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).*\\.(v_proj|q_proj|k_proj|o_proj|up_proj|gate_proj|down_proj)$",
|
| 27 |
+
"task_type": "CAUSAL_LM",
|
| 28 |
+
"trainable_token_indices": null,
|
| 29 |
+
"use_dora": false,
|
| 30 |
+
"use_rslora": false
|
| 31 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bac3dac42ff2b33774177629b363ef51a6ffe2511b32abc1920081abedfd5813
|
| 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": 1024,
|
| 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 |
+
"task1+3_dpo_absgt14_noidk__pari0.8_0608.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/task1+3_dpo_absgt14_noidk__pari0.8_0608/v0-20250610-025054",
|
| 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": 4,
|
| 95 |
+
"per_device_eval_batch_size": 4,
|
| 96 |
+
"per_gpu_train_batch_size": null,
|
| 97 |
+
"per_gpu_eval_batch_size": null,
|
| 98 |
+
"gradient_accumulation_steps": 16,
|
| 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": 2.0,
|
| 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/task1+3_dpo_absgt14_noidk__pari0.8_0608/v0-20250610-025054/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 0x77cc31936d40>, 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/task1+3_dpo_absgt14_noidk__pari0.8_0608/v0-20250610-025054', 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=4, per_device_eval_batch_size=4, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=16, 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=2.0, 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/task1+3_dpo_absgt14_noidk__pari0.8_0608/v0-20250610-025054/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/task1+3_dpo_absgt14_noidk__pari0.8_0608/v0-20250610-025054', 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=1024, 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:bf849250e153a0954d7061fc3d461246955b3b3a5820022d96382b1557016165
|
| 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": 0.9032743695897629,
|
| 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": 0.0060218291305984195,
|
| 14 |
+
"grad_norm": 7.15871000289917,
|
| 15 |
+
"learning_rate": 1.1764705882352942e-05,
|
| 16 |
+
"logits/chosen": -0.32459306716918945,
|
| 17 |
+
"logits/rejected": -0.5000401735305786,
|
| 18 |
+
"logps/chosen": -3.3705074787139893,
|
| 19 |
+
"logps/rejected": -19.54801368713379,
|
| 20 |
+
"loss": 1.1142030954360962,
|
| 21 |
+
"memory(GiB)": 46.13,
|
| 22 |
+
"nll_loss": 0.4210559129714966,
|
| 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)": 0.005428
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.012043658261196839,
|
| 32 |
+
"grad_norm": 6.311911106109619,
|
| 33 |
+
"learning_rate": 2.3529411764705884e-05,
|
| 34 |
+
"logits/chosen": -0.41802966594696045,
|
| 35 |
+
"logits/rejected": -0.5096024870872498,
|
| 36 |
+
"logps/chosen": -8.828028678894043,
|
| 37 |
+
"logps/rejected": -13.557395935058594,
|
| 38 |
+
"loss": 1.4290869235992432,
|
| 39 |
+
"memory(GiB)": 46.13,
|
| 40 |
+
"nll_loss": 0.7359397411346436,
|
| 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)": 0.005545
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.018065487391795258,
|
| 50 |
+
"grad_norm": 6.693342685699463,
|
| 51 |
+
"learning_rate": 3.529411764705883e-05,
|
| 52 |
+
"logits/chosen": -0.3694347143173218,
|
| 53 |
+
"logits/rejected": -0.4798021912574768,
|
| 54 |
+
"logps/chosen": -6.542965412139893,
|
| 55 |
+
"logps/rejected": -17.62507438659668,
|
| 56 |
+
"loss": 1.2220425605773926,
|
| 57 |
+
"memory(GiB)": 46.13,
|
| 58 |
+
"nll_loss": 0.5347560048103333,
|
| 59 |
+
"rewards/accuracies": 0.734375,
|
| 60 |
+
"rewards/chosen": 0.007337508723139763,
|
| 61 |
+
"rewards/margins": 0.011953208595514297,
|
| 62 |
+
"rewards/rejected": -0.004615699406713247,
|
| 63 |
+
"step": 3,
|
| 64 |
+
"train_speed(iter/s)": 0.005572
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.024087316522393678,
|
| 68 |
+
"grad_norm": 4.8458251953125,
|
| 69 |
+
"learning_rate": 4.705882352941177e-05,
|
| 70 |
+
"logits/chosen": -0.36237555742263794,
|
| 71 |
+
"logits/rejected": -0.5201014280319214,
|
| 72 |
+
"logps/chosen": -5.53680944442749,
|
| 73 |
+
"logps/rejected": -16.475929260253906,
|
| 74 |
+
"loss": 1.2521781921386719,
|
| 75 |
+
"memory(GiB)": 46.13,
|
| 76 |
+
"nll_loss": 0.579364538192749,
|
| 77 |
+
"rewards/accuracies": 0.734375,
|
| 78 |
+
"rewards/chosen": 0.01270484272390604,
|
| 79 |
+
"rewards/margins": 0.04416074603796005,
|
| 80 |
+
"rewards/rejected": -0.03145590424537659,
|
| 81 |
+
"step": 4,
|
| 82 |
+
"train_speed(iter/s)": 0.005594
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.030109145652992095,
|
| 86 |
+
"grad_norm": 3.684136152267456,
|
| 87 |
+
"learning_rate": 5.882352941176471e-05,
|
| 88 |
+
"logits/chosen": -0.3457816541194916,
|
| 89 |
+
"logits/rejected": -0.48865967988967896,
|
| 90 |
+
"logps/chosen": -4.437034606933594,
|
| 91 |
+
"logps/rejected": -20.260162353515625,
|
| 92 |
+
"loss": 0.9991621375083923,
|
| 93 |
+
"memory(GiB)": 46.13,
|
| 94 |
+
"nll_loss": 0.38671040534973145,
|
| 95 |
+
"rewards/accuracies": 0.734375,
|
| 96 |
+
"rewards/chosen": 0.002155877649784088,
|
| 97 |
+
"rewards/margins": 0.2208954095840454,
|
| 98 |
+
"rewards/rejected": -0.21873953938484192,
|
| 99 |
+
"step": 5,
|
| 100 |
+
"train_speed(iter/s)": 0.005606
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.036130974783590515,
|
| 104 |
+
"grad_norm": 6.8094706535339355,
|
| 105 |
+
"learning_rate": 7.058823529411765e-05,
|
| 106 |
+
"logits/chosen": -0.37496131658554077,
|
| 107 |
+
"logits/rejected": -0.499918133020401,
|
| 108 |
+
"logps/chosen": -5.312520980834961,
|
| 109 |
+
"logps/rejected": -18.596221923828125,
|
| 110 |
+
"loss": 1.1320774555206299,
|
| 111 |
+
"memory(GiB)": 46.13,
|
| 112 |
+
"nll_loss": 0.5494582056999207,
|
| 113 |
+
"rewards/accuracies": 0.75,
|
| 114 |
+
"rewards/chosen": -0.11241161823272705,
|
| 115 |
+
"rewards/margins": 0.37036794424057007,
|
| 116 |
+
"rewards/rejected": -0.4827795624732971,
|
| 117 |
+
"step": 6,
|
| 118 |
+
"train_speed(iter/s)": 0.005617
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.042152803914188935,
|
| 122 |
+
"grad_norm": 6.563366889953613,
|
| 123 |
+
"learning_rate": 8.23529411764706e-05,
|
| 124 |
+
"logits/chosen": -0.3803170323371887,
|
| 125 |
+
"logits/rejected": -0.43647128343582153,
|
| 126 |
+
"logps/chosen": -12.155686378479004,
|
| 127 |
+
"logps/rejected": -19.252357482910156,
|
| 128 |
+
"loss": 1.275801420211792,
|
