"logits/chosen": -0.6557337641716003,
|
| 863 |
+
"logits/rejected": -0.6595371961593628,
|
| 864 |
+
"logps/chosen": -6.262232780456543,
|
| 865 |
+
"logps/rejected": -22.550495147705078,
|
| 866 |
+
"loss": 0.844598650932312,
|
| 867 |
+
"memory(GiB)": 46.73,
|
| 868 |
+
"nll_loss": 0.32042089104652405,
|
| 869 |
+
"rewards/accuracies": 0.71875,
|
| 870 |
+
"rewards/chosen": 0.098446786403656,
|
| 871 |
+
"rewards/margins": 1.1042643785476685,
|
| 872 |
+
"rewards/rejected": -1.0058175325393677,
|
| 873 |
+
"step": 48,
|
| 874 |
+
"train_speed(iter/s)": 0.005444
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.14006252791424742,
|
| 878 |
+
"grad_norm": 5.518631458282471,
|
| 879 |
+
"learning_rate": 0.0001849056603773585,
|
| 880 |
+
"logits/chosen": -0.7261903285980225,
|
| 881 |
+
"logits/rejected": -0.7333111763000488,
|
| 882 |
+
"logps/chosen": -6.155282020568848,
|
| 883 |
+
"logps/rejected": -28.635177612304688,
|
| 884 |
+
"loss": 0.7839219570159912,
|
| 885 |
+
"memory(GiB)": 46.73,
|
| 886 |
+
"nll_loss": 0.3167538046836853,
|
| 887 |
+
"rewards/accuracies": 0.75,
|
| 888 |
+
"rewards/chosen": 0.18123595416545868,
|
| 889 |
+
"rewards/margins": 1.343546986579895,
|
| 890 |
+
"rewards/rejected": -1.1623109579086304,
|
| 891 |
+
"step": 49,
|
| 892 |
+
"train_speed(iter/s)": 0.00544
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.14292094685127288,
|
| 896 |
+
"grad_norm": 5.368433475494385,
|
| 897 |
+
"learning_rate": 0.00018867924528301889,
|
| 898 |
+
"logits/chosen": -0.7460846304893494,
|
| 899 |
+
"logits/rejected": -0.7433331608772278,
|
| 900 |
+
"logps/chosen": -8.476225852966309,
|
| 901 |
+
"logps/rejected": -20.479339599609375,
|
| 902 |
+
"loss": 1.041689395904541,
|
| 903 |
+
"memory(GiB)": 46.73,
|
| 904 |
+
"nll_loss": 0.5162184834480286,
|
| 905 |
+
"rewards/accuracies": 0.6875,
|
| 906 |
+
"rewards/chosen": 0.07575411349534988,
|
| 907 |
+
"rewards/margins": 0.8544278144836426,
|
| 908 |
+
"rewards/rejected": -0.7786736488342285,
|
| 909 |
+
"step": 50,
|
| 910 |
+
"train_speed(iter/s)": 0.005439
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.14577936578829834,
|
| 914 |
+
"grad_norm": 12.666760444641113,
|
| 915 |
+
"learning_rate": 0.00019245283018867927,
|
| 916 |
+
"logits/chosen": -0.7423916459083557,
|
| 917 |
+
"logits/rejected": -0.7536158561706543,
|
| 918 |
+
"logps/chosen": -4.518327713012695,
|
| 919 |
+
"logps/rejected": -25.71565818786621,
|
| 920 |
+
"loss": 0.8285016417503357,
|
| 921 |
+
"memory(GiB)": 46.73,
|
| 922 |
+
"nll_loss": 0.38353338837623596,
|
| 923 |
+
"rewards/accuracies": 0.6875,
|
| 924 |
+
"rewards/chosen": 0.25002920627593994,
|
| 925 |
+
"rewards/margins": 1.0534642934799194,
|
| 926 |
+
"rewards/rejected": -0.8034350872039795,
|
| 927 |
+
"step": 51,
|
| 928 |
+
"train_speed(iter/s)": 0.005432
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.1486377847253238,
|
| 932 |
+
"grad_norm": 5.63993501663208,
|
| 933 |
+
"learning_rate": 0.00019622641509433963,
|
| 934 |
+
"logits/chosen": -0.7661195993423462,
|
| 935 |
+
"logits/rejected": -0.7684649229049683,
|
| 936 |
+
"logps/chosen": -5.303972244262695,
|
| 937 |
+
"logps/rejected": -20.424297332763672,
|
| 938 |
+
"loss": 0.7760183811187744,
|
| 939 |
+
"memory(GiB)": 46.73,
|
| 940 |
+
"nll_loss": 0.30550119280815125,
|
| 941 |
+
"rewards/accuracies": 0.84375,
|
| 942 |
+
"rewards/chosen": 0.17865437269210815,
|
| 943 |
+
"rewards/margins": 0.9522275328636169,
|
| 944 |
+
"rewards/rejected": -0.7735730409622192,
|
| 945 |
+
"step": 52,
|
| 946 |
+
"train_speed(iter/s)": 0.005436
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.15149620366234925,
|
| 950 |
+
"grad_norm": 3.208770275115967,
|
| 951 |
+
"learning_rate": 0.0002,
|
| 952 |
+
"logits/chosen": -0.7615375518798828,
|
| 953 |
+
"logits/rejected": -0.7639837861061096,
|
| 954 |
+
"logps/chosen": -7.793338775634766,
|
| 955 |
+
"logps/rejected": -22.267044067382812,
|
| 956 |
+
"loss": 0.6889612078666687,
|
| 957 |
+
"memory(GiB)": 46.73,
|
| 958 |
+
"nll_loss": 0.2784683108329773,
|
| 959 |
+
"rewards/accuracies": 0.78125,
|
| 960 |
+
"rewards/chosen": 0.21318240463733673,
|
| 961 |
+
"rewards/margins": 1.1274638175964355,
|
| 962 |
+
"rewards/rejected": -0.9142814874649048,
|
| 963 |
+
"step": 53,
|
| 964 |
+
"train_speed(iter/s)": 0.005436
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.1543546225993747,
|
| 968 |
+
"grad_norm": 4.603616714477539,
|
| 969 |
+
"learning_rate": 0.00019999950054470762,
|
| 970 |
+
"logits/chosen": -0.7416525483131409,
|
| 971 |
+
"logits/rejected": -0.7468435168266296,
|
| 972 |
+
"logps/chosen": -11.753125190734863,
|
| 973 |
+
"logps/rejected": -23.24603271484375,
|
| 974 |
+
"loss": 0.9534075856208801,
|
| 975 |
+
"memory(GiB)": 46.73,
|
| 976 |
+
"nll_loss": 0.42802760004997253,
|
| 977 |
+
"rewards/accuracies": 0.65625,
|
| 978 |
+
"rewards/chosen": 0.2138495296239853,
|
| 979 |
+
"rewards/margins": 0.8852367997169495,
|
| 980 |
+
"rewards/rejected": -0.6713871955871582,
|
| 981 |
+
"step": 54,
|
| 982 |
+
"train_speed(iter/s)": 0.005437
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.15721304153640017,
|
| 986 |
+
"grad_norm": 3.226914405822754,
|
| 987 |
+
"learning_rate": 0.00019999800218381956,
|
| 988 |
+
"logits/chosen": -0.782919704914093,
|
| 989 |
+
"logits/rejected": -0.7837254405021667,
|
| 990 |
+
"logps/chosen": -10.437933921813965,
|
| 991 |
+
"logps/rejected": -21.950359344482422,
|
| 992 |
+
"loss": 0.7319959998130798,
|
| 993 |
+
"memory(GiB)": 46.73,
|
| 994 |
+
"nll_loss": 0.28649967908859253,
|
| 995 |
+
"rewards/accuracies": 0.78125,
|
| 996 |
+
"rewards/chosen": 0.17146320641040802,
|
| 997 |
+
"rewards/margins": 0.9873074889183044,
|
| 998 |
+
"rewards/rejected": -0.8158442974090576,
|
| 999 |
+
"step": 55,
|
| 1000 |
+
"train_speed(iter/s)": 0.005435
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.16007146047342563,
|
| 1004 |
+
"grad_norm": 5.957598686218262,
|
| 1005 |
+
"learning_rate": 0.00019999550493230315,
|
| 1006 |
+
"logits/chosen": -0.7622566819190979,
|
| 1007 |
+
"logits/rejected": -0.7720569968223572,
|
| 1008 |
+
"logps/chosen": -9.299139022827148,
|
| 1009 |
+
"logps/rejected": -25.902114868164062,
|
| 1010 |
+
"loss": 1.0733510255813599,
|
| 1011 |
+
"memory(GiB)": 46.73,
|
| 1012 |
+
"nll_loss": 0.5738880634307861,
|
| 1013 |
+
"rewards/accuracies": 0.75,
|
| 1014 |
+
"rewards/chosen": 0.1327313929796219,
|
| 1015 |
+
"rewards/margins": 1.0297049283981323,
|
| 1016 |
+
"rewards/rejected": -0.8969735503196716,
|
| 1017 |
+
"step": 56,
|
| 1018 |
+
"train_speed(iter/s)": 0.005434
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.1629298794104511,
|
| 1022 |
+
"grad_norm": 4.4198760986328125,
|
| 1023 |
+
"learning_rate": 0.00019999200881510367,
|
| 1024 |
+
"logits/chosen": -0.7777268886566162,
|
| 1025 |
+
"logits/rejected": -0.7817560434341431,
|
| 1026 |
+
"logps/chosen": -8.051362037658691,
|
| 1027 |
+
"logps/rejected": -16.21352767944336,
|
| 1028 |
+
"loss": 0.9994087815284729,
|
| 1029 |
+
"memory(GiB)": 46.73,
|
| 1030 |
+
"nll_loss": 0.5157659649848938,
|
| 1031 |
+
"rewards/accuracies": 0.71875,
|
| 1032 |
+
"rewards/chosen": 0.3178602159023285,
|
| 1033 |
+
"rewards/margins": 0.7738386392593384,
|
| 1034 |
+
"rewards/rejected": -0.4559784233570099,
|
| 1035 |
+
"step": 57,
|
| 1036 |
+
"train_speed(iter/s)": 0.005437
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.16578829834747655,
|
| 1040 |
+
"grad_norm": 10.918800354003906,
|
| 1041 |
+
"learning_rate": 0.00019998751386714417,
|
| 1042 |
+
"logits/chosen": -0.7735524773597717,
|
| 1043 |
+
"logits/rejected": -0.7762777209281921,
|
| 1044 |
+
"logps/chosen": -6.026726245880127,
|
| 1045 |
+
"logps/rejected": -20.445030212402344,
|
| 1046 |
+
"loss": 0.8424608707427979,
|
| 1047 |
+
"memory(GiB)": 46.73,
|
| 1048 |
+
"nll_loss": 0.4571954309940338,
|
| 1049 |
+
"rewards/accuracies": 0.875,
|
| 1050 |
+
"rewards/chosen": 0.3362487852573395,
|
| 1051 |
+
"rewards/margins": 1.0223908424377441,
|
| 1052 |
+
"rewards/rejected": -0.6861420273780823,
|
| 1053 |
+
"step": 58,
|
| 1054 |
+
"train_speed(iter/s)": 0.005438
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.168646717284502,
|
| 1058 |
+
"grad_norm": 6.003457546234131,
|
| 1059 |
+
"learning_rate": 0.00019998202013332526,
|
| 1060 |
+
"logits/chosen": -0.7671990394592285,
|
| 1061 |
+
"logits/rejected": -0.7679510712623596,
|
| 1062 |
+
"logps/chosen": -7.208020210266113,
|
| 1063 |
+
"logps/rejected": -16.45644187927246,
|
| 1064 |
+
"loss": 0.8400717377662659,
|
| 1065 |
+
"memory(GiB)": 46.73,
|
| 1066 |
+
"nll_loss": 0.3789638876914978,
|
| 1067 |
+
"rewards/accuracies": 0.8125,
|
| 1068 |
+
"rewards/chosen": 0.19236475229263306,
|
| 1069 |
+
"rewards/margins": 0.8348343372344971,
|
| 1070 |
+
"rewards/rejected": -0.6424695253372192,
|
| 1071 |
+
"step": 59,
|
| 1072 |
+
"train_speed(iter/s)": 0.005439
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 0.17150513622152747,
|
| 1076 |
+
"grad_norm": 2.5898213386535645,
|
| 1077 |
+
"learning_rate": 0.00019997552766852432,
|
| 1078 |
+
"logits/chosen": -0.800130307674408,
|
| 1079 |
+
"logits/rejected": -0.8043981790542603,
|
| 1080 |
+
"logps/chosen": -6.655481815338135,
|
| 1081 |
+
"logps/rejected": -24.4234619140625,
|
| 1082 |
+
"loss": 0.6945982575416565,
|
| 1083 |
+
"memory(GiB)": 46.73,
|
| 1084 |
+
"nll_loss": 0.3340861201286316,
|
| 1085 |
+
"rewards/accuracies": 0.875,
|
| 1086 |
+
"rewards/chosen": 0.3088517487049103,
|
| 1087 |
+
"rewards/margins": 1.2354176044464111,
|
| 1088 |
+
"rewards/rejected": -0.9265658855438232,
|
| 1089 |
+
"step": 60,
|
| 1090 |
+
"train_speed(iter/s)": 0.005436
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.17436355515855292,
|
| 1094 |
+
"grad_norm": 3.626788377761841,
|
| 1095 |
+
"learning_rate": 0.00019996803653759532,
|
| 1096 |
+
"logits/chosen": -0.7841392159461975,
|
| 1097 |
+
"logits/rejected": -0.7904000282287598,
|