| 129 |
+
"memory(GiB)": 46.13,
|
| 130 |
+
"nll_loss": 0.6294467449188232,
|
| 131 |
+
"rewards/accuracies": 0.609375,
|
| 132 |
+
"rewards/chosen": -0.2581723928451538,
|
| 133 |
+
"rewards/margins": 0.26009732484817505,
|
| 134 |
+
"rewards/rejected": -0.5182697176933289,
|
| 135 |
+
"step": 7,
|
| 136 |
+
"train_speed(iter/s)": 0.005615
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.048174633044787356,
|
| 140 |
+
"grad_norm": 5.275461673736572,
|
| 141 |
+
"learning_rate": 9.411764705882353e-05,
|
| 142 |
+
"logits/chosen": -0.30787912011146545,
|
| 143 |
+
"logits/rejected": -0.420376718044281,
|
| 144 |
+
"logps/chosen": -7.271793842315674,
|
| 145 |
+
"logps/rejected": -17.376087188720703,
|
| 146 |
+
"loss": 1.2506061792373657,
|
| 147 |
+
"memory(GiB)": 46.13,
|
| 148 |
+
"nll_loss": 0.6039291024208069,
|
| 149 |
+
"rewards/accuracies": 0.609375,
|
| 150 |
+
"rewards/chosen": -0.07998734712600708,
|
| 151 |
+
"rewards/margins": 0.2555825710296631,
|
| 152 |
+
"rewards/rejected": -0.3355698883533478,
|
| 153 |
+
"step": 8,
|
| 154 |
+
"train_speed(iter/s)": 0.005615
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.054196462175385776,
|
| 158 |
+
"grad_norm": 3.0540287494659424,
|
| 159 |
+
"learning_rate": 0.00010588235294117647,
|
| 160 |
+
"logits/chosen": -0.30379173159599304,
|
| 161 |
+
"logits/rejected": -0.41216912865638733,
|
| 162 |
+
"logps/chosen": -6.820768356323242,
|
| 163 |
+
"logps/rejected": -16.3853816986084,
|
| 164 |
+
"loss": 1.1634117364883423,
|
| 165 |
+
"memory(GiB)": 46.13,
|
| 166 |
+
"nll_loss": 0.5367689728736877,
|
| 167 |
+
"rewards/accuracies": 0.65625,
|
| 168 |
+
"rewards/chosen": 0.04989926144480705,
|
| 169 |
+
"rewards/margins": 0.1963551640510559,
|
| 170 |
+
"rewards/rejected": -0.14645591378211975,
|
| 171 |
+
"step": 9,
|
| 172 |
+
"train_speed(iter/s)": 0.005615
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.06021829130598419,
|
| 176 |
+
"grad_norm": 2.3425395488739014,
|
| 177 |
+
"learning_rate": 0.00011764705882352942,
|
| 178 |
+
"logits/chosen": -0.3369804918766022,
|
| 179 |
+
"logits/rejected": -0.4523884356021881,
|
| 180 |
+
"logps/chosen": -7.222663402557373,
|
| 181 |
+
"logps/rejected": -17.98470687866211,
|
| 182 |
+
"loss": 1.0803574323654175,
|
| 183 |
+
"memory(GiB)": 46.13,
|
| 184 |
+
"nll_loss": 0.4635474979877472,
|
| 185 |
+
"rewards/accuracies": 0.75,
|
| 186 |
+
"rewards/chosen": 0.0936385914683342,
|
| 187 |
+
"rewards/margins": 0.19075465202331543,
|
| 188 |
+
"rewards/rejected": -0.09711606800556183,
|
| 189 |
+
"step": 10,
|
| 190 |
+
"train_speed(iter/s)": 0.005616
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.06624012043658262,
|
| 194 |
+
"grad_norm": 2.2927637100219727,
|
| 195 |
+
"learning_rate": 0.00012941176470588237,
|
| 196 |
+
"logits/chosen": -0.37337860465049744,
|
| 197 |
+
"logits/rejected": -0.4716382920742035,
|
| 198 |
+
"logps/chosen": -8.593006134033203,
|
| 199 |
+
"logps/rejected": -17.979745864868164,
|
| 200 |
+
"loss": 1.1164621114730835,
|
| 201 |
+
"memory(GiB)": 46.13,
|
| 202 |
+
"nll_loss": 0.49796244502067566,
|
| 203 |
+
"rewards/accuracies": 0.625,
|
| 204 |
+
"rewards/chosen": 0.06070924177765846,
|
| 205 |
+
"rewards/margins": 0.19886666536331177,
|
| 206 |
+
"rewards/rejected": -0.1381574273109436,
|
| 207 |
+
"step": 11,
|
| 208 |
+
"train_speed(iter/s)": 0.005619
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.07226194956718103,
|
| 212 |
+
"grad_norm": 2.380297899246216,
|
| 213 |
+
"learning_rate": 0.0001411764705882353,
|
| 214 |
+
"logits/chosen": -0.4056618809700012,
|
| 215 |
+
"logits/rejected": -0.47582757472991943,
|
| 216 |
+
"logps/chosen": -7.429853916168213,
|
| 217 |
+
"logps/rejected": -16.317821502685547,
|
| 218 |
+
"loss": 1.167059302330017,
|
| 219 |
+
"memory(GiB)": 46.13,
|
| 220 |
+
"nll_loss": 0.5555555820465088,
|
| 221 |
+
"rewards/accuracies": 0.65625,
|
| 222 |
+
"rewards/chosen": -0.0010902846697717905,
|
| 223 |
+
"rewards/margins": 0.21717122197151184,
|
| 224 |
+
"rewards/rejected": -0.2182615101337433,
|
| 225 |
+
"step": 12,
|
| 226 |
+
"train_speed(iter/s)": 0.005622
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.07828377869777944,
|
| 230 |
+
"grad_norm": 1.7392877340316772,
|
| 231 |
+
"learning_rate": 0.00015294117647058822,
|
| 232 |
+
"logits/chosen": -0.4110736846923828,
|
| 233 |
+
"logits/rejected": -0.4770917296409607,
|
| 234 |
+
"logps/chosen": -8.422821044921875,
|
| 235 |
+
"logps/rejected": -16.356956481933594,
|
| 236 |
+
"loss": 1.1002569198608398,
|
| 237 |
+
"memory(GiB)": 46.13,
|
| 238 |
+
"nll_loss": 0.5357962250709534,
|
| 239 |
+
"rewards/accuracies": 0.796875,
|
| 240 |
+
"rewards/chosen": 0.12992499768733978,
|
| 241 |
+
"rewards/margins": 0.3372637927532196,
|
| 242 |
+
"rewards/rejected": -0.20733879506587982,
|
| 243 |
+
"step": 13,
|
| 244 |
+
"train_speed(iter/s)": 0.005625
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.08430560782837787,
|
| 248 |
+
"grad_norm": 5.562183856964111,
|
| 249 |
+
"learning_rate": 0.0001647058823529412,
|
| 250 |
+
"logits/chosen": -0.32676243782043457,
|
| 251 |
+
"logits/rejected": -0.4272843599319458,
|
| 252 |
+
"logps/chosen": -8.393424034118652,
|
| 253 |
+
"logps/rejected": -19.974212646484375,
|
| 254 |
+
"loss": 1.156798005104065,
|
| 255 |
+
"memory(GiB)": 46.13,
|
| 256 |
+
"nll_loss": 0.6415989995002747,
|
| 257 |
+
"rewards/accuracies": 0.78125,
|
| 258 |
+
"rewards/chosen": -0.03989575803279877,
|
| 259 |
+
"rewards/margins": 0.5587274432182312,
|
| 260 |
+
"rewards/rejected": -0.5986231565475464,
|
| 261 |
+
"step": 14,
|
| 262 |
+
"train_speed(iter/s)": 0.005627
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.09032743695897628,
|
| 266 |
+
"grad_norm": 3.5578579902648926,
|
| 267 |
+
"learning_rate": 0.00017647058823529413,
|
| 268 |
+
"logits/chosen": -0.36884137988090515,
|
| 269 |
+
"logits/rejected": -0.47317999601364136,
|
| 270 |
+
"logps/chosen": -6.998829364776611,
|
| 271 |
+
"logps/rejected": -18.39483642578125,
|
| 272 |
+
"loss": 1.265438199043274,
|
| 273 |
+
"memory(GiB)": 46.13,
|
| 274 |
+
"nll_loss": 0.7606177926063538,
|
| 275 |
+
"rewards/accuracies": 0.859375,
|
| 276 |
+
"rewards/chosen": 0.002129599452018738,
|
| 277 |
+
"rewards/margins": 0.5497386455535889,
|
| 278 |
+
"rewards/rejected": -0.5476090908050537,
|
| 279 |
+
"step": 15,
|
| 280 |
+
"train_speed(iter/s)": 0.00563
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.09634926608957471,
|
| 284 |
+
"grad_norm": 3.6414742469787598,
|
| 285 |
+
"learning_rate": 0.00018823529411764707,
|
| 286 |
+
"logits/chosen": -0.3789006471633911,
|
| 287 |
+
"logits/rejected": -0.4743453860282898,
|
| 288 |
+
"logps/chosen": -6.895637512207031,
|
| 289 |
+
"logps/rejected": -20.256675720214844,
|
| 290 |
+
"loss": 1.064023494720459,
|
| 291 |
+
"memory(GiB)": 46.13,
|
| 292 |
+
"nll_loss": 0.5308047533035278,
|
| 293 |
+
"rewards/accuracies": 0.8125,
|
| 294 |
+
"rewards/chosen": 0.055957891047000885,
|
| 295 |
+
"rewards/margins": 0.40152865648269653,
|
| 296 |
+
"rewards/rejected": -0.34557074308395386,
|
| 297 |
+
"step": 16,
|
| 298 |
+
"train_speed(iter/s)": 0.005632
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.10237109522017313,
|
| 302 |
+
"grad_norm": 3.263599157333374,
|
| 303 |
+
"learning_rate": 0.0002,
|
| 304 |
+
"logits/chosen": -0.28460752964019775,
|
| 305 |
+
"logits/rejected": -0.42314791679382324,
|
| 306 |
+
"logps/chosen": -5.295862197875977,
|
| 307 |
+
"logps/rejected": -21.666519165039062,
|
| 308 |
+
"loss": 0.9750788807868958,
|
| 309 |
+
"memory(GiB)": 46.13,
|
| 310 |
+
"nll_loss": 0.5031211376190186,
|
| 311 |
+
"rewards/accuracies": 0.90625,
|
| 312 |
+
"rewards/chosen": 0.08322687447071075,
|
| 313 |
+
"rewards/margins": 0.5906113386154175,
|
| 314 |
+
"rewards/rejected": -0.5073844194412231,
|
| 315 |
+
"step": 17,
|
| 316 |
+
"train_speed(iter/s)": 0.005633
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.10839292435077155,
|
| 320 |
+
"grad_norm": 2.0240275859832764,
|
| 321 |
+
"learning_rate": 0.00019999502669559432,
|
| 322 |
+
"logits/chosen": -0.3283552825450897,
|
| 323 |
+
"logits/rejected": -0.40596985816955566,
|
| 324 |
+
"logps/chosen": -6.424624919891357,
|
| 325 |
+
"logps/rejected": -17.71906852722168,
|
| 326 |
+
"loss": 0.9781989455223083,
|
| 327 |
+
"memory(GiB)": 46.13,
|
| 328 |
+
"nll_loss": 0.47217291593551636,
|
| 329 |
+
"rewards/accuracies": 0.796875,
|
| 330 |
+
"rewards/chosen": -0.0207663644105196,
|
| 331 |
+
"rewards/margins": 0.5476341247558594,
|
| 332 |
+
"rewards/rejected": -0.568400502204895,
|
| 333 |
+
"step": 18,
|
| 334 |
+
"train_speed(iter/s)": 0.005634
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.11441475348136997,
|
| 338 |
+
"grad_norm": 1.682889699935913,
|
| 339 |
+
"learning_rate": 0.00019998010727705236,
|
| 340 |
+
"logits/chosen": -0.2821420729160309,
|
| 341 |
+
"logits/rejected": -0.38918256759643555,
|
| 342 |
+
"logps/chosen": -5.882609844207764,
|
| 343 |
+
"logps/rejected": -20.25113296508789,
|
| 344 |
+
"loss": 0.9472936391830444,
|
| 345 |
+
"memory(GiB)": 46.13,
|
| 346 |
+
"nll_loss": 0.4797293543815613,
|
| 347 |
+
"rewards/accuracies": 0.78125,
|
| 348 |
+
"rewards/chosen": 0.06502628326416016,
|
| 349 |
+
"rewards/margins": 0.6772950291633606,
|
| 350 |
+
"rewards/rejected": -0.6122686862945557,
|
| 351 |
+
"step": 19,
|
| 352 |
+
"train_speed(iter/s)": 0.005633
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.12043658261196838,
|
| 356 |
+
"grad_norm": 2.9386143684387207,
|
| 357 |
+
"learning_rate": 0.00019995524322835034,
|
| 358 |
+
"logits/chosen": -0.2649421691894531,
|
| 359 |
+
"logits/rejected": -0.34010952711105347,
|
| 360 |
+
"logps/chosen": -10.95669174194336,
|
| 361 |
+
"logps/rejected": -24.044157028198242,
|
| 362 |
+
"loss": 1.1134816408157349,
|
| 363 |
+
"memory(GiB)": 46.13,
|
| 364 |
+
"nll_loss": 0.546178936958313,
|
| 365 |
+
"rewards/accuracies": 0.796875,
|
| 366 |
+
"rewards/chosen": -0.18016880750656128,
|
| 367 |
+
"rewards/margins": 0.5494988560676575,
|
| 368 |
+
"rewards/rejected": -0.729667603969574,
|
| 369 |
+
"step": 20,
|
| 370 |
+
"train_speed(iter/s)": 0.00563
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.1264584117425668,
|
| 374 |
+
"grad_norm": 1.6700323820114136,
|
| 375 |
+
"learning_rate": 0.00019992043702261793,
|
| 376 |
+
"logits/chosen": -0.2561277151107788,
|
| 377 |
+
"logits/rejected": -0.3126121163368225,
|
| 378 |
+
"logps/chosen": -6.485073566436768,
|
| 379 |
+
"logps/rejected": -19.772449493408203,
|
| 380 |
+
"loss": 0.9205772876739502,
|
| 381 |
+
"memory(GiB)": 46.13,
|
| 382 |
+
"nll_loss": 0.42214635014533997,
|
| 383 |
+
"rewards/accuracies": 0.765625,
|
| 384 |
+
"rewards/chosen": -0.0028960183262825012,
|
| 385 |
+
"rewards/margins": 0.6638913750648499,
|
| 386 |
+
"rewards/rejected": -0.6667873859405518,
|
| 387 |
+
"step": 21,
|
| 388 |
+
"train_speed(iter/s)": 0.00563
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.13248024087316523,
|
| 392 |
+
"grad_norm": 2.0114383697509766,
|
| 393 |
+
"learning_rate": 0.00019987569212189224,
|
| 394 |
+
"logits/chosen": -0.2601704001426697,
|
| 395 |
+
"logits/rejected": -0.3789623975753784,
|
| 396 |
+
"logps/chosen": -6.645152568817139,
|
| 397 |
+
"logps/rejected": -21.876020431518555,
|
| 398 |
+
"loss": 1.1017377376556396,
|
| 399 |
+
"memory(GiB)": 46.13,
|
| 400 |
+
"nll_loss": 0.6103405356407166,
|
| 401 |
+
"rewards/accuracies": 0.828125,
|
| 402 |
+
"rewards/chosen": 0.04465393349528313,
|
| 403 |
+
"rewards/margins": 0.6134660840034485,
|
| 404 |
+
"rewards/rejected": -0.5688121318817139,
|
| 405 |
+
"step": 22,
|
| 406 |
+
"train_speed(iter/s)": 0.005629
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.13850207000376363,
|
| 410 |
+
"grad_norm": 3.031888961791992,
|
| 411 |
+
"learning_rate": 0.0001998210129767735,
|
| 412 |
+
"logits/chosen": -0.26405903697013855,
|
| 413 |
+
"logits/rejected": -0.3669392466545105,
|
| 414 |
+
"logps/chosen": -6.921745300292969,
|
| 415 |
+
"logps/rejected": -21.70663070678711,
|
| 416 |
+
"loss": 0.9795225858688354,
|
| 417 |
+
"memory(GiB)": 46.13,
|
| 418 |
+
"nll_loss": 0.4959058463573456,
|
| 419 |
+
"rewards/accuracies": 0.796875,
|
| 420 |
+
"rewards/chosen": 0.10062012821435928,
|
| 421 |
+
"rewards/margins": 0.6365569829940796,
|
| 422 |
+
"rewards/rejected": -0.5359368324279785,
|
| 423 |
+
"step": 23,
|
| 424 |
+
"train_speed(iter/s)": 0.005629
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.14452389913436206,
|
| 428 |
+
"grad_norm": 2.15742564201355,
|
| 429 |
+
"learning_rate": 0.00019975640502598244,
|
| 430 |
+
"logits/chosen": -0.2354922890663147,
|
| 431 |
+
"logits/rejected": -0.36152032017707825,
|
| 432 |
+
"logps/chosen": -4.718395709991455,
|
| 433 |
+
"logps/rejected": -24.57033348083496,
|
| 434 |
+
"loss": 0.7706338763237,
|
| 435 |
+
"memory(GiB)": 46.13,
|
| 436 |
+
"nll_loss": 0.3560357987880707,
|
| 437 |
+
"rewards/accuracies": 0.84375,
|
| 438 |