| 1098 |
+
"logps/chosen": -4.378767013549805,
|
| 1099 |
+
"logps/rejected": -26.130855560302734,
|
| 1100 |
+
"loss": 0.5890128016471863,
|
| 1101 |
+
"memory(GiB)": 46.73,
|
| 1102 |
+
"nll_loss": 0.2802823781967163,
|
| 1103 |
+
"rewards/accuracies": 0.9375,
|
| 1104 |
+
"rewards/chosen": 0.2776252329349518,
|
| 1105 |
+
"rewards/margins": 1.491828203201294,
|
| 1106 |
+
"rewards/rejected": -1.2142030000686646,
|
| 1107 |
+
"step": 61,
|
| 1108 |
+
"train_speed(iter/s)": 0.005438
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.17722197409557838,
|
| 1112 |
+
"grad_norm": 6.747097492218018,
|
| 1113 |
+
"learning_rate": 0.00019995954681536798,
|
| 1114 |
+
"logits/chosen": -0.7469812035560608,
|
| 1115 |
+
"logits/rejected": -0.7426640391349792,
|
| 1116 |
+
"logps/chosen": -11.407801628112793,
|
| 1117 |
+
"logps/rejected": -19.764530181884766,
|
| 1118 |
+
"loss": 1.1509681940078735,
|
| 1119 |
+
"memory(GiB)": 46.73,
|
| 1120 |
+
"nll_loss": 0.4816472828388214,
|
| 1121 |
+
"rewards/accuracies": 0.71875,
|
| 1122 |
+
"rewards/chosen": -0.023781299591064453,
|
| 1123 |
+
"rewards/margins": 0.7125381231307983,
|
| 1124 |
+
"rewards/rejected": -0.7363194227218628,
|
| 1125 |
+
"step": 62,
|
| 1126 |
+
"train_speed(iter/s)": 0.005439
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 0.18008039303260384,
|
| 1130 |
+
"grad_norm": 6.9256978034973145,
|
| 1131 |
+
"learning_rate": 0.00019995005858664696,
|
| 1132 |
+
"logits/chosen": -0.819698691368103,
|
| 1133 |
+
"logits/rejected": -0.8221041560173035,
|
| 1134 |
+
"logps/chosen": -7.0722761154174805,
|
| 1135 |
+
"logps/rejected": -30.66091537475586,
|
| 1136 |
+
"loss": 0.9850404262542725,
|
| 1137 |
+
"memory(GiB)": 46.73,
|
| 1138 |
+
"nll_loss": 0.42992380261421204,
|
| 1139 |
+
"rewards/accuracies": 0.75,
|
| 1140 |
+
"rewards/chosen": 0.07954245805740356,
|
| 1141 |
+
"rewards/margins": 1.697967767715454,
|
| 1142 |
+
"rewards/rejected": -1.6184254884719849,
|
| 1143 |
+
"step": 63,
|
| 1144 |
+
"train_speed(iter/s)": 0.005439
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.1829388119696293,
|
| 1148 |
+
"grad_norm": 5.373042106628418,
|
| 1149 |
+
"learning_rate": 0.00019993957194621128,
|
| 1150 |
+
"logits/chosen": -0.8020437955856323,
|
| 1151 |
+
"logits/rejected": -0.8022991418838501,
|
| 1152 |
+
"logps/chosen": -12.086634635925293,
|
| 1153 |
+
"logps/rejected": -21.337501525878906,
|
| 1154 |
+
"loss": 0.9992907643318176,
|
| 1155 |
+
"memory(GiB)": 46.73,
|
| 1156 |
+
"nll_loss": 0.46714556217193604,
|
| 1157 |
+
"rewards/accuracies": 0.78125,
|
| 1158 |
+
"rewards/chosen": 0.11581222712993622,
|
| 1159 |
+
"rewards/margins": 1.0597875118255615,
|
| 1160 |
+
"rewards/rejected": -0.9439753293991089,
|
| 1161 |
+
"step": 64,
|
| 1162 |
+
"train_speed(iter/s)": 0.005441
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.18579723090665476,
|
| 1166 |
+
"grad_norm": 3.77294921875,
|
| 1167 |
+
"learning_rate": 0.00019992808699881303,
|
| 1168 |
+
"logits/chosen": -0.7432379722595215,
|
| 1169 |
+
"logits/rejected": -0.7440240979194641,
|
| 1170 |
+
"logps/chosen": -6.54131555557251,
|
| 1171 |
+
"logps/rejected": -25.4437255859375,
|
| 1172 |
+
"loss": 0.8025262951850891,
|
| 1173 |
+
"memory(GiB)": 46.73,
|
| 1174 |
+
"nll_loss": 0.4345034658908844,
|
| 1175 |
+
"rewards/accuracies": 0.90625,
|
| 1176 |
+
"rewards/chosen": 0.1653081476688385,
|
| 1177 |
+
"rewards/margins": 1.261300802230835,
|
| 1178 |
+
"rewards/rejected": -1.0959926843643188,
|
| 1179 |
+
"step": 65,
|
| 1180 |
+
"train_speed(iter/s)": 0.005442
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.18865564984368022,
|
| 1184 |
+
"grad_norm": 3.9820404052734375,
|
| 1185 |
+
"learning_rate": 0.0001999156038591766,
|
| 1186 |
+
"logits/chosen": -0.754156231880188,
|
| 1187 |
+
"logits/rejected": -0.7619881629943848,
|
| 1188 |
+
"logps/chosen": -6.165823459625244,
|
| 1189 |
+
"logps/rejected": -23.689210891723633,
|
| 1190 |
+
"loss": 0.7558284997940063,
|
| 1191 |
+
"memory(GiB)": 46.73,
|
| 1192 |
+
"nll_loss": 0.35462892055511475,
|
| 1193 |
+
"rewards/accuracies": 0.84375,
|
| 1194 |
+
"rewards/chosen": 0.18967051804065704,
|
| 1195 |
+
"rewards/margins": 0.9685844779014587,
|
| 1196 |
+
"rewards/rejected": -0.7789139151573181,
|
| 1197 |
+
"step": 66,
|
| 1198 |
+
"train_speed(iter/s)": 0.005441
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.19151406878070568,
|
| 1202 |
+
"grad_norm": 4.361135482788086,
|
| 1203 |
+
"learning_rate": 0.00019990212265199738,
|
| 1204 |
+
"logits/chosen": -0.7575560212135315,
|
| 1205 |
+
"logits/rejected": -0.7635327577590942,
|
| 1206 |
+
"logps/chosen": -4.563057899475098,
|
| 1207 |
+
"logps/rejected": -24.856014251708984,
|
| 1208 |
+
"loss": 0.7012404203414917,
|
| 1209 |
+
"memory(GiB)": 46.73,
|
| 1210 |
+
"nll_loss": 0.3292723000049591,
|
| 1211 |
+
"rewards/accuracies": 0.84375,
|
| 1212 |
+
"rewards/chosen": 0.360738068819046,
|
| 1213 |
+
"rewards/margins": 1.0550967454910278,
|
| 1214 |
+
"rewards/rejected": -0.6943586468696594,
|
| 1215 |
+
"step": 67,
|
| 1216 |
+
"train_speed(iter/s)": 0.005441
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.19437248771773113,
|
| 1220 |
+
"grad_norm": 8.01318359375,
|
| 1221 |
+
"learning_rate": 0.00019988764351194056,
|
| 1222 |
+
"logits/chosen": -0.7201259136199951,
|
| 1223 |
+
"logits/rejected": -0.7213173508644104,
|
| 1224 |
+
"logps/chosen": -14.019023895263672,
|
| 1225 |
+
"logps/rejected": -23.463457107543945,
|
| 1226 |
+
"loss": 1.155617594718933,
|
| 1227 |
+
"memory(GiB)": 46.73,
|
| 1228 |
+
"nll_loss": 0.6643218994140625,
|
| 1229 |
+
"rewards/accuracies": 0.78125,
|
| 1230 |
+
"rewards/chosen": 0.16900399327278137,
|
| 1231 |
+
"rewards/margins": 0.8244603276252747,
|
| 1232 |
+
"rewards/rejected": -0.6554563641548157,
|
| 1233 |
+
"step": 68,
|
| 1234 |
+
"train_speed(iter/s)": 0.005442
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.1972309066547566,
|
| 1238 |
+
"grad_norm": 4.880486488342285,
|
| 1239 |
+
"learning_rate": 0.00019987216658363984,
|
| 1240 |
+
"logits/chosen": -0.7603176832199097,
|
| 1241 |
+
"logits/rejected": -0.7611086368560791,
|
| 1242 |
+
"logps/chosen": -7.871481418609619,
|
| 1243 |
+
"logps/rejected": -18.355331420898438,
|
| 1244 |
+
"loss": 0.9050899744033813,
|
| 1245 |
+
"memory(GiB)": 46.73,
|
| 1246 |
+
"nll_loss": 0.4671905040740967,
|
| 1247 |
+
"rewards/accuracies": 0.90625,
|
| 1248 |
+
"rewards/chosen": 0.14834776520729065,
|
| 1249 |
+
"rewards/margins": 0.9282243251800537,
|
| 1250 |
+
"rewards/rejected": -0.7798765897750854,
|
| 1251 |
+
"step": 69,
|
| 1252 |
+
"train_speed(iter/s)": 0.005441
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.20008932559178205,
|
| 1256 |
+
"grad_norm": 36.09653854370117,
|
| 1257 |
+
"learning_rate": 0.00019985569202169584,
|
| 1258 |
+
"logits/chosen": -0.826109766960144,
|
| 1259 |
+
"logits/rejected": -0.8319188952445984,
|
| 1260 |
+
"logps/chosen": -7.292363166809082,
|
| 1261 |
+
"logps/rejected": -19.279315948486328,
|
| 1262 |
+
"loss": 0.906982421875,
|
| 1263 |
+
"memory(GiB)": 46.73,
|
| 1264 |
+
"nll_loss": 0.4692971408367157,
|
| 1265 |
+
"rewards/accuracies": 0.8125,
|
| 1266 |
+
"rewards/chosen": 0.17047010362148285,
|
| 1267 |
+
"rewards/margins": 0.9899492859840393,
|
| 1268 |
+
"rewards/rejected": -0.8194791674613953,
|
| 1269 |
+
"step": 70,
|
| 1270 |
+
"train_speed(iter/s)": 0.005443
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.2029477445288075,
|
| 1274 |
+
"grad_norm": 5.051456451416016,
|
| 1275 |
+
"learning_rate": 0.00019983821999067475,
|
| 1276 |
+
"logits/chosen": -0.7659405469894409,
|
| 1277 |
+
"logits/rejected": -0.773281991481781,
|
| 1278 |
+
"logps/chosen": -7.996345520019531,
|
| 1279 |
+
"logps/rejected": -16.987951278686523,
|
| 1280 |
+
"loss": 0.9166132211685181,
|
| 1281 |
+
"memory(GiB)": 46.73,
|
| 1282 |
+
"nll_loss": 0.43807780742645264,
|
| 1283 |
+
"rewards/accuracies": 0.71875,
|
| 1284 |
+
"rewards/chosen": 0.2428373396396637,
|
| 1285 |
+
"rewards/margins": 0.7172515988349915,
|
| 1286 |
+
"rewards/rejected": -0.47441428899765015,
|
| 1287 |
+
"step": 71,
|
| 1288 |
+
"train_speed(iter/s)": 0.005444
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 0.20580616346583297,
|
| 1292 |
+
"grad_norm": 3.3546969890594482,
|
| 1293 |
+
"learning_rate": 0.00019981975066510655,
|
| 1294 |
+
"logits/chosen": -0.7702159285545349,
|
| 1295 |
+
"logits/rejected": -0.7756153345108032,
|
| 1296 |
+
"logps/chosen": -3.096975088119507,
|
| 1297 |
+
"logps/rejected": -24.7138671875,
|
| 1298 |
+
"loss": 0.6350318193435669,
|
| 1299 |
+
"memory(GiB)": 46.73,
|
| 1300 |
+
"nll_loss": 0.20447328686714172,
|
| 1301 |
+
"rewards/accuracies": 0.8125,
|
| 1302 |
+
"rewards/chosen": 0.23610374331474304,
|
| 1303 |
+
"rewards/margins": 1.1568939685821533,
|
| 1304 |
+
"rewards/rejected": -0.9207903146743774,
|
| 1305 |
+
"step": 72,
|
| 1306 |
+
"train_speed(iter/s)": 0.005441
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.20866458240285843,
|
| 1310 |
+
"grad_norm": 3.3056812286376953,
|
| 1311 |
+
"learning_rate": 0.00019980028422948322,
|
| 1312 |
+
"logits/chosen": -0.7785294651985168,
|
| 1313 |
+
"logits/rejected": -0.7796329259872437,
|
| 1314 |
+
"logps/chosen": -6.459807395935059,
|
| 1315 |
+
"logps/rejected": -18.2646541595459,
|
| 1316 |
+
"loss": 0.7826670408248901,
|
| 1317 |
+
"memory(GiB)": 46.73,
|
| 1318 |
+
"nll_loss": 0.299204021692276,
|
| 1319 |
+
"rewards/accuracies": 0.71875,
|
| 1320 |
+
"rewards/chosen": 0.17048794031143188,
|
| 1321 |
+
"rewards/margins": 0.8538780212402344,
|
| 1322 |
+
"rewards/rejected": -0.6833901405334473,
|
| 1323 |
+
"step": 73,
|
| 1324 |
+
"train_speed(iter/s)": 0.005441
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.2115230013398839,
|
| 1328 |
+
"grad_norm": 2.804248094558716,
|
| 1329 |
+
"learning_rate": 0.00019977982087825713,
|
| 1330 |
+
"logits/chosen": -0.7178574204444885,
|
| 1331 |
+