+
"rewards/chosen": 0.09812817722558975,
|
| 439 |
+
"rewards/margins": 0.9238024950027466,
|
| 440 |
+
"rewards/rejected": -0.825674295425415,
|
| 441 |
+
"step": 24,
|
| 442 |
+
"train_speed(iter/s)": 0.005629
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.1505457282649605,
|
| 446 |
+
"grad_norm": 3.1512320041656494,
|
| 447 |
+
"learning_rate": 0.0001996818746958191,
|
| 448 |
+
"logits/chosen": -0.21456243097782135,
|
| 449 |
+
"logits/rejected": -0.34130796790122986,
|
| 450 |
+
"logps/chosen": -6.996722221374512,
|
| 451 |
+
"logps/rejected": -30.622554779052734,
|
| 452 |
+
"loss": 0.8399394750595093,
|
| 453 |
+
"memory(GiB)": 46.13,
|
| 454 |
+
"nll_loss": 0.49700063467025757,
|
| 455 |
+
"rewards/accuracies": 0.84375,
|
| 456 |
+
"rewards/chosen": -0.024456650018692017,
|
| 457 |
+
"rewards/margins": 1.4392318725585938,
|
| 458 |
+
"rewards/rejected": -1.4636887311935425,
|
| 459 |
+
"step": 25,
|
| 460 |
+
"train_speed(iter/s)": 0.005628
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.1565675573955589,
|
| 464 |
+
"grad_norm": 2.9886889457702637,
|
| 465 |
+
"learning_rate": 0.00019959742939952392,
|
| 466 |
+
"logits/chosen": -0.2622988224029541,
|
| 467 |
+
"logits/rejected": -0.3251183331012726,
|
| 468 |
+
"logps/chosen": -12.495084762573242,
|
| 469 |
+
"logps/rejected": -25.176090240478516,
|
| 470 |
+
"loss": 1.1550263166427612,
|
| 471 |
+
"memory(GiB)": 46.13,
|
| 472 |
+
"nll_loss": 0.7094165086746216,
|
| 473 |
+
"rewards/accuracies": 0.8125,
|
| 474 |
+
"rewards/chosen": -0.19754531979560852,
|
| 475 |
+
"rewards/margins": 1.1228585243225098,
|
| 476 |
+
"rewards/rejected": -1.3204039335250854,
|
| 477 |
+
"step": 26,
|
| 478 |
+
"train_speed(iter/s)": 0.00563
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.16258938652615731,
|
| 482 |
+
"grad_norm": 2.7348685264587402,
|
| 483 |
+
"learning_rate": 0.00019950307753654017,
|
| 484 |
+
"logits/chosen": -0.23797279596328735,
|
| 485 |
+
"logits/rejected": -0.3161388337612152,
|
| 486 |
+
"logps/chosen": -8.345640182495117,
|
| 487 |
+
"logps/rejected": -24.480266571044922,
|
| 488 |
+
"loss": 1.0113354921340942,
|
| 489 |
+
"memory(GiB)": 46.13,
|
| 490 |
+
"nll_loss": 0.5643945336341858,
|
| 491 |
+
"rewards/accuracies": 0.859375,
|
| 492 |
+
"rewards/chosen": 0.05142675340175629,
|
| 493 |
+
"rewards/margins": 1.0910050868988037,
|
| 494 |
+
"rewards/rejected": -1.0395784378051758,
|
| 495 |
+
"step": 27,
|
| 496 |
+
"train_speed(iter/s)": 0.00563
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.16861121565675574,
|
| 500 |
+
"grad_norm": 3.4209959506988525,
|
| 501 |
+
"learning_rate": 0.00019939882849167852,
|
| 502 |
+
"logits/chosen": -0.2834317684173584,
|
| 503 |
+
"logits/rejected": -0.4013361930847168,
|
| 504 |
+
"logps/chosen": -6.27006196975708,
|
| 505 |
+
"logps/rejected": -23.9411678314209,
|
| 506 |
+
"loss": 0.9620530605316162,
|
| 507 |
+
"memory(GiB)": 46.13,
|
| 508 |
+
"nll_loss": 0.5452341437339783,
|
| 509 |
+
"rewards/accuracies": 0.84375,
|
| 510 |
+
"rewards/chosen": 0.08509417623281479,
|
| 511 |
+
"rewards/margins": 1.0887510776519775,
|
| 512 |
+
"rewards/rejected": -1.0036569833755493,
|
| 513 |
+
"step": 28,
|
| 514 |
+
"train_speed(iter/s)": 0.005632
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.17463304478735417,
|
| 518 |
+
"grad_norm": 3.021099805831909,
|
| 519 |
+
"learning_rate": 0.00019928469263418374,
|
| 520 |
+
"logits/chosen": -0.29641368985176086,
|
| 521 |
+
"logits/rejected": -0.3944772779941559,
|
| 522 |
+
"logps/chosen": -6.065124988555908,
|
| 523 |
+
"logps/rejected": -19.367380142211914,
|
| 524 |
+
"loss": 0.8545750379562378,
|
| 525 |
+
"memory(GiB)": 46.13,
|
| 526 |
+
"nll_loss": 0.3696524500846863,
|
| 527 |
+
"rewards/accuracies": 0.859375,
|
| 528 |
+
"rewards/chosen": -0.008128602989017963,
|
| 529 |
+
"rewards/margins": 0.7214774489402771,
|
| 530 |
+
"rewards/rejected": -0.729606032371521,
|
| 531 |
+
"step": 29,
|
| 532 |
+
"train_speed(iter/s)": 0.005633
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.18065487391795257,
|
| 536 |
+
"grad_norm": 2.237170457839966,
|
| 537 |
+
"learning_rate": 0.00019916068131670302,
|
| 538 |
+
"logits/chosen": -0.2665432393550873,
|
| 539 |
+
"logits/rejected": -0.41999131441116333,
|
| 540 |
+
"logps/chosen": -3.6829960346221924,
|
| 541 |
+
"logps/rejected": -24.998981475830078,
|
| 542 |
+
"loss": 0.7719818353652954,
|
| 543 |
+
"memory(GiB)": 46.13,
|
| 544 |
+
"nll_loss": 0.3786780834197998,
|
| 545 |
+
"rewards/accuracies": 0.890625,
|
| 546 |
+
"rewards/chosen": 0.15382343530654907,
|
| 547 |
+
"rewards/margins": 1.0922845602035522,
|
| 548 |
+
"rewards/rejected": -0.9384610652923584,
|
| 549 |
+
"step": 30,
|
| 550 |
+
"train_speed(iter/s)": 0.005633
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.186676703048551,
|
| 554 |
+
"grad_norm": 2.299628496170044,
|
| 555 |
+
"learning_rate": 0.00019902680687415705,
|
| 556 |
+
"logits/chosen": -0.33755987882614136,
|
| 557 |
+
"logits/rejected": -0.37323832511901855,
|
| 558 |
+
"logps/chosen": -11.234481811523438,
|
| 559 |
+
"logps/rejected": -24.92943000793457,
|
| 560 |
+
"loss": 1.0785183906555176,
|
| 561 |
+
"memory(GiB)": 46.13,
|
| 562 |
+
"nll_loss": 0.6123806834220886,
|
| 563 |
+
"rewards/accuracies": 0.765625,
|
| 564 |
+
"rewards/chosen": 0.03546600416302681,
|
| 565 |
+
"rewards/margins": 0.9260610342025757,
|
| 566 |
+
"rewards/rejected": -0.890595018863678,
|
| 567 |
+
"step": 31,
|
| 568 |
+
"train_speed(iter/s)": 0.005633
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.19269853217914942,
|
| 572 |
+
"grad_norm": 2.0755157470703125,
|
| 573 |
+
"learning_rate": 0.00019888308262251285,
|
| 574 |
+
"logits/chosen": -0.2504952847957611,
|
| 575 |
+
"logits/rejected": -0.3852018713951111,
|
| 576 |
+
"logps/chosen": -5.893071174621582,
|
| 577 |
+
"logps/rejected": -33.010154724121094,
|
| 578 |
+
"loss": 0.7333768010139465,
|
| 579 |
+
"memory(GiB)": 46.13,
|
| 580 |
+
"nll_loss": 0.4039708375930786,
|
| 581 |
+
"rewards/accuracies": 0.859375,
|
| 582 |
+
"rewards/chosen": -0.07270973920822144,
|
| 583 |
+
"rewards/margins": 1.7266591787338257,
|
| 584 |
+
"rewards/rejected": -1.7993686199188232,
|
| 585 |
+
"step": 32,
|
| 586 |
+
"train_speed(iter/s)": 0.005634
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.19872036130974782,
|
| 590 |
+
"grad_norm": 3.686795711517334,
|
| 591 |
+
"learning_rate": 0.00019872952285745959,
|
| 592 |
+
"logits/chosen": -0.30149099230766296,
|
| 593 |
+
"logits/rejected": -0.4157900810241699,
|
| 594 |
+
"logps/chosen": -7.993706703186035,
|
| 595 |
+
"logps/rejected": -27.117816925048828,
|
| 596 |
+
"loss": 1.0753562450408936,
|
| 597 |
+
"memory(GiB)": 46.13,
|
| 598 |
+
"nll_loss": 0.6016828417778015,
|
| 599 |
+
"rewards/accuracies": 0.796875,
|
| 600 |
+
"rewards/chosen": -0.10546419769525528,
|
| 601 |
+
"rewards/margins": 1.0370161533355713,
|
| 602 |
+
"rewards/rejected": -1.1424803733825684,
|
| 603 |
+
"step": 33,
|
| 604 |
+
"train_speed(iter/s)": 0.005634
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.20474219044034625,
|
| 608 |
+
"grad_norm": 3.085048198699951,
|
| 609 |
+
"learning_rate": 0.0001985661428529863,
|
| 610 |
+
"logits/chosen": -0.31711238622665405,
|
| 611 |
+
"logits/rejected": -0.41008925437927246,
|
| 612 |
+
"logps/chosen": -6.8641557693481445,
|
| 613 |
+
"logps/rejected": -27.393281936645508,
|
| 614 |
+
"loss": 1.054256558418274,
|
| 615 |
+
"memory(GiB)": 46.13,
|
| 616 |
+
"nll_loss": 0.6109166741371155,
|
| 617 |
+
"rewards/accuracies": 0.765625,
|
| 618 |
+
"rewards/chosen": -0.06303416192531586,
|
| 619 |
+
"rewards/margins": 1.1662019491195679,
|
| 620 |
+
"rewards/rejected": -1.229236125946045,
|
| 621 |
+
"step": 34,
|
| 622 |
+
"train_speed(iter/s)": 0.005634
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.21076401957094468,
|
| 626 |
+
"grad_norm": 1.7560195922851562,
|
| 627 |
+
"learning_rate": 0.00019839295885986296,
|
| 628 |
+
"logits/chosen": -0.337604820728302,
|
| 629 |
+
"logits/rejected": -0.39799487590789795,
|
| 630 |
+
"logps/chosen": -6.864170551300049,
|
| 631 |
+
"logps/rejected": -20.806344985961914,
|
| 632 |
+
"loss": 0.9619123339653015,
|
| 633 |
+
"memory(GiB)": 46.13,
|
| 634 |
+
"nll_loss": 0.48513123393058777,
|
| 635 |
+
"rewards/accuracies": 0.828125,
|
| 636 |
+
"rewards/chosen": 0.10183387249708176,
|
| 637 |
+
"rewards/margins": 0.7188172340393066,
|
| 638 |
+
"rewards/rejected": -0.6169832944869995,
|
| 639 |
+
"step": 35,
|
| 640 |
+
"train_speed(iter/s)": 0.005634
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.2167858487015431,
|
| 644 |
+
"grad_norm": 2.6364471912384033,
|
| 645 |
+
"learning_rate": 0.0001982099881040239,
|
| 646 |
+
"logits/chosen": -0.36966580152511597,
|
| 647 |
+
"logits/rejected": -0.43631935119628906,
|
| 648 |
+
"logps/chosen": -6.647594451904297,
|
| 649 |
+
"logps/rejected": -21.291913986206055,
|
| 650 |
+
"loss": 0.9380239844322205,
|
| 651 |
+
"memory(GiB)": 46.13,
|
| 652 |
+
"nll_loss": 0.4613775312900543,
|
| 653 |
+
"rewards/accuracies": 0.78125,
|
| 654 |
+
"rewards/chosen": 0.06358329951763153,
|
| 655 |
+
"rewards/margins": 0.7893975973129272,
|
| 656 |
+
"rewards/rejected": -0.7258143424987793,
|
| 657 |
+
"step": 36,
|
| 658 |
+
"train_speed(iter/s)": 0.005634
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.2228076778321415,
|
| 662 |
+
"grad_norm": 2.354177474975586,
|
| 663 |
+
"learning_rate": 0.00019801724878485438,
|
| 664 |
+
"logits/chosen": -0.36403897404670715,
|
| 665 |
+
"logits/rejected": -0.45796939730644226,
|
| 666 |
+
"logps/chosen": -5.62083625793457,
|
| 667 |
+
"logps/rejected": -23.648834228515625,
|
| 668 |
+
"loss": 0.9243438243865967,
|
| 669 |
+
"memory(GiB)": 46.13,
|
| 670 |
+
"nll_loss": 0.46970799565315247,
|
| 671 |
+
"rewards/accuracies": 0.828125,
|
| 672 |
+
"rewards/chosen": 0.06422886252403259,
|
| 673 |
+
"rewards/margins": 0.8610265851020813,
|
| 674 |
+
"rewards/rejected": -0.7967977523803711,
|
| 675 |
+
"step": 37,
|
| 676 |
+
"train_speed(iter/s)": 0.005634
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.22882950696273993,
|
| 680 |
+
"grad_norm": 2.2988836765289307,
|
| 681 |
+
"learning_rate": 0.00019781476007338058,
|
| 682 |
+
"logits/chosen": -0.3704385459423065,
|
| 683 |
+
"logits/rejected": -0.47973009943962097,
|
| 684 |
+
"logps/chosen": -4.231719017028809,
|
| 685 |
+
"logps/rejected": -20.56240463256836,
|
| 686 |
+
"loss": 0.916353702545166,
|
| 687 |
+
"memory(GiB)": 46.13,
|
| 688 |
+
"nll_loss": 0.4209356904029846,
|
| 689 |
+
"rewards/accuracies": 0.859375,
|
| 690 |
+
"rewards/chosen": 0.06186368316411972,
|
| 691 |
+
"rewards/margins": 0.7635293006896973,
|
| 692 |
+
"rewards/rejected": -0.7016656398773193,
|
| 693 |
+
"step": 38,
|
| 694 |
+
"train_speed(iter/s)": 0.005635
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.23485133609333836,
|
| 698 |
+
"grad_norm": 2.009106397628784,
|
| 699 |
+
"learning_rate": 0.00019760254211036244,
|
| 700 |
+
"logits/chosen": -0.3372359573841095,
|
| 701 |
+
"logits/rejected": -0.453107088804245,
|
| 702 |
+
"logps/chosen": -5.785426616668701,
|
| 703 |
+
"logps/rejected": -23.19124412536621,
|
| 704 |
+
"loss": 0.9124959707260132,
|
| 705 |
+
"memory(GiB)": 46.13,
|
| 706 |
+
"nll_loss": 0.44838690757751465,
|
| 707 |
+
"rewards/accuracies": 0.859375,
|
| 708 |
+
"rewards/chosen": 0.04832564294338226,
|
| 709 |
+
"rewards/margins": 0.937610387802124,
|
| 710 |
+
"rewards/rejected": -0.8892846703529358,
|
| 711 |
+
"step": 39,
|
| 712 |
+
"train_speed(iter/s)": 0.005635
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.24087316522393676,
|
| 716 |
+
"grad_norm": 2.5412213802337646,
|
| 717 |
+
"learning_rate": 0.00019738061600429064,
|
| 718 |
+
"logits/chosen": -0.39427676796913147,
|
| 719 |
+
"logits/rejected": -0.48203372955322266,
|
| 720 |
+
"logps/chosen": -5.785412311553955,
|
| 721 |
+
"logps/rejected": -19.031665802001953,
|
| 722 |
+
"loss": 0.9683349132537842,
|
| 723 |
+
"memory(GiB)": 46.13,
|
| 724 |
+
"nll_loss": 0.5110080242156982,
|
| 725 |
+
"rewards/accuracies": 0.84375,
|
| 726 |
+
"rewards/chosen": 0.022132933139801025,
|
| 727 |
+
"rewards/margins": 0.8074027299880981,
|
| 728 |
+
"rewards/rejected": -0.7852697372436523,
|
| 729 |
+
"step": 40,
|
| 730 |
+
"train_speed(iter/s)": 0.005637
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.24689499435453519,
|
| 734 |
+
"grad_norm": 2.0049972534179688,
|
| 735 |
+
"learning_rate": 0.00019714900382928675,
|
| 736 |
+
"logits/chosen": -0.3044354021549225,
|
| 737 |
+
"logits/rejected": -0.41962650418281555,
|
| 738 |
+
"logps/chosen": -4.612288475036621,
|
| 739 |
+
"logps/rejected": -24.228839874267578,
|
| 740 |
+
"loss": 0.8179229497909546,
|
| 741 |
+
"memory(GiB)": 46.13,
|
| 742 |
+
"nll_loss": 0.4278942048549652,
|
| 743 |
+
"rewards/accuracies": 0.8125,
|
| 744 |
+
"rewards/chosen": 0.07943032681941986,
|
| 745 |
+
"rewards/margins": 1.1561895608901978,
|