"logits/rejected": -0.7222320437431335,
|
| 1332 |
+
"logps/chosen": -5.382746696472168,
|
| 1333 |
+
"logps/rejected": -21.648345947265625,
|
| 1334 |
+
"loss": 0.6176108717918396,
|
| 1335 |
+
"memory(GiB)": 46.73,
|
| 1336 |
+
"nll_loss": 0.2642488479614258,
|
| 1337 |
+
"rewards/accuracies": 0.90625,
|
| 1338 |
+
"rewards/chosen": 0.3241485357284546,
|
| 1339 |
+
"rewards/margins": 1.1783626079559326,
|
| 1340 |
+
"rewards/rejected": -0.854214072227478,
|
| 1341 |
+
"step": 74,
|
| 1342 |
+
"train_speed(iter/s)": 0.005441
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.21438142027690935,
|
| 1346 |
+
"grad_norm": 2.3271028995513916,
|
| 1347 |
+
"learning_rate": 0.0001997583608158388,
|
| 1348 |
+
"logits/chosen": -0.7665284872055054,
|
| 1349 |
+
"logits/rejected": -0.7763013243675232,
|
| 1350 |
+
"logps/chosen": -3.597620725631714,
|
| 1351 |
+
"logps/rejected": -28.286640167236328,
|
| 1352 |
+
"loss": 0.4727986454963684,
|
| 1353 |
+
"memory(GiB)": 46.73,
|
| 1354 |
+
"nll_loss": 0.1368505358695984,
|
| 1355 |
+
"rewards/accuracies": 0.90625,
|
| 1356 |
+
"rewards/chosen": 0.36978626251220703,
|
| 1357 |
+
"rewards/margins": 1.460551142692566,
|
| 1358 |
+
"rewards/rejected": -1.0907649993896484,
|
| 1359 |
+
"step": 75,
|
| 1360 |
+
"train_speed(iter/s)": 0.005442
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.2172398392139348,
|
| 1364 |
+
"grad_norm": 5.8310160636901855,
|
| 1365 |
+
"learning_rate": 0.00019973590425659507,
|
| 1366 |
+
"logits/chosen": -0.7936246991157532,
|
| 1367 |
+
"logits/rejected": -0.7950795888900757,
|
| 1368 |
+
"logps/chosen": -8.558622360229492,
|
| 1369 |
+
"logps/rejected": -22.39399528503418,
|
| 1370 |
+
"loss": 0.7566131949424744,
|
| 1371 |
+
"memory(GiB)": 46.73,
|
| 1372 |
+
"nll_loss": 0.4084041714668274,
|
| 1373 |
+
"rewards/accuracies": 0.875,
|
| 1374 |
+
"rewards/chosen": 0.26708710193634033,
|
| 1375 |
+
"rewards/margins": 1.3221920728683472,
|
| 1376 |
+
"rewards/rejected": -1.0551048517227173,
|
| 1377 |
+
"step": 76,
|
| 1378 |
+
"train_speed(iter/s)": 0.00544
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.22009825815096026,
|
| 1382 |
+
"grad_norm": 4.89293909072876,
|
| 1383 |
+
"learning_rate": 0.0001997124514248469,
|
| 1384 |
+
"logits/chosen": -0.8093288540840149,
|
| 1385 |
+
"logits/rejected": -0.8106516003608704,
|
| 1386 |
+
"logps/chosen": -12.884543418884277,
|
| 1387 |
+
"logps/rejected": -22.5157413482666,
|
| 1388 |
+
"loss": 1.130676507949829,
|
| 1389 |
+
"memory(GiB)": 46.73,
|
| 1390 |
+
"nll_loss": 0.7112829685211182,
|
| 1391 |
+
"rewards/accuracies": 0.8125,
|
| 1392 |
+
"rewards/chosen": 0.10567857325077057,
|
| 1393 |
+
"rewards/margins": 1.2312006950378418,
|
| 1394 |
+
"rewards/rejected": -1.1255221366882324,
|
| 1395 |
+
"step": 77,
|
| 1396 |
+
"train_speed(iter/s)": 0.005441
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.2229566770879857,
|
| 1400 |
+
"grad_norm": 7.881620407104492,
|
| 1401 |
+
"learning_rate": 0.00019968800255486713,
|
| 1402 |
+
"logits/chosen": -0.8169571161270142,
|
| 1403 |
+
"logits/rejected": -0.8179275393486023,
|
| 1404 |
+
"logps/chosen": -6.7093825340271,
|
| 1405 |
+
"logps/rejected": -29.32290267944336,
|
| 1406 |
+
"loss": 0.8154858350753784,
|
| 1407 |
+
"memory(GiB)": 46.73,
|
| 1408 |
+
"nll_loss": 0.3944161534309387,
|
| 1409 |
+
"rewards/accuracies": 0.8125,
|
| 1410 |
+
"rewards/chosen": 0.16114036738872528,
|
| 1411 |
+
"rewards/margins": 1.4641916751861572,
|
| 1412 |
+
"rewards/rejected": -1.3030513525009155,
|
| 1413 |
+
"step": 78,
|
| 1414 |
+
"train_speed(iter/s)": 0.005441
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.22581509602501115,
|
| 1418 |
+
"grad_norm": 2.513998031616211,
|
| 1419 |
+
"learning_rate": 0.00019966255789087808,
|
| 1420 |
+
"logits/chosen": -0.8547464609146118,
|
| 1421 |
+
"logits/rejected": -0.8608615398406982,
|
| 1422 |
+
"logps/chosen": -7.297125339508057,
|
| 1423 |
+
"logps/rejected": -27.81803321838379,
|
| 1424 |
+
"loss": 0.6003709435462952,
|
| 1425 |
+
"memory(GiB)": 46.73,
|
| 1426 |
+
"nll_loss": 0.29905110597610474,
|
| 1427 |
+
"rewards/accuracies": 0.9375,
|
| 1428 |
+
"rewards/chosen": 0.13682395219802856,
|
| 1429 |
+
"rewards/margins": 1.487520456314087,
|
| 1430 |
+
"rewards/rejected": -1.3506965637207031,
|
| 1431 |
+
"step": 79,
|
| 1432 |
+
"train_speed(iter/s)": 0.005437
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 0.2286735149620366,
|
| 1436 |
+
"grad_norm": 22.424015045166016,
|
| 1437 |
+
"learning_rate": 0.0001996361176870492,
|
| 1438 |
+
"logits/chosen": -0.8499366044998169,
|
| 1439 |
+
"logits/rejected": -0.863770604133606,
|
| 1440 |
+
"logps/chosen": -8.103764533996582,
|
| 1441 |
+
"logps/rejected": -35.18703842163086,
|
| 1442 |
+
"loss": 0.7495014071464539,
|
| 1443 |
+
"memory(GiB)": 46.73,
|
| 1444 |
+
"nll_loss": 0.5106410980224609,
|
| 1445 |
+
"rewards/accuracies": 0.9375,
|
| 1446 |
+
"rewards/chosen": 0.2272297590970993,
|
| 1447 |
+
"rewards/margins": 1.9457027912139893,
|
| 1448 |
+
"rewards/rejected": -1.7184730768203735,
|
| 1449 |
+
"step": 80,
|
| 1450 |
+
"train_speed(iter/s)": 0.005436
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.23153193389906207,
|
| 1454 |
+
"grad_norm": 4.0974273681640625,
|
| 1455 |
+
"learning_rate": 0.00019960868220749448,
|
| 1456 |
+
"logits/chosen": -0.8633654117584229,
|
| 1457 |
+
"logits/rejected": -0.8727495670318604,
|
| 1458 |
+
"logps/chosen": -5.683876991271973,
|
| 1459 |
+
"logps/rejected": -34.967037200927734,
|
| 1460 |
+
"loss": 0.6187952756881714,
|
| 1461 |
+
"memory(GiB)": 46.73,
|
| 1462 |
+
"nll_loss": 0.32927608489990234,
|
| 1463 |
+
"rewards/accuracies": 0.90625,
|
| 1464 |
+
"rewards/chosen": 0.2094997614622116,
|
| 1465 |
+
"rewards/margins": 2.0794010162353516,
|
| 1466 |
+
"rewards/rejected": -1.8699012994766235,
|
| 1467 |
+
"step": 81,
|
| 1468 |
+
"train_speed(iter/s)": 0.005436
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 0.23439035283608753,
|
| 1472 |
+
"grad_norm": 3.275705575942993,
|
| 1473 |
+
"learning_rate": 0.00019958025172626986,
|
| 1474 |
+
"logits/chosen": -0.8937872648239136,
|
| 1475 |
+
"logits/rejected": -0.9034286737442017,
|
| 1476 |
+
"logps/chosen": -11.346511840820312,
|
| 1477 |
+
"logps/rejected": -35.471927642822266,
|
| 1478 |
+
"loss": 0.7466462850570679,
|
| 1479 |
+
"memory(GiB)": 46.73,
|
| 1480 |
+
"nll_loss": 0.4821043014526367,
|
| 1481 |
+
"rewards/accuracies": 0.9375,
|
| 1482 |
+
"rewards/chosen": 0.2012697458267212,
|
| 1483 |
+
"rewards/margins": 1.9956668615341187,
|
| 1484 |
+
"rewards/rejected": -1.7943971157073975,
|
| 1485 |
+
"step": 82,
|
| 1486 |
+
"train_speed(iter/s)": 0.005436
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.237248771773113,
|
| 1490 |
+
"grad_norm": 8.550602912902832,
|
| 1491 |
+
"learning_rate": 0.0001995508265273704,
|
| 1492 |
+
"logits/chosen": -0.9247998595237732,
|
| 1493 |
+
"logits/rejected": -0.934203028678894,
|
| 1494 |
+
"logps/chosen": -7.0234270095825195,
|
| 1495 |
+
"logps/rejected": -40.43362808227539,
|
| 1496 |
+
"loss": 0.6714224219322205,
|
| 1497 |
+
"memory(GiB)": 46.73,
|
| 1498 |
+
"nll_loss": 0.36930039525032043,
|
| 1499 |
+
"rewards/accuracies": 0.84375,
|
| 1500 |
+
"rewards/chosen": -0.015845157206058502,
|
| 1501 |
+
"rewards/margins": 2.3306310176849365,
|
| 1502 |
+
"rewards/rejected": -2.3464760780334473,
|
| 1503 |
+
"step": 83,
|
| 1504 |
+
"train_speed(iter/s)": 0.005436
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 0.24010719071013845,
|
| 1508 |
+
"grad_norm": 10.380730628967285,
|
| 1509 |
+
"learning_rate": 0.00019952040690472748,
|
| 1510 |
+
"logits/chosen": -0.89071124792099,
|
| 1511 |
+
"logits/rejected": -0.896521806716919,
|
| 1512 |
+
"logps/chosen": -9.274986267089844,
|
| 1513 |
+
"logps/rejected": -31.167112350463867,
|
| 1514 |
+
"loss": 0.9556330442428589,
|
| 1515 |
+
"memory(GiB)": 46.73,
|
| 1516 |
+
"nll_loss": 0.4726044833660126,
|
| 1517 |
+
"rewards/accuracies": 0.78125,
|
| 1518 |
+
"rewards/chosen": 0.055123914033174515,
|
| 1519 |
+
"rewards/margins": 1.99386465549469,
|
| 1520 |
+
"rewards/rejected": -1.9387409687042236,
|
| 1521 |
+
"step": 84,
|
| 1522 |
+
"train_speed(iter/s)": 0.005435
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.2429656096471639,
|
| 1526 |
+
"grad_norm": 59.343536376953125,
|
| 1527 |
+
"learning_rate": 0.00019948899316220604,
|
| 1528 |
+
"logits/chosen": -0.86629319190979,
|
| 1529 |
+
"logits/rejected": -0.8767050504684448,
|
| 1530 |
+
"logps/chosen": -6.04016637802124,
|
| 1531 |
+
"logps/rejected": -28.20957374572754,
|
| 1532 |
+
"loss": 0.725101888179779,
|
| 1533 |
+
"memory(GiB)": 46.73,
|
| 1534 |
+
"nll_loss": 0.40447694063186646,
|
| 1535 |
+
"rewards/accuracies": 0.8125,
|
| 1536 |
+
"rewards/chosen": 0.29112857580184937,
|
| 1537 |
+
"rewards/margins": 1.845672369003296,
|
| 1538 |
+
"rewards/rejected": -1.5545437335968018,
|
| 1539 |
+
"step": 85,
|
| 1540 |
+
"train_speed(iter/s)": 0.005436
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.24582402858418936,
|
| 1544 |
+
"grad_norm": 6.300822734832764,
|
| 1545 |
+
"learning_rate": 0.0001994565856136012,
|
| 1546 |
+
"logits/chosen": -0.8398523330688477,
|
| 1547 |
+
"logits/rejected": -0.8439276814460754,
|
| 1548 |
+
"logps/chosen": -7.2902045249938965,
|
| 1549 |
+
"logps/rejected": -32.82334899902344,
|
| 1550 |
+
"loss": 0.5805257558822632,
|
| 1551 |
+
"memory(GiB)": 46.73,
|
| 1552 |
+
"nll_loss": 0.4190253019332886,
|
| 1553 |
+
"rewards/accuracies": 0.9375,
|
| 1554 |
+
"rewards/chosen": 0.3955605626106262,
|
| 1555 |
+
"rewards/margins": 2.423994302749634,
|
| 1556 |
+
"rewards/rejected": -2.0284337997436523,
|
| 1557 |
+
"step": 86,
|
| 1558 |
+
"train_speed(iter/s)": 0.005436
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.24868244752121482,
|
| 1562 |
+
"grad_norm": 43.768619537353516,
|
| 1563 |
+
"learning_rate": 0.0001994231845826354,
|
| 1564 |
+
"logits/chosen": -0.8710224628448486,
|
| 1565 |
+
"logits/rejected": -0.8721168637275696,
|
| 1566 |
+
"logps/chosen": -7.397428035736084,
|
| 1567 |
+
"logps/rejected": -29.24224853515625,
|
| 1568 |
+
"loss": 1.1686397790908813,
|
| 1569 |
+
"memory(GiB)": 46.73,
|
| 1570 |
+
"nll_loss": 0.7285480499267578,
|
| 1571 |
+
"rewards/accuracies": 0.8125,
|
| 1572 |
+
"rewards/chosen": -0.11831004917621613,
|
| 1573 |
+
"rewards/margins": 1.2224053144454956,
|
| 1574 |
+
"rewards/rejected": -1.3407154083251953,
|
| 1575 |
+
"step": 87,
|
| 1576 |
+
"train_speed(iter/s)": 0.005436
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 0.2515408664582403,
|
| 1580 |
+
"grad_norm": 39.749202728271484,
|
| 1581 |
+
"learning_rate": 0.00019938879040295508,
|
| 1582 |
+
"logits/chosen": -0.8551483750343323,
|
| 1583 |
+
"logits/rejected": -0.8542779684066772,
|
| 1584 |
+
"logps/chosen": -10.534215927124023,
|
| 1585 |
+
"logps/rejected": -26.982810974121094,
|
| 1586 |
+
"loss": 1.2515333890914917,
|
| 1587 |
+
"memory(GiB)": 46.73,
|
| 1588 |
+
"nll_loss": 0.7181237936019897,
|
| 1589 |
+
"rewards/accuracies": 0.71875,
|
| 1590 |
+
"rewards/chosen": -0.03403427451848984,
|
| 1591 |
+
"rewards/margins": 1.2902536392211914,
|
| 1592 |
+
"rewards/rejected": -1.32428777217865,
|
| 1593 |
+
"step": 88,
|
| 1594 |
+
"train_speed(iter/s)": 0.005437
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.25439928539526574,
|
| 1598 |
+
"grad_norm": 10.098331451416016,
|
| 1599 |
+
"learning_rate": 0.00019935340341812737,
|
| 1600 |
+
"logits/chosen": -0.8135808706283569,
|
| 1601 |
+
"logits/rejected": -0.8137494325637817,
|
| 1602 |
+
"logps/chosen": -9.00113582611084,
|
| 1603 |
+
"logps/rejected": -22.338396072387695,
|
| 1604 |
+
"loss": 0.9040060639381409,
|
| 1605 |
+
"memory(GiB)": 46.73,
|
| 1606 |
+
"nll_loss": 0.5397258400917053,
|
| 1607 |
+
"rewards/accuracies": 0.84375,
|
| 1608 |
+
"rewards/chosen": 0.04357346147298813,
|
| 1609 |
+
"rewards/margins": 1.2031430006027222,
|
| 1610 |
+
"rewards/rejected": -1.159569501876831,
|
| 1611 |
+
"step": 89,
|
| 1612 |
+
"train_speed(iter/s)": 0.005435
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.2572577043322912,
|
| 1616 |
+
"grad_norm": 8.489130973815918,
|
| 1617 |
+
"learning_rate": 0.00019931702398163657,
|
| 1618 |
+
"logits/chosen": -0.8590521216392517,
|
| 1619 |
+
"logits/rejected": -0.8651595711708069,
|
| 1620 |
+
"logps/chosen": -8.849437713623047,
|
| 1621 |
+
"logps/rejected": -27.54432487487793,
|
| 1622 |
+
"loss": 0.8707167506217957,
|
| 1623 |
+
"memory(GiB)": 46.73,
|
| 1624 |
+
"nll_loss": 0.48353832960128784,
|
| 1625 |
+
"rewards/accuracies": 0.8125,
|
| 1626 |
+
"rewards/chosen": 0.06561100482940674,
|
| 1627 |
+
"rewards/margins": 1.1353689432144165,
|
| 1628 |
+
"rewards/rejected": -1.0697578191757202,
|
| 1629 |
+
"step": 90,
|
| 1630 |
+
"train_speed(iter/s)": 0.005437
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 0.26011612326931666,
|
| 1634 |
+
"grad_norm": 5.018461227416992,
|
| 1635 |
+
"learning_rate": 0.00019927965245688072,
|
| 1636 |
+
"logits/chosen": -0.8591816425323486,
|
| 1637 |
+
"logits/rejected": -0.8660668730735779,
|
| 1638 |
+
"logps/chosen": -11.013117790222168,
|
| 1639 |
+
"logps/rejected": -27.660642623901367,
|
| 1640 |
+
"loss": 0.9352516531944275,
|
| 1641 |
+
"memory(GiB)": 46.73,
|
| 1642 |
+
"nll_loss": 0.541546106338501,
|
| 1643 |
+
"rewards/accuracies": 0.78125,
|
| 1644 |
+
"rewards/chosen": 0.2031691074371338,
|
| 1645 |
+
"rewards/margins": 1.437905192375183,
|
| 1646 |
+
"rewards/rejected": -1.2347362041473389,
|
| 1647 |
+
"step": 91,
|
| 1648 |
+
"train_speed(iter/s)": 0.005437
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.26297454220634214,
|
| 1652 |
+
"grad_norm": 6.94728946685791,
|
| 1653 |
+
"learning_rate": 0.00019924128921716797,
|
| 1654 |
+
"logits/chosen": -0.8400283455848694,
|
| 1655 |
+
"logits/rejected": -0.8450547456741333,
|
| 1656 |
+
"logps/chosen": -9.49151611328125,
|
| 1657 |
+
"logps/rejected": -28.902881622314453,
|
| 1658 |
+
"loss": 0.806583821773529,
|
| 1659 |
+
"memory(GiB)": 46.73,
|
| 1660 |
+
"nll_loss": 0.3471754193305969,
|
| 1661 |
+
"rewards/accuracies": 0.78125,
|
| 1662 |
+
"rewards/chosen": 0.21784308552742004,
|
| 1663 |
+
"rewards/margins": 1.679865837097168,
|
| 1664 |
+
"rewards/rejected": -1.4620226621627808,
|
| 1665 |
+
"step": 92,
|
| 1666 |
+
"train_speed(iter/s)": 0.005436
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.2658329611433676,
|
| 1670 |
+
"grad_norm": 2.8959178924560547,
|
| 1671 |
+
"learning_rate": 0.00019920193464571275,
|
| 1672 |
+
"logits/chosen": -0.839047372341156,
|
| 1673 |
+
"logits/rejected": -0.8450105786323547,
|
| 1674 |
+
"logps/chosen": -10.59231948852539,
|
| 1675 |
+
"logps/rejected": -29.6372013092041,
|
| 1676 |
+
"loss": 0.7759805917739868,
|
| 1677 |
+
"memory(GiB)": 46.73,
|
| 1678 |
+
"nll_loss": 0.5248351097106934,
|
| 1679 |
+
"rewards/accuracies": 0.90625,
|
| 1680 |
+
"rewards/chosen": 0.3599660098552704,
|
| 1681 |
+
"rewards/margins": 1.950143575668335,
|
| 1682 |
+
"rewards/rejected": -1.5901774168014526,
|
| 1683 |
+
"step": 93,
|
| 1684 |
+
"train_speed(iter/s)": 0.005437
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.268691380080393,
|
| 1688 |
+
"grad_norm": 5.318148612976074,
|
| 1689 |
+
"learning_rate": 0.00019916158913563207,
|
| 1690 |
+
"logits/chosen": -0.8162392377853394,
|
| 1691 |
+
"logits/rejected": -0.8246886730194092,
|
| 1692 |
+
"logps/chosen": -5.6260833740234375,
|
| 1693 |
+
"logps/rejected": -29.538055419921875,
|
| 1694 |
+
"loss": 0.6956499218940735,
|
| 1695 |
+
"memory(GiB)": 46.73,
|
| 1696 |
+
"nll_loss": 0.38102442026138306,
|
| 1697 |
+
"rewards/accuracies": 0.84375,
|
| 1698 |
+
"rewards/chosen": 0.2292197048664093,
|
| 1699 |
+
"rewards/margins": 1.7469617128372192,
|
| 1700 |
+
"rewards/rejected": -1.5177420377731323,
|
| 1701 |
+
"step": 94,
|
| 1702 |
+
"train_speed(iter/s)": 0.005437
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.2715497990174185,
|
| 1706 |
+
"grad_norm": 4.5264787673950195,
|
| 1707 |
+
"learning_rate": 0.00019912025308994148,
|
| 1708 |
+
"logits/chosen": -0.8437933325767517,
|
| 1709 |
+
"logits/rejected": -0.8463299870491028,
|
| 1710 |
+
"logps/chosen": -8.831324577331543,
|
| 1711 |
+
"logps/rejected": -26.610977172851562,
|
| 1712 |
+
"loss": 0.7661141157150269,
|
| 1713 |
+
"memory(GiB)": 46.73,
|
| 1714 |
+
"nll_loss": 0.4225902557373047,
|
| 1715 |
+
"rewards/accuracies": 0.875,
|
| 1716 |
+
"rewards/chosen": 0.12178793549537659,
|
| 1717 |
+
"rewards/margins": 1.467708945274353,
|
| 1718 |
+
"rewards/rejected": -1.3459210395812988,
|
| 1719 |
+
"step": 95,
|
| 1720 |
+
"train_speed(iter/s)": 0.005436
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 0.2744082179544439,
|
| 1724 |
+
"grad_norm": 5.978212356567383,
|
| 1725 |
+
"learning_rate": 0.00019907792692155113,
|
| 1726 |
+
"logits/chosen": -0.7815445065498352,
|
| 1727 |
+
"logits/rejected": -0.78977370262146,
|
| 1728 |
+
"logps/chosen": -6.673672199249268,
|
| 1729 |
+
"logps/rejected": -33.78847122192383,
|
| 1730 |
+
"loss": 0.7695144414901733,
|
| 1731 |
+
"memory(GiB)": 46.73,
|
| 1732 |
+
"nll_loss": 0.5254613757133484,
|
| 1733 |
+
"rewards/accuracies": 0.9375,
|
| 1734 |
+
"rewards/chosen": 0.2026345133781433,
|
| 1735 |
+
"rewards/margins": 1.8294543027877808,
|
| 1736 |
+
"rewards/rejected": -1.6268197298049927,
|
| 1737 |
+
"step": 96,
|
| 1738 |
+
"train_speed(iter/s)": 0.005435
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.2772666368914694,
|
| 1742 |
+
"grad_norm": 9.170755386352539,
|
| 1743 |
+
"learning_rate": 0.00019903461105326154,
|
| 1744 |
+
"logits/chosen": -0.8260455131530762,
|
| 1745 |
+
"logits/rejected": -0.830398678779602,
|
| 1746 |
+
"logps/chosen": -6.689513206481934,
|
| 1747 |
+
"logps/rejected": -22.671979904174805,
|
| 1748 |
+
"loss": 0.7065067291259766,
|
| 1749 |
+
"memory(GiB)": 46.73,
|
| 1750 |
+
"nll_loss": 0.32698267698287964,
|
| 1751 |
+
"rewards/accuracies": 0.8125,
|
| 1752 |
+
"rewards/chosen": 0.2753945589065552,
|
| 1753 |
+
"rewards/margins": 1.3633196353912354,
|
| 1754 |
+
"rewards/rejected": -1.0879249572753906,
|
| 1755 |
+
"step": 97,
|
| 1756 |
+
"train_speed(iter/s)": 0.005435
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.28012505582849484,
|
| 1760 |
+
"grad_norm": 3.521716594696045,
|
| 1761 |
+
"learning_rate": 0.00019899030591775957,
|
| 1762 |
+
"logits/chosen": -0.7963160872459412,
|
| 1763 |
+
"logits/rejected": -0.8014420866966248,
|
| 1764 |
+
"logps/chosen": -7.337916851043701,
|
| 1765 |
+
"logps/rejected": -28.08383560180664,
|
| 1766 |
+
"loss": 0.61917644739151,
|
| 1767 |
+
"memory(GiB)": 46.73,
|
| 1768 |
+
"nll_loss": 0.3110499680042267,
|
| 1769 |
+
"rewards/accuracies": 0.84375,
|
| 1770 |
+
"rewards/chosen": 0.23168900609016418,
|
| 1771 |
+
"rewards/margins": 1.7576744556427002,
|
| 1772 |
+
"rewards/rejected": -1.525985598564148,
|
| 1773 |
+
"step": 98,
|
| 1774 |
+
"train_speed(iter/s)": 0.005436
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.2829834747655203,
|
| 1778 |
+
"grad_norm": 5.793614864349365,
|
| 1779 |
+
"learning_rate": 0.00019894501195761392,
|
| 1780 |
+
"logits/chosen": -0.8137191534042358,
|
| 1781 |
+
"logits/rejected": -0.8227260708808899,
|
| 1782 |
+
"logps/chosen": -6.954023838043213,
|
| 1783 |
+
"logps/rejected": -35.75700378417969,
|
| 1784 |
+
"loss": 0.6259573698043823,
|
| 1785 |
+
"memory(GiB)": 46.73,
|
| 1786 |
+
"nll_loss": 0.33839166164398193,
|
| 1787 |
+
"rewards/accuracies": 0.875,
|
| 1788 |
+
"rewards/chosen": 0.26017990708351135,
|
| 1789 |
+
"rewards/margins": 2.326916217803955,
|
| 1790 |
+
"rewards/rejected": -2.0667362213134766,
|
| 1791 |
+
"step": 99,
|
| 1792 |
+
"train_speed(iter/s)": 0.005437
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 0.28584189370254576,
|
| 1796 |
+
"grad_norm": 5.138144493103027,
|
| 1797 |
+
"learning_rate": 0.00019889872962527072,
|
| 1798 |