| 746 |
+
"rewards/rejected": -1.0767593383789062,
|
| 747 |
+
"step": 41,
|
| 748 |
+
"train_speed(iter/s)": 0.005636
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.2529168234851336,
|
| 752 |
+
"grad_norm": 2.0874667167663574,
|
| 753 |
+
"learning_rate": 0.0001969077286229078,
|
| 754 |
+
"logits/chosen": -0.2736699879169464,
|
| 755 |
+
"logits/rejected": -0.4168338179588318,
|
| 756 |
+
"logps/chosen": -7.539179801940918,
|
| 757 |
+
"logps/rejected": -30.339994430541992,
|
| 758 |
+
"loss": 0.9962757229804993,
|
| 759 |
+
"memory(GiB)": 46.13,
|
| 760 |
+
"nll_loss": 0.6084927320480347,
|
| 761 |
+
"rewards/accuracies": 0.796875,
|
| 762 |
+
"rewards/chosen": 0.046928200870752335,
|
| 763 |
+
"rewards/margins": 1.4784610271453857,
|
| 764 |
+
"rewards/rejected": -1.431532859802246,
|
| 765 |
+
"step": 42,
|
| 766 |
+
"train_speed(iter/s)": 0.005636
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.258938652615732,
|
| 770 |
+
"grad_norm": 3.2748661041259766,
|
| 771 |
+
"learning_rate": 0.00019665681438385473,
|
| 772 |
+
"logits/chosen": -0.26249000430107117,
|
| 773 |
+
"logits/rejected": -0.3955192565917969,
|
| 774 |
+
"logps/chosen": -6.726841449737549,
|
| 775 |
+
"logps/rejected": -26.447267532348633,
|
| 776 |
+
"loss": 0.9923527836799622,
|
| 777 |
+
"memory(GiB)": 46.13,
|
| 778 |
+
"nll_loss": 0.5636712312698364,
|
| 779 |
+
"rewards/accuracies": 0.859375,
|
| 780 |
+
"rewards/chosen": -0.02616805024445057,
|
| 781 |
+
"rewards/margins": 1.2797741889953613,
|
| 782 |
+
"rewards/rejected": -1.3059422969818115,
|
| 783 |
+
"step": 43,
|
| 784 |
+
"train_speed(iter/s)": 0.005637
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.26496048174633047,
|
| 788 |
+
"grad_norm": 4.499722957611084,
|
| 789 |
+
"learning_rate": 0.00019639628606958533,
|
| 790 |
+
"logits/chosen": -0.3380294442176819,
|
| 791 |
+
"logits/rejected": -0.4303723871707916,
|
| 792 |
+
"logps/chosen": -7.925315856933594,
|
| 793 |
+
"logps/rejected": -23.916906356811523,
|
| 794 |
+
"loss": 1.136991024017334,
|
| 795 |
+
"memory(GiB)": 46.13,
|
| 796 |
+
"nll_loss": 0.6508135795593262,
|
| 797 |
+
"rewards/accuracies": 0.765625,
|
| 798 |
+
"rewards/chosen": -0.1142488494515419,
|
| 799 |
+
"rewards/margins": 1.0733367204666138,
|
| 800 |
+
"rewards/rejected": -1.1875855922698975,
|
| 801 |
+
"step": 44,
|
| 802 |
+
"train_speed(iter/s)": 0.005638
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.27098231087692887,
|
| 806 |
+
"grad_norm": 3.695734739303589,
|
| 807 |
+
"learning_rate": 0.0001961261695938319,
|
| 808 |
+
"logits/chosen": -0.2465212196111679,
|
| 809 |
+
"logits/rejected": -0.39715391397476196,
|
| 810 |
+
"logps/chosen": -7.8324666023254395,
|
| 811 |
+
"logps/rejected": -24.97869110107422,
|
| 812 |
+
"loss": 0.9623777270317078,
|
| 813 |
+
"memory(GiB)": 46.13,
|
| 814 |
+
"nll_loss": 0.5307109951972961,
|
| 815 |
+
"rewards/accuracies": 0.828125,
|
| 816 |
+
"rewards/chosen": 0.08475106954574585,
|
| 817 |
+
"rewards/margins": 0.9390297532081604,
|
| 818 |
+
"rewards/rejected": -0.8542786836624146,
|
| 819 |
+
"step": 45,
|
| 820 |
+
"train_speed(iter/s)": 0.005638
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.27700414000752727,
|
| 824 |
+
"grad_norm": 3.1841602325439453,
|
| 825 |
+
"learning_rate": 0.00019584649182402357,
|
| 826 |
+
"logits/chosen": -0.22775670886039734,
|
| 827 |
+
"logits/rejected": -0.3454330265522003,
|
| 828 |
+
"logps/chosen": -6.5436859130859375,
|
| 829 |
+
"logps/rejected": -24.402027130126953,
|
| 830 |
+
"loss": 1.0284608602523804,
|
| 831 |
+
"memory(GiB)": 46.13,
|
| 832 |
+
"nll_loss": 0.5513083338737488,
|
| 833 |
+
"rewards/accuracies": 0.75,
|
| 834 |
+
"rewards/chosen": 0.05229588598012924,
|
| 835 |
+
"rewards/margins": 0.9507507085800171,
|
| 836 |
+
"rewards/rejected": -0.8984547853469849,
|
| 837 |
+
"step": 46,
|
| 838 |
+
"train_speed(iter/s)": 0.005637
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.2830259691381257,
|
| 842 |
+
"grad_norm": 2.0325071811676025,
|
| 843 |
+
"learning_rate": 0.0001955572805786141,
|
| 844 |
+
"logits/chosen": -0.19194024801254272,
|
| 845 |
+
"logits/rejected": -0.32474127411842346,
|
| 846 |
+
"logps/chosen": -4.646670341491699,
|
| 847 |
+
"logps/rejected": -27.415727615356445,
|
| 848 |
+
"loss": 0.7930896878242493,
|
| 849 |
+
"memory(GiB)": 46.13,
|
| 850 |
+
"nll_loss": 0.38733866810798645,
|
| 851 |
+
"rewards/accuracies": 0.828125,
|
| 852 |
+
"rewards/chosen": 0.11078543961048126,
|
| 853 |
+
"rewards/margins": 1.1206859350204468,
|
| 854 |
+
"rewards/rejected": -1.009900450706482,
|
| 855 |
+
"step": 47,
|
| 856 |
+
"train_speed(iter/s)": 0.005637
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.2890477982687241,
|
| 860 |
+
"grad_norm": 2.07770037651062,
|
| 861 |
+
"learning_rate": 0.0001952585646243146,
|
| 862 |
+
"logits/chosen": -0.25676336884498596,
|
| 863 |
+
"logits/rejected": -0.3494555950164795,
|
| 864 |
+
"logps/chosen": -7.388433933258057,
|
| 865 |
+
"logps/rejected": -23.070579528808594,
|
| 866 |
+
"loss": 0.8534475564956665,
|
| 867 |
+
"memory(GiB)": 46.13,
|
| 868 |
+
"nll_loss": 0.42261388897895813,
|
| 869 |
+
"rewards/accuracies": 0.796875,
|
| 870 |
+
"rewards/chosen": 0.006194199435412884,
|
| 871 |
+
"rewards/margins": 0.9558709859848022,
|
| 872 |
+
"rewards/rejected": -0.9496768712997437,
|
| 873 |
+
"step": 48,
|
| 874 |
+
"train_speed(iter/s)": 0.005637
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.2950696273993225,
|
| 878 |
+
"grad_norm": 1.6862146854400635,
|
| 879 |
+
"learning_rate": 0.00019495037367323262,
|
| 880 |
+
"logits/chosen": -0.22853028774261475,
|
| 881 |
+
"logits/rejected": -0.3647540211677551,
|
| 882 |
+
"logps/chosen": -3.3972930908203125,
|
| 883 |
+
"logps/rejected": -25.391393661499023,
|
| 884 |
+
"loss": 0.7584111094474792,
|
| 885 |
+
"memory(GiB)": 46.13,
|
| 886 |
+
"nll_loss": 0.35165295004844666,
|
| 887 |
+
"rewards/accuracies": 0.796875,
|
| 888 |
+
"rewards/chosen": 0.12043334543704987,
|
| 889 |
+
"rewards/margins": 1.1430784463882446,
|
| 890 |
+
"rewards/rejected": -1.022645115852356,
|
| 891 |
+
"step": 49,
|
| 892 |
+
"train_speed(iter/s)": 0.005638
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.301091456529921,
|
| 896 |
+
"grad_norm": 3.027331829071045,
|
| 897 |
+
"learning_rate": 0.00019463273837991643,
|
| 898 |
+
"logits/chosen": -0.2110910564661026,
|
| 899 |
+
"logits/rejected": -0.38482916355133057,
|
| 900 |
+
"logps/chosen": -4.55904483795166,