+
"logits/chosen": -0.8491114377975464,
|
| 1799 |
+
"logits/rejected": -0.8503983020782471,
|
| 1800 |
+
"logps/chosen": -5.965326309204102,
|
| 1801 |
+
"logps/rejected": -24.436853408813477,
|
| 1802 |
+
"loss": 0.9054074287414551,
|
| 1803 |
+
"memory(GiB)": 46.73,
|
| 1804 |
+
"nll_loss": 0.3898293972015381,
|
| 1805 |
+
"rewards/accuracies": 0.65625,
|
| 1806 |
+
"rewards/chosen": 0.10434619337320328,
|
| 1807 |
+
"rewards/margins": 1.369442105293274,
|
| 1808 |
+
"rewards/rejected": -1.265095829963684,
|
| 1809 |
+
"step": 100,
|
| 1810 |
+
"train_speed(iter/s)": 0.005437
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.28870031263957124,
|
| 1814 |
+
"grad_norm": 3.6206700801849365,
|
| 1815 |
+
"learning_rate": 0.00019885145938304907,
|
| 1816 |
+
"logits/chosen": -0.8245471119880676,
|
| 1817 |
+
"logits/rejected": -0.8329594135284424,
|
| 1818 |
+
"logps/chosen": -12.563156127929688,
|
| 1819 |
+
"logps/rejected": -31.796875,
|
| 1820 |
+
"loss": 0.6779420971870422,
|
| 1821 |
+
"memory(GiB)": 46.73,
|
| 1822 |
+
"nll_loss": 0.36515456438064575,
|
| 1823 |
+
"rewards/accuracies": 0.84375,
|
| 1824 |
+
"rewards/chosen": 0.2728801667690277,
|
| 1825 |
+
"rewards/margins": 2.0701961517333984,
|
| 1826 |
+
"rewards/rejected": -1.797316074371338,
|
| 1827 |
+
"step": 101,
|
| 1828 |
+
"train_speed(iter/s)": 0.005432
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 0.2915587315765967,
|
| 1832 |
+
"grad_norm": 4.129011154174805,
|
| 1833 |
+
"learning_rate": 0.0001988032017031364,
|
| 1834 |
+
"logits/chosen": -0.7823147773742676,
|
| 1835 |
+
"logits/rejected": -0.7900741696357727,
|
| 1836 |
+
"logps/chosen": -5.765602111816406,
|
| 1837 |
+
"logps/rejected": -31.836132049560547,
|
| 1838 |
+
"loss": 0.533344030380249,
|
| 1839 |
+
"memory(GiB)": 46.73,
|
| 1840 |
+
"nll_loss": 0.3198188543319702,
|
| 1841 |
+
"rewards/accuracies": 0.9375,
|
| 1842 |
+
"rewards/chosen": 0.3069157004356384,
|
| 1843 |
+
"rewards/margins": 2.2244224548339844,
|
| 1844 |
+
"rewards/rejected": -1.9175068140029907,
|
| 1845 |
+
"step": 102,
|
| 1846 |
+
"train_speed(iter/s)": 0.005432
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.29441715051362216,
|
| 1850 |
+
"grad_norm": 7.285937786102295,
|
| 1851 |
+
"learning_rate": 0.00019875395706758388,
|
| 1852 |
+
"logits/chosen": -0.8199520707130432,
|
| 1853 |
+
"logits/rejected": -0.8266770839691162,
|
| 1854 |
+
"logps/chosen": -6.686117172241211,
|
| 1855 |
+
"logps/rejected": -35.52680969238281,
|
| 1856 |
+
"loss": 0.6140994429588318,
|
| 1857 |
+
"memory(GiB)": 46.73,
|
| 1858 |
+
"nll_loss": 0.33540287613868713,
|
| 1859 |
+
"rewards/accuracies": 0.90625,
|
| 1860 |
+
"rewards/chosen": 0.3744865357875824,
|
| 1861 |
+
"rewards/margins": 2.1863038539886475,
|
| 1862 |
+
"rewards/rejected": -1.8118171691894531,
|
| 1863 |
+
"step": 103,
|
| 1864 |
+
"train_speed(iter/s)": 0.005432
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.2972755694506476,
|
| 1868 |
+
"grad_norm": 4.08017110824585,
|
| 1869 |
+
"learning_rate": 0.0001987037259683013,
|
| 1870 |
+
"logits/chosen": -0.7858948707580566,
|
| 1871 |
+
"logits/rejected": -0.7929791808128357,
|
| 1872 |
+
"logps/chosen": -6.60341215133667,
|
| 1873 |
+
"logps/rejected": -32.91640853881836,
|
| 1874 |
+
"loss": 0.49151238799095154,
|
| 1875 |
+
"memory(GiB)": 46.73,
|
| 1876 |
+
"nll_loss": 0.26134875416755676,
|
| 1877 |
+
"rewards/accuracies": 0.9375,
|
| 1878 |
+
"rewards/chosen": 0.3017127513885498,
|
| 1879 |
+
"rewards/margins": 2.313183546066284,
|
| 1880 |
+
"rewards/rejected": -2.0114707946777344,
|
| 1881 |
+
"step": 104,
|
| 1882 |
+
"train_speed(iter/s)": 0.005432
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.3001339883876731,
|
| 1886 |
+
"grad_norm": 6.110865116119385,
|
| 1887 |
+
"learning_rate": 0.00019865250890705247,
|
| 1888 |
+
"logits/chosen": -0.8439383506774902,
|
| 1889 |
+
"logits/rejected": -0.8605507612228394,
|
| 1890 |
+
"logps/chosen": -5.249790668487549,
|
| 1891 |
+
"logps/rejected": -45.18266677856445,
|
| 1892 |
+
"loss": 0.665439248085022,
|
| 1893 |
+
"memory(GiB)": 46.73,
|
| 1894 |
+
"nll_loss": 0.4175432622432709,
|
| 1895 |
+
"rewards/accuracies": 0.90625,
|
| 1896 |
+
"rewards/chosen": 0.19819766283035278,
|
| 1897 |
+
"rewards/margins": 2.974789619445801,
|
| 1898 |
+
"rewards/rejected": -2.776592254638672,
|
| 1899 |
+
"step": 105,
|
| 1900 |
+
"train_speed(iter/s)": 0.005432
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.3029924073246985,
|
| 1904 |
+
"grad_norm": 8.704270362854004,
|
| 1905 |
+
"learning_rate": 0.00019860030639545003,
|
| 1906 |
+
"logits/chosen": -0.8581717610359192,
|
| 1907 |
+
"logits/rejected": -0.8701122403144836,
|
| 1908 |
+
"logps/chosen": -8.200727462768555,
|
| 1909 |
+
"logps/rejected": -39.06196975708008,
|
| 1910 |
+
"loss": 0.9722508788108826,
|
| 1911 |
+
"memory(GiB)": 46.73,
|
| 1912 |
+
"nll_loss": 0.6783410310745239,
|
| 1913 |
+
"rewards/accuracies": 0.875,
|
| 1914 |
+
"rewards/chosen": 0.09280453622341156,
|
| 1915 |
+
"rewards/margins": 2.520831823348999,
|
| 1916 |
+
"rewards/rejected": -2.4280271530151367,
|
| 1917 |
+
"step": 106,
|
| 1918 |
+
"train_speed(iter/s)": 0.005432
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.305850826261724,
|
| 1922 |
+
"grad_norm": 6.919908046722412,
|
| 1923 |
+
"learning_rate": 0.00019854711895495036,
|
| 1924 |
+
"logits/chosen": -0.8861989974975586,
|
| 1925 |
+
"logits/rejected": -0.8968295454978943,
|
| 1926 |
+
"logps/chosen": -4.962037563323975,
|
| 1927 |
+
"logps/rejected": -40.77632522583008,
|
| 1928 |
+
"loss": 0.5569091439247131,
|
| 1929 |
+
"memory(GiB)": 46.73,
|
| 1930 |
+
"nll_loss": 0.37660181522369385,
|
| 1931 |
+
"rewards/accuracies": 0.96875,
|
| 1932 |
+
"rewards/chosen": 0.30795052647590637,
|
| 1933 |
+
"rewards/margins": 3.064462184906006,
|
| 1934 |
+
"rewards/rejected": -2.756511688232422,
|
| 1935 |
+
"step": 107,
|
| 1936 |
+
"train_speed(iter/s)": 0.005431
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 0.3087092451987494,
|
| 1940 |
+
"grad_norm": 4.996776580810547,
|
| 1941 |
+
"learning_rate": 0.00019849294711684845,
|
| 1942 |
+
"logits/chosen": -0.9625644683837891,
|
| 1943 |
+
"logits/rejected": -0.972130298614502,
|
| 1944 |
+
"logps/chosen": -8.820281028747559,
|
| 1945 |
+
"logps/rejected": -50.21597671508789,
|
| 1946 |
+
"loss": 0.7489598989486694,
|
| 1947 |
+
"memory(GiB)": 46.73,
|
| 1948 |
+
"nll_loss": 0.5024483799934387,
|
| 1949 |
+
"rewards/accuracies": 0.90625,
|
| 1950 |
+
"rewards/chosen": 0.26294729113578796,
|
| 1951 |
+
"rewards/margins": 3.5292611122131348,
|
| 1952 |
+
"rewards/rejected": -3.2663135528564453,
|
| 1953 |
+
"step": 108,
|
| 1954 |
+
"train_speed(iter/s)": 0.005431
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.3115676641357749,
|
| 1958 |
+
"grad_norm": 6.157798767089844,
|
| 1959 |
+
"learning_rate": 0.00019843779142227256,
|
| 1960 |
+
"logits/chosen": -0.8762018084526062,
|
| 1961 |
+
"logits/rejected": -0.882946789264679,
|
| 1962 |
+
"logps/chosen": -7.924221992492676,
|
| 1963 |
+
"logps/rejected": -43.705806732177734,
|
| 1964 |
+
"loss": 0.6497112512588501,
|
| 1965 |
+
"memory(GiB)": 46.73,
|
| 1966 |
+
"nll_loss": 0.3160628080368042,
|
| 1967 |
+
"rewards/accuracies": 0.75,
|
| 1968 |
+
"rewards/chosen": 0.10452399402856827,
|
| 1969 |
+
"rewards/margins": 3.319636106491089,
|
| 1970 |
+
"rewards/rejected": -3.2151122093200684,
|
| 1971 |
+
"step": 109,
|
| 1972 |
+
"train_speed(iter/s)": 0.005433
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 0.31442608307280034,
|
| 1976 |
+
"grad_norm": 15.488557815551758,
|
| 1977 |
+
"learning_rate": 0.00019838165242217875,
|
| 1978 |
+
"logits/chosen": -0.863703727722168,
|
| 1979 |
+
"logits/rejected": -0.8713690042495728,
|
| 1980 |
+
"logps/chosen": -6.335309028625488,
|
| 1981 |
+
"logps/rejected": -46.15715026855469,
|
| 1982 |
+
"loss": 0.871519923210144,
|
| 1983 |
+
"memory(GiB)": 46.73,
|
| 1984 |
+
"nll_loss": 0.5997005701065063,
|
| 1985 |
+
"rewards/accuracies": 0.875,
|
| 1986 |
+
"rewards/chosen": 0.15678179264068604,
|
| 1987 |
+
"rewards/margins": 3.3330183029174805,
|
| 1988 |
+
"rewards/rejected": -3.176236629486084,
|
| 1989 |
+
"step": 110,
|
| 1990 |
+
"train_speed(iter/s)": 0.005432
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.31728450200982583,
|
| 1994 |
+
"grad_norm": 4.969946384429932,
|
| 1995 |
+
"learning_rate": 0.0001983245306773454,
|
| 1996 |
+
"logits/chosen": -0.8029311895370483,
|
| 1997 |
+
"logits/rejected": -0.8069216012954712,
|
| 1998 |
+
"logps/chosen": -5.91574764251709,
|
| 1999 |
+
"logps/rejected": -26.575307846069336,
|
| 2000 |
+
"loss": 0.6512956619262695,
|
| 2001 |
+
"memory(GiB)": 46.73,
|
| 2002 |
+
"nll_loss": 0.2922568917274475,
|
| 2003 |
+
"rewards/accuracies": 0.90625,
|
| 2004 |
+
"rewards/chosen": 0.20303387939929962,
|
| 2005 |
+
"rewards/margins": 1.892715573310852,
|
| 2006 |
+
"rewards/rejected": -1.6896816492080688,
|
| 2007 |
+
"step": 111,
|
| 2008 |
+
"train_speed(iter/s)": 0.005433
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.32014292094685126,
|
| 2012 |
+
"grad_norm": 6.895030498504639,
|
| 2013 |
+
"learning_rate": 0.0001982664267583677,
|
| 2014 |
+
"logits/chosen": -0.8168566226959229,
|
| 2015 |
+
"logits/rejected": -0.8292732238769531,
|
| 2016 |
+
"logps/chosen": -5.602714538574219,
|
| 2017 |
+
"logps/rejected": -31.50632667541504,
|
| 2018 |
+
"loss": 0.8574928641319275,
|
| 2019 |
+
"memory(GiB)": 46.73,
|
| 2020 |
+
"nll_loss": 0.5075439810752869,
|
| 2021 |
+
"rewards/accuracies": 0.8125,
|
| 2022 |
+
"rewards/chosen": 0.010208956897258759,
|
| 2023 |
+
"rewards/margins": 1.8891350030899048,
|
| 2024 |
+
"rewards/rejected": -1.8789262771606445,
|
| 2025 |
+
"step": 112,
|
| 2026 |
+
"train_speed(iter/s)": 0.005432
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.32300133988387675,
|
| 2030 |
+
"grad_norm": 4.541954040527344,
|
| 2031 |
+
"learning_rate": 0.00019820734124565182,
|
| 2032 |
+
"logits/chosen": -0.7733545899391174,
|
| 2033 |
+
"logits/rejected": -0.7771452069282532,
|
| 2034 |
+
"logps/chosen": -6.299251079559326,
|
| 2035 |
+
"logps/rejected": -27.075977325439453,
|
| 2036 |
+
"loss": 0.7081068754196167,
|
| 2037 |
+
"memory(GiB)": 46.73,
|
| 2038 |
+
"nll_loss": 0.3878493905067444,
|
| 2039 |
+
"rewards/accuracies": 0.84375,
|
| 2040 |
+
"rewards/chosen": 0.13715356588363647,
|
| 2041 |
+
"rewards/margins": 1.6064281463623047,
|
| 2042 |
+
"rewards/rejected": -1.4692747592926025,
|
| 2043 |
+
"step": 113,
|
| 2044 |
+
"train_speed(iter/s)": 0.005431
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.3258597588209022,
|
| 2048 |
+
"grad_norm": 3.1780240535736084,
|
| 2049 |
+
"learning_rate": 0.00019814727472940917,
|
| 2050 |
+
"logits/chosen": -0.7537604570388794,
|
| 2051 |
+
"logits/rejected": -0.7559649348258972,
|
| 2052 |
+
"logps/chosen": -11.733123779296875,
|
| 2053 |
+
"logps/rejected": -27.673603057861328,
|
| 2054 |
+
"loss": 0.71772301197052,
|
| 2055 |
+
"memory(GiB)": 46.73,
|
| 2056 |
+
"nll_loss": 0.42100292444229126,
|
| 2057 |
+
"rewards/accuracies": 0.96875,
|
| 2058 |
+
"rewards/chosen": 0.21125301718711853,
|
| 2059 |
+
"rewards/margins": 1.6915286779403687,
|
| 2060 |
+
"rewards/rejected": -1.4802757501602173,
|
| 2061 |
+
"step": 114,
|
| 2062 |
+
"train_speed(iter/s)": 0.005429
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.32871817775792767,
|
| 2066 |
+
"grad_norm": 3.8751778602600098,
|
| 2067 |
+
"learning_rate": 0.00019808622780965066,
|
| 2068 |
+
"logits/chosen": -0.7520322799682617,
|
| 2069 |
+
"logits/rejected": -0.7543449401855469,
|
| 2070 |
+
"logps/chosen": -7.260441303253174,
|
| 2071 |
+
"logps/rejected": -29.127504348754883,
|
| 2072 |
+
"loss": 0.6945881843566895,
|
| 2073 |
+
"memory(GiB)": 46.73,
|
| 2074 |
+
"nll_loss": 0.3524607717990875,
|
| 2075 |
+
"rewards/accuracies": 0.84375,
|
| 2076 |
+
"rewards/chosen": 0.26789340376853943,
|
| 2077 |
+
"rewards/margins": 1.691644549369812,
|
| 2078 |
+
"rewards/rejected": -1.4237512350082397,
|
| 2079 |
+
"step": 115,
|
| 2080 |
+
"train_speed(iter/s)": 0.005429
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 0.3315765966949531,
|
| 2084 |
+
"grad_norm": 3.0139174461364746,
|
| 2085 |
+
"learning_rate": 0.0001980242010961803,
|
| 2086 |
+
"logits/chosen": -0.7901269793510437,
|
| 2087 |
+
"logits/rejected": -0.7972939014434814,
|
| 2088 |
+
"logps/chosen": -7.289261341094971,
|
| 2089 |
+
"logps/rejected": -47.759063720703125,
|
| 2090 |
+
"loss": 0.5331582427024841,
|
| 2091 |
+
"memory(GiB)": 46.73,
|
| 2092 |
+
"nll_loss": 0.38446128368377686,
|
| 2093 |
+
"rewards/accuracies": 1.0,
|
| 2094 |
+
"rewards/chosen": 0.3029555678367615,
|
| 2095 |
+
"rewards/margins": 2.9576172828674316,
|
| 2096 |
+
"rewards/rejected": -2.6546616554260254,
|
| 2097 |
+
"step": 116,
|
| 2098 |
+
"train_speed(iter/s)": 0.005429
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.3344350156319786,
|
| 2102 |
+
"grad_norm": 3.233109712600708,
|
| 2103 |
+
"learning_rate": 0.00019796119520858955,
|
| 2104 |
+
"logits/chosen": -0.7568274140357971,
|
| 2105 |
+
"logits/rejected": -0.7630717158317566,
|
| 2106 |
+
"logps/chosen": -10.064506530761719,
|
| 2107 |
+
"logps/rejected": -32.50635528564453,
|
| 2108 |
+
"loss": 0.789489209651947,
|
| 2109 |
+
"memory(GiB)": 46.73,
|
| 2110 |
+
"nll_loss": 0.4839721918106079,
|
| 2111 |
+
"rewards/accuracies": 0.84375,
|
| 2112 |
+
"rewards/chosen": 0.3997339606285095,
|
| 2113 |
+
"rewards/margins": 2.3508408069610596,
|
| 2114 |
+
"rewards/rejected": -1.9511069059371948,
|
| 2115 |
+
"step": 117,
|
| 2116 |
+
"train_speed(iter/s)": 0.005429
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.337293434569004,
|
| 2120 |
+
"grad_norm": 6.901320934295654,
|
| 2121 |
+
"learning_rate": 0.0001978972107762509,
|
| 2122 |
+
"logits/chosen": -0.7462923526763916,
|
| 2123 |
+
"logits/rejected": -0.7441357970237732,
|
| 2124 |
+
"logps/chosen": -10.976001739501953,
|
| 2125 |
+
"logps/rejected": -31.10492706298828,
|
| 2126 |
+
"loss": 0.7512857913970947,
|
| 2127 |
+
"memory(GiB)": 46.73,
|
| 2128 |
+
"nll_loss": 0.43077993392944336,
|
| 2129 |
+
"rewards/accuracies": 0.875,
|
| 2130 |
+
"rewards/chosen": 0.1536616086959839,
|
| 2131 |
+
"rewards/margins": 2.106844186782837,
|
| 2132 |
+
"rewards/rejected": -1.953182578086853,
|
| 2133 |
+
"step": 118,
|
| 2134 |
+
"train_speed(iter/s)": 0.00543
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.3401518535060295,
|
| 2138 |
+
"grad_norm": 2.286742925643921,
|
| 2139 |
+
"learning_rate": 0.00019783224843831162,
|
| 2140 |
+
"logits/chosen": -0.7183930277824402,
|
| 2141 |
+
"logits/rejected": -0.7182736396789551,
|
| 2142 |
+
"logps/chosen": -8.731864929199219,
|
| 2143 |
+
"logps/rejected": -33.98211669921875,
|
| 2144 |
+
"loss": 0.6520930528640747,
|
| 2145 |
+
"memory(GiB)": 46.73,
|
| 2146 |
+
"nll_loss": 0.3684110939502716,
|
| 2147 |
+
"rewards/accuracies": 0.84375,
|
| 2148 |
+
"rewards/chosen": 0.2964346706867218,
|
| 2149 |
+
"rewards/margins": 2.4515674114227295,
|
| 2150 |
+
"rewards/rejected": -2.15513277053833,
|
| 2151 |
+
"step": 119,
|
| 2152 |
+
"train_speed(iter/s)": 0.005432
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.34301027244305493,
|
| 2156 |
+
"grad_norm": 4.936792850494385,
|
| 2157 |
+
"learning_rate": 0.00019776630884368738,
|
| 2158 |
+
"logits/chosen": -0.7660694122314453,
|
| 2159 |
+
"logits/rejected": -0.776424765586853,
|
| 2160 |
+
"logps/chosen": -4.871070861816406,
|
| 2161 |
+
"logps/rejected": -47.67111587524414,
|
| 2162 |
+
"loss": 0.4515693187713623,
|
| 2163 |
+
"memory(GiB)": 46.73,
|
| 2164 |
+
"nll_loss": 0.30822524428367615,
|
| 2165 |
+
"rewards/accuracies": 0.9375,
|
| 2166 |
+
"rewards/chosen": 0.38517045974731445,
|
| 2167 |
+
"rewards/margins": 3.7489230632781982,
|
| 2168 |
+
"rewards/rejected": -3.363752841949463,
|
| 2169 |
+
"step": 120,
|
| 2170 |
+
"train_speed(iter/s)": 0.005431
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.3458686913800804,
|
| 2174 |
+
"grad_norm": 9.613821983337402,
|
| 2175 |
+
"learning_rate": 0.00019769939265105573,
|
| 2176 |
+
"logits/chosen": -0.7732415795326233,
|
| 2177 |
+
"logits/rejected": -0.7863773703575134,
|
| 2178 |
+
"logps/chosen": -4.910383701324463,
|
| 2179 |
+
"logps/rejected": -44.046382904052734,
|
| 2180 |
+
"loss": 0.6085942983627319,
|
| 2181 |
+
"memory(GiB)": 46.73,
|
| 2182 |
+
"nll_loss": 0.3416149616241455,
|
| 2183 |
+
"rewards/accuracies": 0.84375,
|
| 2184 |
+
"rewards/chosen": 0.3152579069137573,
|
| 2185 |
+
"rewards/margins": 3.380157470703125,
|
| 2186 |
+
"rewards/rejected": -3.0648996829986572,
|
| 2187 |
+
"step": 121,
|
| 2188 |
+
"train_speed(iter/s)": 0.005432
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 0.34872711031710585,
|
| 2192 |
+
"grad_norm": 5.3062744140625,
|
| 2193 |
+
"learning_rate": 0.00019763150052884966,
|
| 2194 |
+
"logits/chosen": -0.7940014004707336,
|
| 2195 |
+
"logits/rejected": -0.8029736280441284,
|
| 2196 |
+
"logps/chosen": -3.551081895828247,
|
| 2197 |
+
"logps/rejected": -38.06512451171875,
|
| 2198 |
+
"loss": 0.42784249782562256,
|
| 2199 |
+
"memory(GiB)": 46.73,
|
| 2200 |
+
"nll_loss": 0.25387704372406006,
|
| 2201 |
+
"rewards/accuracies": 0.96875,
|
| 2202 |
+
"rewards/chosen": 0.37009117007255554,
|
| 2203 |
+
"rewards/margins": 2.841196298599243,
|
| 2204 |
+
"rewards/rejected": -2.471104860305786,
|
| 2205 |
+
"step": 122,
|
| 2206 |
+
"train_speed(iter/s)": 0.005431
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.35158552925413133,
|
| 2210 |
+
"grad_norm": 1.533804178237915,
|
| 2211 |
+
"learning_rate": 0.0001975626331552507,
|
| 2212 |
+
"logits/chosen": -0.8097442984580994,
|
| 2213 |
+
"logits/rejected": -0.8176737427711487,
|
| 2214 |
+
"logps/chosen": -3.7899668216705322,
|
| 2215 |
+
"logps/rejected": -51.7594108581543,
|
| 2216 |
+
"loss": 0.3041819632053375,
|
| 2217 |
+
"memory(GiB)": 46.73,
|
| 2218 |
+
"nll_loss": 0.2387474924325943,
|
| 2219 |
+
"rewards/accuracies": 1.0,
|
| 2220 |
+
"rewards/chosen": 0.35475367307662964,
|
| 2221 |
+
"rewards/margins": 4.306023597717285,
|
| 2222 |
+
"rewards/rejected": -3.9512698650360107,
|
| 2223 |
+
"step": 123,
|
| 2224 |
+
"train_speed(iter/s)": 0.00543
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.35444394819115677,
|
| 2228 |
+
"grad_norm": 4.373802661895752,
|
| 2229 |
+
"learning_rate": 0.00019749279121818235,
|
| 2230 |
+
"logits/chosen": -0.7605909705162048,
|
| 2231 |
+
"logits/rejected": -0.770943284034729,
|
| 2232 |
+
"logps/chosen": -8.83084774017334,
|
| 2233 |
+
"logps/rejected": -59.24419021606445,
|
| 2234 |
+
"loss": 0.5374789834022522,
|
| 2235 |
+
"memory(GiB)": 46.73,
|
| 2236 |
+
"nll_loss": 0.39307644963264465,
|
| 2237 |
+
"rewards/accuracies": 0.96875,
|
| 2238 |
+
"rewards/chosen": 0.14716514945030212,
|
| 2239 |
+
"rewards/margins": 4.3210930824279785,
|
| 2240 |
+
"rewards/rejected": -4.173928260803223,
|
| 2241 |
+
"step": 124,
|
| 2242 |
+
"train_speed(iter/s)": 0.00543
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.3573023671281822,
|
| 2246 |
+
"grad_norm": 6.3563103675842285,
|
| 2247 |
+
"learning_rate": 0.0001974219754153032,
|
| 2248 |
+
"logits/chosen": -0.8258897066116333,
|
| 2249 |
+
"logits/rejected": -0.8272631168365479,
|
| 2250 |
+
"logps/chosen": -7.77381706237793,
|
| 2251 |
+
"logps/rejected": -41.343414306640625,
|
| 2252 |
+
"loss": 0.6204044818878174,
|
| 2253 |
+
"memory(GiB)": 46.73,
|
| 2254 |
+
"nll_loss": 0.33825787901878357,
|
| 2255 |
+
"rewards/accuracies": 0.84375,
|
| 2256 |
+
"rewards/chosen": 0.12217142432928085,
|
| 2257 |
+
"rewards/margins": 3.0349438190460205,
|
| 2258 |
+
"rewards/rejected": -2.9127724170684814,
|
| 2259 |
+
"step": 125,
|
| 2260 |
+
"train_speed(iter/s)": 0.005431
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.3601607860652077,
|
| 2264 |
+
"grad_norm": 9.85932731628418,
|