|
| 901 |
+
"logps/rejected": -31.877506256103516,
|
| 902 |
+
"loss": 0.675049901008606,
|
| 903 |
+
"memory(GiB)": 46.13,
|
| 904 |
+
"nll_loss": 0.32799023389816284,
|
| 905 |
+
"rewards/accuracies": 0.875,
|
| 906 |
+
"rewards/chosen": 0.05019056051969528,
|
| 907 |
+
"rewards/margins": 1.6790026426315308,
|
| 908 |
+
"rewards/rejected": -1.6288120746612549,
|
| 909 |
+
"step": 50,
|
| 910 |
+
"train_speed(iter/s)": 0.005638
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.3071132856605194,
|
| 914 |
+
"grad_norm": 2.950259208679199,
|
| 915 |
+
"learning_rate": 0.00019430569033830605,
|
| 916 |
+
"logits/chosen": -0.19638891518115997,
|
| 917 |
+
"logits/rejected": -0.3516000807285309,
|
| 918 |
+
"logps/chosen": -7.3598952293396,
|
| 919 |
+
"logps/rejected": -32.14203643798828,
|
| 920 |
+
"loss": 0.9957353472709656,
|
| 921 |
+
"memory(GiB)": 46.13,
|
| 922 |
+
"nll_loss": 0.6411410570144653,
|
| 923 |
+
"rewards/accuracies": 0.796875,
|
| 924 |
+
"rewards/chosen": -0.031813204288482666,
|
| 925 |
+
"rewards/margins": 1.6832952499389648,
|
| 926 |
+
"rewards/rejected": -1.7151083946228027,
|
| 927 |
+
"step": 51,
|
| 928 |
+
"train_speed(iter/s)": 0.005628
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.3131351147911178,
|
| 932 |
+
"grad_norm": 3.4817123413085938,
|
| 933 |
+
"learning_rate": 0.00019396926207859084,
|
| 934 |
+
"logits/chosen": -0.2925015687942505,
|
| 935 |
+
"logits/rejected": -0.4453135132789612,
|
| 936 |
+
"logps/chosen": -6.357159614562988,
|
| 937 |
+
"logps/rejected": -34.39080047607422,
|
| 938 |
+
"loss": 0.9899907112121582,
|
| 939 |
+
"memory(GiB)": 46.13,
|
| 940 |
+
"nll_loss": 0.6184481382369995,
|
| 941 |
+
"rewards/accuracies": 0.828125,
|
| 942 |
+
"rewards/chosen": -0.06884388625621796,
|
| 943 |
+
"rewards/margins": 1.8842136859893799,
|
| 944 |
+
"rewards/rejected": -1.9530572891235352,
|
| 945 |
+
"step": 52,
|
| 946 |
+
"train_speed(iter/s)": 0.005629
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.31915694392171623,
|
| 950 |
+
"grad_norm": 3.4239518642425537,
|
| 951 |
+
"learning_rate": 0.00019362348706397373,
|
| 952 |
+
"logits/chosen": -0.22639772295951843,
|
| 953 |
+
"logits/rejected": -0.4221537709236145,
|
| 954 |
+
"logps/chosen": -5.600902080535889,
|
| 955 |
+
"logps/rejected": -36.40532684326172,
|
| 956 |
+
"loss": 0.8088786602020264,
|
| 957 |
+
"memory(GiB)": 46.13,
|
| 958 |
+
"nll_loss": 0.4771510660648346,
|
| 959 |
+
"rewards/accuracies": 0.859375,
|
| 960 |
+
"rewards/chosen": 0.0569700188934803,
|
| 961 |
+
"rewards/margins": 1.8676366806030273,
|
| 962 |
+
"rewards/rejected": -1.810666561126709,
|
| 963 |
+
"step": 53,
|
| 964 |
+
"train_speed(iter/s)": 0.005628
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.32517877305231463,
|
| 968 |
+
"grad_norm": 3.8685495853424072,
|
| 969 |
+
"learning_rate": 0.00019326839968734279,
|
| 970 |
+
"logits/chosen": -0.2569487690925598,
|
| 971 |
+
"logits/rejected": -0.40343666076660156,
|
| 972 |
+
"logps/chosen": -5.900696277618408,
|
| 973 |
+
"logps/rejected": -36.28150177001953,
|
| 974 |
+
"loss": 0.7994194030761719,
|
| 975 |
+
"memory(GiB)": 46.13,
|
| 976 |
+
"nll_loss": 0.43986907601356506,
|
| 977 |
+
"rewards/accuracies": 0.8125,
|
| 978 |
+
"rewards/chosen": 0.030100831761956215,
|
| 979 |
+
"rewards/margins": 1.9989190101623535,
|
| 980 |
+
"rewards/rejected": -1.968818187713623,
|
| 981 |
+
"step": 54,
|
| 982 |
+
"train_speed(iter/s)": 0.005629
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.3312006021829131,
|
| 986 |
+
"grad_norm": 3.0355138778686523,
|
| 987 |
+
"learning_rate": 0.00019290403526785025,
|
| 988 |
+
"logits/chosen": -0.25347286462783813,
|
| 989 |
+
"logits/rejected": -0.3757534623146057,
|
| 990 |
+
"logps/chosen": -7.442959308624268,
|
| 991 |
+
"logps/rejected": -32.45537567138672,
|
| 992 |
+
"loss": 0.903397798538208,
|
| 993 |
+
"memory(GiB)": 46.13,
|
| 994 |
+
"nll_loss": 0.531738817691803,
|
| 995 |
+
"rewards/accuracies": 0.859375,
|
| 996 |
+
"rewards/chosen": 0.0414905920624733,
|
| 997 |
+
"rewards/margins": 1.6564455032348633,
|
| 998 |
+
"rewards/rejected": -1.614954948425293,
|
| 999 |
+
"step": 55,
|
| 1000 |
+
"train_speed(iter/s)": 0.005629
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.3372224313135115,
|
| 1004 |
+
"grad_norm": 3.739659070968628,
|
| 1005 |
+
"learning_rate": 0.00019253043004739968,
|
| 1006 |
+
"logits/chosen": -0.20877386629581451,
|
| 1007 |
+
"logits/rejected": -0.37541496753692627,
|
| 1008 |
+
"logps/chosen": -7.142071723937988,
|
| 1009 |
+
"logps/rejected": -32.15766906738281,
|
| 1010 |
+
"loss": 0.9619933366775513,
|
| 1011 |
+
"memory(GiB)": 46.13,
|
| 1012 |
+
"nll_loss": 0.4930208921432495,
|
| 1013 |
+
"rewards/accuracies": 0.8125,
|
| 1014 |
+
"rewards/chosen": -0.08227778971195221,
|
| 1015 |
+
"rewards/margins": 1.5348973274230957,
|
| 1016 |
+
"rewards/rejected": -1.6171751022338867,
|
| 1017 |
+
"step": 56,
|
| 1018 |
+
"train_speed(iter/s)": 0.005628
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.3432442604441099,
|
| 1022 |
+
"grad_norm": 2.4167847633361816,
|
| 1023 |
+
"learning_rate": 0.00019214762118704076,
|
| 1024 |
+
"logits/chosen": -0.32553619146347046,
|
| 1025 |
+
"logits/rejected": -0.44049111008644104,
|
| 1026 |
+
"logps/chosen": -6.96627140045166,
|
| 1027 |
+
"logps/rejected": -24.777843475341797,
|
| 1028 |
+
"loss": 0.8523041605949402,
|
| 1029 |
+
"memory(GiB)": 46.13,
|
| 1030 |
+
"nll_loss": 0.44423431158065796,
|
| 1031 |
+
"rewards/accuracies": 0.828125,
|
| 1032 |
+
"rewards/chosen": 0.08760333806276321,
|
| 1033 |
+
"rewards/margins": 1.0987879037857056,
|
| 1034 |
+
"rewards/rejected": -1.011184573173523,
|
| 1035 |
+
"step": 57,
|
| 1036 |
+
"train_speed(iter/s)": 0.005629
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.34926608957470834,
|
| 1040 |
+
"grad_norm": 1.849902629852295,
|
| 1041 |
+
"learning_rate": 0.00019175564676327339,
|
| 1042 |
+
"logits/chosen": -0.29927390813827515,
|
| 1043 |
+
"logits/rejected": -0.4095182418823242,
|
| 1044 |
+
"logps/chosen": -6.824729919433594,
|
| 1045 |
+
"logps/rejected": -31.636598587036133,
|
| 1046 |
+
"loss": 0.827016294002533,
|
| 1047 |
+
"memory(GiB)": 46.13,
|
| 1048 |
+
"nll_loss": 0.432888925075531,
|
| 1049 |
+
"rewards/accuracies": 0.828125,
|
| 1050 |
+
"rewards/chosen": 0.10196143388748169,
|
| 1051 |
+
"rewards/margins": 1.6724345684051514,
|
| 1052 |
+
"rewards/rejected": -1.5704729557037354,
|
| 1053 |
+
"step": 58,
|