| 2265 |
+
"learning_rate": 0.00019735018645399967,
|
| 2266 |
+
"logits/chosen": -0.8187968730926514,
|
| 2267 |
+
"logits/rejected": -0.8283690214157104,
|
| 2268 |
+
"logps/chosen": -4.5190534591674805,
|
| 2269 |
+
"logps/rejected": -45.119789123535156,
|
| 2270 |
+
"loss": 0.5073397755622864,
|
| 2271 |
+
"memory(GiB)": 46.73,
|
| 2272 |
+
"nll_loss": 0.33468902111053467,
|
| 2273 |
+
"rewards/accuracies": 0.875,
|
| 2274 |
+
"rewards/chosen": 0.34614282846450806,
|
| 2275 |
+
"rewards/margins": 3.5951905250549316,
|
| 2276 |
+
"rewards/rejected": -3.2490477561950684,
|
| 2277 |
+
"step": 126,
|
| 2278 |
+
"train_speed(iter/s)": 0.005432
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.3630192050022331,
|
| 2282 |
+
"grad_norm": 3.943042039871216,
|
| 2283 |
+
"learning_rate": 0.00019727742505137936,
|
| 2284 |
+
"logits/chosen": -0.7813420295715332,
|
| 2285 |
+
"logits/rejected": -0.7975005507469177,
|
| 2286 |
+
"logps/chosen": -7.530571937561035,
|
| 2287 |
+
"logps/rejected": -53.51607894897461,
|
| 2288 |
+
"loss": 0.46567243337631226,
|
| 2289 |
+
"memory(GiB)": 46.73,
|
| 2290 |
+
"nll_loss": 0.2742045819759369,
|
| 2291 |
+
"rewards/accuracies": 0.90625,
|
| 2292 |
+
"rewards/chosen": 0.23301076889038086,
|
| 2293 |
+
"rewards/margins": 3.9482827186584473,
|
| 2294 |
+
"rewards/rejected": -3.7152719497680664,
|
| 2295 |
+
"step": 127,
|
| 2296 |
+
"train_speed(iter/s)": 0.005432
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 0.3658776239392586,
|
| 2300 |
+
"grad_norm": 1.9621978998184204,
|
| 2301 |
+
"learning_rate": 0.00019720369193426365,
|
| 2302 |
+
"logits/chosen": -0.7805514335632324,
|
| 2303 |
+
"logits/rejected": -0.7947923541069031,
|
| 2304 |
+
"logps/chosen": -4.593271255493164,
|
| 2305 |
+
"logps/rejected": -51.02714157104492,
|
| 2306 |
+
"loss": 0.3543168008327484,
|
| 2307 |
+
"memory(GiB)": 46.73,
|
| 2308 |
+
"nll_loss": 0.2695479393005371,
|
| 2309 |
+
"rewards/accuracies": 0.96875,
|
| 2310 |
+
"rewards/chosen": 0.3205646574497223,
|
| 2311 |
+
"rewards/margins": 3.938343048095703,
|
| 2312 |
+
"rewards/rejected": -3.6177780628204346,
|
| 2313 |
+
"step": 128,
|
| 2314 |
+
"train_speed(iter/s)": 0.005433
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.36873604287628403,
|
| 2318 |
+
"grad_norm": 6.375858783721924,
|
| 2319 |
+
"learning_rate": 0.00019712898783918038,
|
| 2320 |
+
"logits/chosen": -0.7720431089401245,
|
| 2321 |
+
"logits/rejected": -0.7774226069450378,
|
| 2322 |
+
"logps/chosen": -3.9679479598999023,
|
| 2323 |
+
"logps/rejected": -39.059242248535156,
|
| 2324 |
+
"loss": 0.42859622836112976,
|
| 2325 |
+
"memory(GiB)": 46.73,
|
| 2326 |
+
"nll_loss": 0.2718777060508728,
|
| 2327 |
+
"rewards/accuracies": 0.96875,
|
| 2328 |
+
"rewards/chosen": 0.335873007774353,
|
| 2329 |
+
"rewards/margins": 3.0524649620056152,
|
| 2330 |
+
"rewards/rejected": -2.7165920734405518,
|
| 2331 |
+
"step": 129,
|
| 2332 |
+
"train_speed(iter/s)": 0.005434
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.3715944618133095,
|
| 2336 |
+
"grad_norm": 20.90467643737793,
|
| 2337 |
+
"learning_rate": 0.00019705331351235674,
|
| 2338 |
+
"logits/chosen": -0.7668591737747192,
|
| 2339 |
+
"logits/rejected": -0.7685776948928833,
|
| 2340 |
+
"logps/chosen": -9.782769203186035,
|
| 2341 |
+
"logps/rejected": -33.2745361328125,
|
| 2342 |
+
"loss": 1.310682773590088,
|
| 2343 |
+
"memory(GiB)": 46.73,
|
| 2344 |
+
"nll_loss": 0.8834658861160278,
|
| 2345 |
+
"rewards/accuracies": 0.78125,
|
| 2346 |
+
"rewards/chosen": -0.039194077253341675,
|
| 2347 |
+
"rewards/margins": 2.1344308853149414,
|
| 2348 |
+
"rewards/rejected": -2.1736247539520264,
|
| 2349 |
+
"step": 130,
|
| 2350 |
+
"train_speed(iter/s)": 0.005434
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 0.37445288075033495,
|
| 2354 |
+
"grad_norm": 2.3716464042663574,
|
| 2355 |
+
"learning_rate": 0.00019697666970971153,
|
| 2356 |
+
"logits/chosen": -0.7852880954742432,
|
| 2357 |
+
"logits/rejected": -0.7899134755134583,
|
| 2358 |
+
"logps/chosen": -6.234109878540039,
|
| 2359 |
+
"logps/rejected": -42.3105354309082,
|
| 2360 |
+
"loss": 0.4260920286178589,
|
| 2361 |
+
"memory(GiB)": 46.73,
|
| 2362 |
+
"nll_loss": 0.25257518887519836,
|
| 2363 |
+
"rewards/accuracies": 0.9375,
|
| 2364 |
+
"rewards/chosen": 0.3239039182662964,
|
| 2365 |
+
"rewards/margins": 3.0808892250061035,
|
| 2366 |
+
"rewards/rejected": -2.7569851875305176,
|
| 2367 |
+
"step": 131,
|
| 2368 |
+
"train_speed(iter/s)": 0.005433
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.37731129968736044,
|
| 2372 |
+
"grad_norm": 17.186786651611328,
|
| 2373 |
+
"learning_rate": 0.00019689905719684782,
|
| 2374 |
+
"logits/chosen": -0.7246553301811218,
|
| 2375 |
+
"logits/rejected": -0.739700198173523,
|
| 2376 |
+
"logps/chosen": -4.994725227355957,
|
| 2377 |
+
"logps/rejected": -37.61749267578125,
|
| 2378 |
+
"loss": 0.7901899814605713,
|
| 2379 |
+
"memory(GiB)": 46.73,
|
| 2380 |
+
"nll_loss": 0.529437243938446,
|
| 2381 |
+
"rewards/accuracies": 0.96875,
|
| 2382 |
+
"rewards/chosen": 0.3914901614189148,
|
| 2383 |
+
"rewards/margins": 2.7029402256011963,
|
| 2384 |
+
"rewards/rejected": -2.311450242996216,
|
| 2385 |
+
"step": 132,
|
| 2386 |
+
"train_speed(iter/s)": 0.005432
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.38016971862438587,
|
| 2390 |
+
"grad_norm": 19.44753646850586,
|
| 2391 |
+
"learning_rate": 0.00019682047674904525,
|
| 2392 |
+
"logits/chosen": -0.7204717993736267,
|
| 2393 |
+
"logits/rejected": -0.7237159013748169,
|
| 2394 |
+
"logps/chosen": -10.331472396850586,
|
| 2395 |
+
"logps/rejected": -44.90771484375,
|
| 2396 |
+
"loss": 0.9498412609100342,
|
| 2397 |
+
"memory(GiB)": 46.73,
|
| 2398 |
+
"nll_loss": 0.6981635093688965,
|
| 2399 |
+
"rewards/accuracies": 0.90625,
|
| 2400 |
+
"rewards/chosen": 0.13750669360160828,
|
| 2401 |
+
"rewards/margins": 3.0414295196533203,
|
| 2402 |
+
"rewards/rejected": -2.9039225578308105,
|
| 2403 |
+
"step": 133,
|
| 2404 |
+
"train_speed(iter/s)": 0.005433
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.38302813756141135,
|
| 2408 |
+
"grad_norm": 2.630051612854004,
|
| 2409 |
+
"learning_rate": 0.00019674092915125218,
|
| 2410 |
+
"logits/chosen": -0.7359256148338318,
|
| 2411 |
+
"logits/rejected": -0.7523387670516968,
|
| 2412 |
+
"logps/chosen": -5.2792582511901855,
|
| 2413 |
+
"logps/rejected": -53.244354248046875,
|
| 2414 |
+
"loss": 0.37174469232559204,
|
| 2415 |
+
"memory(GiB)": 46.73,
|
| 2416 |
+
"nll_loss": 0.19531044363975525,
|
| 2417 |
+
"rewards/accuracies": 0.9375,
|
| 2418 |
+
"rewards/chosen": 0.35951948165893555,
|
| 2419 |
+
"rewards/margins": 3.9146037101745605,
|
| 2420 |
+
"rewards/rejected": -3.555084228515625,
|
| 2421 |
+
"step": 134,
|
| 2422 |
+
"train_speed(iter/s)": 0.005432
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.3858865564984368,
|
| 2426 |
+
"grad_norm": 8.38878059387207,
|
| 2427 |
+
"learning_rate": 0.00019666041519807802,
|
| 2428 |
+
"logits/chosen": -0.7512345910072327,
|
| 2429 |
+
"logits/rejected": -0.7590532302856445,
|
| 2430 |
+
"logps/chosen": -7.108774662017822,
|
| 2431 |
+
"logps/rejected": -49.24845886230469,
|
| 2432 |
+
"loss": 0.515214741230011,
|
| 2433 |
+
"memory(GiB)": 46.73,
|
| 2434 |
+
"nll_loss": 0.35548365116119385,
|
| 2435 |
+
"rewards/accuracies": 0.9375,
|
| 2436 |
+
"rewards/chosen": 0.11580058932304382,
|
| 2437 |
+
"rewards/margins": 3.6866679191589355,
|
| 2438 |
+
"rewards/rejected": -3.5708670616149902,
|
| 2439 |
+
"step": 135,
|
| 2440 |
+
"train_speed(iter/s)": 0.005431
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 0.38874497543546227,
|
| 2444 |
+
"grad_norm": 7.971051216125488,
|
| 2445 |
+
"learning_rate": 0.0001965789356937852,
|
| 2446 |
+
"logits/chosen": -0.7898239493370056,
|
| 2447 |
+
"logits/rejected": -0.7973160743713379,
|
| 2448 |
+
"logps/chosen": -9.885854721069336,
|
| 2449 |
+
"logps/rejected": -55.670047760009766,
|
| 2450 |
+
"loss": 0.7229015827178955,
|
| 2451 |
+
"memory(GiB)": 46.73,
|
| 2452 |
+
"nll_loss": 0.43035101890563965,
|
| 2453 |
+
"rewards/accuracies": 0.875,
|
| 2454 |
+
"rewards/chosen": 0.13693107664585114,
|
| 2455 |
+
"rewards/margins": 3.9661924839019775,
|
| 2456 |
+
"rewards/rejected": -3.829261541366577,
|
| 2457 |
+
"step": 136,
|
| 2458 |
+
"train_speed(iter/s)": 0.00543
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.3916033943724877,
|
| 2462 |
+
"grad_norm": 6.154712677001953,
|
| 2463 |
+
"learning_rate": 0.00019649649145228102,
|
| 2464 |
+
"logits/chosen": -0.7895013689994812,
|
| 2465 |
+
"logits/rejected": -0.7943228483200073,
|
| 2466 |
+
"logps/chosen": -5.192661285400391,
|
| 2467 |
+
"logps/rejected": -39.91571044921875,
|
| 2468 |
+
"loss": 0.5592948794364929,
|
| 2469 |
+
"memory(GiB)": 46.73,
|
| 2470 |
+
"nll_loss": 0.22298313677310944,
|
| 2471 |
+
"rewards/accuracies": 0.8125,
|
| 2472 |
+
"rewards/chosen": 0.19678261876106262,
|
| 2473 |
+
"rewards/margins": 2.95813250541687,
|
| 2474 |
+
"rewards/rejected": -2.761349678039551,
|
| 2475 |
+
"step": 137,
|
| 2476 |
+
"train_speed(iter/s)": 0.005431
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.3944618133095132,
|
| 2480 |
+
"grad_norm": 2.1893486976623535,
|
| 2481 |
+
"learning_rate": 0.0001964130832971098,
|
| 2482 |
+
"logits/chosen": -0.7978731393814087,
|
| 2483 |
+
"logits/rejected": -0.808353066444397,
|
| 2484 |
+
"logps/chosen": -7.741363048553467,
|
| 2485 |
+
"logps/rejected": -49.584014892578125,
|
| 2486 |
+
"loss": 0.4424561560153961,
|
| 2487 |
+
"memory(GiB)": 46.73,
|
| 2488 |
+
"nll_loss": 0.3187883198261261,
|
| 2489 |
+
"rewards/accuracies": 0.96875,
|
| 2490 |
+
"rewards/chosen": 0.3768758773803711,
|
| 2491 |
+
"rewards/margins": 3.4152517318725586,
|
| 2492 |
+
"rewards/rejected": -3.0383756160736084,
|
| 2493 |
+
"step": 138,
|
| 2494 |
+
"train_speed(iter/s)": 0.005431
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.3973202322465386,
|
| 2498 |
+
"grad_norm": 3.585315465927124,
|
| 2499 |
+
"learning_rate": 0.0001963287120614444,
|
| 2500 |
+
"logits/chosen": -0.8294897079467773,
|
| 2501 |
+
"logits/rejected": -0.8371224403381348,
|
| 2502 |
+
"logps/chosen": -4.645942211151123,
|
| 2503 |
+
"logps/rejected": -41.52901840209961,
|
| 2504 |
+
"loss": 0.5509465336799622,
|
| 2505 |
+
"memory(GiB)": 46.73,
|
| 2506 |
+
"nll_loss": 0.3694899380207062,
|
| 2507 |
+
"rewards/accuracies": 0.9375,
|
| 2508 |
+
"rewards/chosen": 0.15256014466285706,
|
| 2509 |
+
"rewards/margins": 3.042027473449707,
|
| 2510 |
+
"rewards/rejected": -2.8894670009613037,
|
| 2511 |
+
"step": 139,
|
| 2512 |
+
"train_speed(iter/s)": 0.005432
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.4001786511835641,
|
| 2516 |
+
"grad_norm": 3.1583364009857178,
|
| 2517 |
+
"learning_rate": 0.00019624337858807807,
|
| 2518 |
+
"logits/chosen": -0.8058665990829468,
|
| 2519 |
+
"logits/rejected": -0.802970290184021,
|
| 2520 |
+
"logps/chosen": -12.615983009338379,
|
| 2521 |
+
"logps/rejected": -42.576560974121094,
|
| 2522 |
+
"loss": 0.47237128019332886,
|
| 2523 |
+
"memory(GiB)": 46.73,
|
| 2524 |
+
"nll_loss": 0.28210973739624023,
|
| 2525 |
+
"rewards/accuracies": 0.875,
|
| 2526 |
+
"rewards/chosen": -0.08335088193416595,
|
| 2527 |
+
"rewards/margins": 2.9341583251953125,
|
| 2528 |
+
"rewards/rejected": -3.0175094604492188,
|
| 2529 |
+
"step": 140,
|
| 2530 |
+
"train_speed(iter/s)": 0.005432
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.40303707012058954,
|
| 2534 |
+
"grad_norm": 1.9583823680877686,
|
| 2535 |
+
"learning_rate": 0.00019615708372941587,
|
| 2536 |
+
"logits/chosen": -0.7773740887641907,
|
| 2537 |
+
"logits/rejected": -0.7813900113105774,
|
| 2538 |
+
"logps/chosen": -7.43025016784668,
|
| 2539 |
+
"logps/rejected": -48.81037521362305,
|
| 2540 |
+
"loss": 0.45620426535606384,
|
| 2541 |
+
"memory(GiB)": 46.73,
|
| 2542 |
+
"nll_loss": 0.2858808636665344,
|
| 2543 |
+
"rewards/accuracies": 0.90625,
|
| 2544 |
+
"rewards/chosen": 0.14979784190654755,
|
| 2545 |
+
"rewards/margins": 3.327134847640991,
|
| 2546 |
+
"rewards/rejected": -3.177337169647217,
|
| 2547 |
+
"step": 141,
|
| 2548 |
+
"train_speed(iter/s)": 0.005434
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.405895489057615,
|
| 2552 |
+
"grad_norm": 17.376474380493164,
|
| 2553 |
+
"learning_rate": 0.00019606982834746627,
|
| 2554 |
+
"logits/chosen": -0.8055461645126343,
|
| 2555 |
+
"logits/rejected": -0.8150217533111572,
|
| 2556 |
+
"logps/chosen": -9.388490676879883,
|
| 2557 |
+
"logps/rejected": -43.86671447753906,
|
| 2558 |
+
"loss": 1.0983664989471436,
|
| 2559 |
+
"memory(GiB)": 46.73,
|
| 2560 |
+
"nll_loss": 0.7692541480064392,
|
| 2561 |
+
"rewards/accuracies": 0.90625,
|
| 2562 |
+
"rewards/chosen": 0.1326138973236084,
|
| 2563 |
+
"rewards/margins": 2.8655741214752197,
|
| 2564 |
+
"rewards/rejected": -2.7329602241516113,
|
| 2565 |
+
"step": 142,
|
| 2566 |
+
"train_speed(iter/s)": 0.005433
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 0.40875390799464045,
|
| 2570 |
+
"grad_norm": 35.17894744873047,
|
| 2571 |
+
"learning_rate": 0.00019598161331383257,
|
| 2572 |
+
"logits/chosen": -0.8625121116638184,
|
| 2573 |
+
"logits/rejected": -0.871791422367096,
|
| 2574 |
+
"logps/chosen": -9.16450023651123,
|
| 2575 |
+
"logps/rejected": -47.45623016357422,
|
| 2576 |
+
"loss": 0.7930290102958679,
|
| 2577 |
+
"memory(GiB)": 46.73,
|
| 2578 |
+
"nll_loss": 0.543784499168396,
|
| 2579 |
+
"rewards/accuracies": 0.875,
|
| 2580 |
+
"rewards/chosen": 0.08059316873550415,
|
| 2581 |
+
"rewards/margins": 3.273069381713867,
|
| 2582 |
+
"rewards/rejected": -3.192476272583008,
|
| 2583 |
+
"step": 143,
|
| 2584 |
+
"train_speed(iter/s)": 0.005433
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.41161232693166594,
|
| 2588 |
+
"grad_norm": 4.112196445465088,
|
| 2589 |
+
"learning_rate": 0.00019589243950970402,
|
| 2590 |
+
"logits/chosen": -0.8877297043800354,
|
| 2591 |
+
"logits/rejected": -0.8911035060882568,
|
| 2592 |
+
"logps/chosen": -6.215106964111328,
|
| 2593 |
+
"logps/rejected": -41.89563751220703,
|
| 2594 |
+
"loss": 0.3530213236808777,
|
| 2595 |
+
"memory(GiB)": 46.73,
|
| 2596 |
+
"nll_loss": 0.20008514821529388,
|
| 2597 |
+
"rewards/accuracies": 0.9375,
|
| 2598 |
+
"rewards/chosen": 0.3117275834083557,
|
| 2599 |
+
"rewards/margins": 3.073056221008301,
|
| 2600 |
+
"rewards/rejected": -2.76132869720459,
|
| 2601 |
+
"step": 144,
|
| 2602 |
+
"train_speed(iter/s)": 0.005434
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.41447074586869137,
|
| 2606 |
+
"grad_norm": 12.758796691894531,
|
| 2607 |
+
"learning_rate": 0.00019580230782584722,
|
| 2608 |
+
"logits/chosen": -0.8928901553153992,
|
| 2609 |
+
"logits/rejected": -0.8998609781265259,
|
| 2610 |
+
"logps/chosen": -5.406792640686035,
|
| 2611 |
+
"logps/rejected": -40.006412506103516,
|
| 2612 |
+
"loss": 0.5003563165664673,
|
| 2613 |
+
"memory(GiB)": 46.73,
|
| 2614 |
+
"nll_loss": 0.30671095848083496,
|
| 2615 |
+
"rewards/accuracies": 0.90625,
|
| 2616 |
+
"rewards/chosen": 0.23732616007328033,
|
| 2617 |
+
"rewards/margins": 2.932386636734009,
|
| 2618 |
+
"rewards/rejected": -2.6950607299804688,
|
| 2619 |
+
"step": 145,
|
| 2620 |
+
"train_speed(iter/s)": 0.005434
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.41732916480571686,
|
| 2624 |
+
"grad_norm": 4.014595985412598,
|
| 2625 |
+
"learning_rate": 0.00019571121916259706,
|
| 2626 |
+
"logits/chosen": -0.887598991394043,
|
| 2627 |
+
"logits/rejected": -0.8942646384239197,
|
| 2628 |
+
"logps/chosen": -4.3415961265563965,
|
| 2629 |
+
"logps/rejected": -46.59621810913086,
|
| 2630 |
+
"loss": 0.36891594529151917,
|
| 2631 |
+
"memory(GiB)": 46.73,
|
| 2632 |
+
"nll_loss": 0.2651711106300354,
|
| 2633 |
+
"rewards/accuracies": 0.96875,
|
| 2634 |
+
"rewards/chosen": 0.36169373989105225,
|
| 2635 |
+
"rewards/margins": 3.6561460494995117,
|
| 2636 |
+
"rewards/rejected": -3.294452428817749,
|
| 2637 |
+
"step": 146,
|
| 2638 |
+
"train_speed(iter/s)": 0.005434
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.4201875837427423,
|
| 2642 |
+
"grad_norm": 10.37033462524414,
|
| 2643 |
+
"learning_rate": 0.00019561917442984788,
|
| 2644 |
+
"logits/chosen": -0.8872988224029541,
|
| 2645 |
+
"logits/rejected": -0.8909754753112793,
|
| 2646 |
+
"logps/chosen": -7.036233901977539,
|
| 2647 |
+
"logps/rejected": -44.856361389160156,
|
| 2648 |
+
"loss": 0.5071041584014893,
|
| 2649 |
+
"memory(GiB)": 46.73,
|
| 2650 |
+
"nll_loss": 0.4129847586154938,
|
| 2651 |
+
"rewards/accuracies": 0.96875,
|
| 2652 |
+
"rewards/chosen": 0.29683101177215576,
|
| 2653 |
+
"rewards/margins": 3.6587531566619873,
|
| 2654 |
+
"rewards/rejected": -3.361922025680542,
|
| 2655 |
+
"step": 147,
|
| 2656 |
+
"train_speed(iter/s)": 0.005434
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.4230460026797678,
|
| 2660 |
+
"grad_norm": 12.54053783416748,
|
| 2661 |
+
"learning_rate": 0.00019552617454704428,
|
| 2662 |
+
"logits/chosen": -0.909309983253479,
|
| 2663 |
+
"logits/rejected": -0.9173017144203186,
|
| 2664 |
+
"logps/chosen": -5.930859088897705,
|
| 2665 |
+
"logps/rejected": -46.784332275390625,
|
| 2666 |
+
"loss": 0.6537680625915527,
|
| 2667 |
+
"memory(GiB)": 46.73,
|
| 2668 |
+
"nll_loss": 0.521246075630188,
|
| 2669 |
+
"rewards/accuracies": 0.9375,
|
| 2670 |
+
"rewards/chosen": 0.25368478894233704,
|
| 2671 |
+
"rewards/margins": 3.396470308303833,
|
| 2672 |
+
"rewards/rejected": -3.1427857875823975,
|
| 2673 |
+
"step": 148,
|
| 2674 |
+
"train_speed(iter/s)": 0.005433
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.4259044216167932,
|
| 2678 |
+
"grad_norm": 2.5870954990386963,
|
| 2679 |
+
"learning_rate": 0.00019543222044317188,
|
| 2680 |
+
"logits/chosen": -0.8764325976371765,
|
| 2681 |
+
"logits/rejected": -0.881737232208252,
|
| 2682 |
+
"logps/chosen": -5.638083457946777,
|
| 2683 |
+
"logps/rejected": -47.27805709838867,
|
| 2684 |
+
"loss": 0.3462856113910675,
|
| 2685 |
+
"memory(GiB)": 46.73,
|
| 2686 |
+
"nll_loss": 0.22526228427886963,
|
| 2687 |
+
"rewards/accuracies": 0.96875,
|
| 2688 |
+
"rewards/chosen": 0.3174706697463989,
|
| 2689 |
+
"rewards/margins": 3.5086669921875,
|
| 2690 |
+
"rewards/rejected": -3.1911962032318115,
|
| 2691 |
+
"step": 149,
|
| 2692 |
+
"train_speed(iter/s)": 0.005433
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.4287628405538187,
|
| 2696 |
+
"grad_norm": 9.161517143249512,
|
| 2697 |
+
"learning_rate": 0.00019533731305674818,
|
| 2698 |
+
"logits/chosen": -0.8732975721359253,
|
| 2699 |
+
"logits/rejected": -0.8769663572311401,
|
| 2700 |
+
"logps/chosen": -9.518259048461914,
|
| 2701 |
+
"logps/rejected": -42.666114807128906,
|
| 2702 |
+
"loss": 0.9442892670631409,
|
| 2703 |
+
"memory(GiB)": 46.73,
|
| 2704 |
+
"nll_loss": 0.6460040211677551,
|
| 2705 |
+
"rewards/accuracies": 0.875,
|
| 2706 |
+
"rewards/chosen": 0.07534892857074738,
|
| 2707 |
+
"rewards/margins": 3.1069083213806152,
|
| 2708 |
+
"rewards/rejected": -3.0315592288970947,
|
| 2709 |
+
"step": 150,
|
| 2710 |
+
"train_speed(iter/s)": 0.005433
|
| 2711 |
+
}
|
| 2712 |
+
],
|
| 2713 |
+
"logging_steps": 1,
|
| 2714 |
+
"max_steps": 1047,
|
| 2715 |
+
"num_input_tokens_seen": 0,
|
| 2716 |
+
"num_train_epochs": 3,
|
| 2717 |
+
"save_steps": 50,
|
| 2718 |
+
"stateful_callbacks": {
|
| 2719 |
+
"TrainerControl": {
|
| 2720 |
+
"args": {
|
| 2721 |
+
"should_epoch_stop": false,
|
| 2722 |
+
"should_evaluate": false,
|
| 2723 |
+
"should_log": false,
|
| 2724 |
+
"should_save": true,
|
| 2725 |
+
"should_training_stop": false
|
| 2726 |
+
},
|
| 2727 |
+
"attributes": {}
|
| 2728 |
+
}
|
| 2729 |
+
},
|
| 2730 |
+
"total_flos": 1.0479874209029816e+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 sha256:98fc0ce607ab742b8d6bb90f5dd0a507efd428041b2a1526b6e6ebc0f5e6426e
|
| 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)
|