| 1054 |
+
"train_speed(iter/s)": 0.005629
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.35528791870530674,
|
| 1058 |
+
"grad_norm": 2.3601560592651367,
|
| 1059 |
+
"learning_rate": 0.0001913545457642601,
|
| 1060 |
+
"logits/chosen": -0.31936806440353394,
|
| 1061 |
+
"logits/rejected": -0.42600923776626587,
|
| 1062 |
+
"logps/chosen": -6.396060943603516,
|
| 1063 |
+
"logps/rejected": -26.946306228637695,
|
| 1064 |
+
"loss": 0.9836419224739075,
|
| 1065 |
+
"memory(GiB)": 46.13,
|
| 1066 |
+
"nll_loss": 0.5654138326644897,
|
| 1067 |
+
"rewards/accuracies": 0.84375,
|
| 1068 |
+
"rewards/chosen": 0.013314278796315193,
|
| 1069 |
+
"rewards/margins": 1.2707021236419678,
|
| 1070 |
+
"rewards/rejected": -1.2573878765106201,
|
| 1071 |
+
"step": 59,
|
| 1072 |
+
"train_speed(iter/s)": 0.005629
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 0.36130974783590514,
|
| 1076 |
+
"grad_norm": 2.7022287845611572,
|
| 1077 |
+
"learning_rate": 0.00019094435808594823,
|
| 1078 |
+
"logits/chosen": -0.3565162420272827,
|
| 1079 |
+
"logits/rejected": -0.48129239678382874,
|
| 1080 |
+
"logps/chosen": -4.523666858673096,
|
| 1081 |
+
"logps/rejected": -30.854751586914062,
|
| 1082 |
+
"loss": 0.7463247776031494,
|
| 1083 |
+
"memory(GiB)": 46.13,
|
| 1084 |
+
"nll_loss": 0.41281038522720337,
|
| 1085 |
+
"rewards/accuracies": 0.84375,
|
| 1086 |
+
"rewards/chosen": 0.06385684013366699,
|
| 1087 |
+
"rewards/margins": 1.8442981243133545,
|
| 1088 |
+
"rewards/rejected": -1.7804412841796875,
|
| 1089 |
+
"step": 60,
|
| 1090 |
+
"train_speed(iter/s)": 0.00563
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.3673315769665036,
|
| 1094 |
+
"grad_norm": 2.549546957015991,
|
| 1095 |
+
"learning_rate": 0.0001905251245281015,
|
| 1096 |
+
"logits/chosen": -0.2870449125766754,
|
| 1097 |
+
"logits/rejected": -0.4699952006340027,
|
| 1098 |
+
"logps/chosen": -3.242504596710205,
|
| 1099 |
+
"logps/rejected": -35.58238220214844,
|
| 1100 |
+
"loss": 0.6698994636535645,
|
| 1101 |
+
"memory(GiB)": 46.13,
|
| 1102 |
+
"nll_loss": 0.3660232126712799,
|
| 1103 |
+
"rewards/accuracies": 0.890625,
|
| 1104 |
+
"rewards/chosen": 0.03870222344994545,
|
| 1105 |
+
"rewards/margins": 1.8160982131958008,
|
| 1106 |
+
"rewards/rejected": -1.7773959636688232,
|
| 1107 |
+
"step": 61,
|
| 1108 |
+
"train_speed(iter/s)": 0.005631
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.373353406097102,
|
| 1112 |
+
"grad_norm": 3.344719409942627,
|
| 1113 |
+
"learning_rate": 0.0001900968867902419,
|
| 1114 |
+
"logits/chosen": -0.26915398240089417,
|
| 1115 |
+
"logits/rejected": -0.38059523701667786,
|
| 1116 |
+
"logps/chosen": -6.996611595153809,
|
| 1117 |
+
"logps/rejected": -36.200714111328125,
|
| 1118 |
+
"loss": 0.8631665110588074,
|
| 1119 |
+
"memory(GiB)": 46.13,
|
| 1120 |
+
"nll_loss": 0.5069284439086914,
|
| 1121 |
+
"rewards/accuracies": 0.828125,
|
| 1122 |
+
"rewards/chosen": -0.0650690495967865,
|
| 1123 |
+
"rewards/margins": 1.7960646152496338,
|
| 1124 |
+
"rewards/rejected": -1.8611338138580322,
|
| 1125 |
+
"step": 62,
|
| 1126 |
+
"train_speed(iter/s)": 0.00563
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 0.3793752352277004,
|
| 1130 |
+
"grad_norm": 1.7888697385787964,
|
| 1131 |
+
"learning_rate": 0.0001896596874675021,
|
| 1132 |
+
"logits/chosen": -0.25557130575180054,
|
| 1133 |
+
"logits/rejected": -0.4095707833766937,
|
| 1134 |
+
"logps/chosen": -5.574047088623047,
|
| 1135 |
+
"logps/rejected": -31.464292526245117,
|
| 1136 |
+
"loss": 0.6977128386497498,
|
| 1137 |
+
"memory(GiB)": 46.13,
|
| 1138 |
+
"nll_loss": 0.3594370484352112,
|
| 1139 |
+
"rewards/accuracies": 0.84375,
|
| 1140 |
+
"rewards/chosen": 0.08366499841213226,
|
| 1141 |
+
"rewards/margins": 1.6534584760665894,
|
| 1142 |
+
"rewards/rejected": -1.569793462753296,
|
| 1143 |
+
"step": 63,
|
| 1144 |
+
"train_speed(iter/s)": 0.00563
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.38539706435829885,
|
| 1148 |
+
"grad_norm": 2.8035733699798584,
|
| 1149 |
+
"learning_rate": 0.00018921357004638835,
|
| 1150 |
+
"logits/chosen": -0.22736577689647675,
|
| 1151 |
+
"logits/rejected": -0.3719921410083771,
|
| 1152 |
+
"logps/chosen": -6.317353248596191,
|
| 1153 |
+
"logps/rejected": -30.95716094970703,
|
| 1154 |
+
"loss": 0.8575116991996765,
|
| 1155 |
+
"memory(GiB)": 46.13,
|
| 1156 |
+
"nll_loss": 0.5229077339172363,
|
| 1157 |
+
"rewards/accuracies": 0.875,
|
| 1158 |
+
"rewards/chosen": 0.13013550639152527,
|
| 1159 |
+
"rewards/margins": 1.687712550163269,
|
| 1160 |
+
"rewards/rejected": -1.5575770139694214,
|
| 1161 |
+
"step": 64,
|
| 1162 |
+
"train_speed(iter/s)": 0.00563
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.39141889348889725,
|
| 1166 |
+
"grad_norm": 2.560202121734619,
|
| 1167 |
+
"learning_rate": 0.00018875857890045543,
|
| 1168 |
+
"logits/chosen": -0.24293088912963867,
|
| 1169 |
+
"logits/rejected": -0.35710442066192627,
|
| 1170 |
+
"logps/chosen": -7.503670692443848,
|
| 1171 |
+
"logps/rejected": -32.99870300292969,
|
| 1172 |
+
"loss": 0.8683996796607971,
|
| 1173 |
+
"memory(GiB)": 46.13,
|
| 1174 |
+
"nll_loss": 0.49205920100212097,
|
| 1175 |
+
"rewards/accuracies": 0.796875,
|
| 1176 |
+
"rewards/chosen": 0.03412435203790665,
|
| 1177 |
+
"rewards/margins": 1.4795939922332764,
|
| 1178 |
+
"rewards/rejected": -1.4454697370529175,
|
| 1179 |
+
"step": 65,
|
| 1180 |
+
"train_speed(iter/s)": 0.005629
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.39744072261949565,
|
| 1184 |
+
"grad_norm": 1.993003487586975,
|
| 1185 |
+
"learning_rate": 0.00018829475928589271,
|
| 1186 |
+
"logits/chosen": -0.23541702330112457,
|
| 1187 |
+
"logits/rejected": -0.3272259533405304,
|
| 1188 |
+
"logps/chosen": -5.38151216506958,
|
| 1189 |
+
"logps/rejected": -29.121570587158203,
|
| 1190 |
+
"loss": 0.7320932149887085,
|
| 1191 |
+
"memory(GiB)": 46.13,
|
| 1192 |
+
"nll_loss": 0.36119604110717773,
|
| 1193 |
+
"rewards/accuracies": 0.828125,
|
| 1194 |
+
"rewards/chosen": 0.0803036242723465,
|
| 1195 |
+
"rewards/margins": 1.6109750270843506,
|
| 1196 |
+
"rewards/rejected": -1.53067147731781,
|
| 1197 |
+
"step": 66,
|
| 1198 |
+
"train_speed(iter/s)": 0.00563
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.4034625517500941,
|
| 1202 |
+
"grad_norm": 2.9964606761932373,
|
| 1203 |
+
"learning_rate": 0.00018782215733702286,
|
| 1204 |
+
"logits/chosen": -0.20383372902870178,
|
| 1205 |
+
"logits/rejected": -0.34373044967651367,
|
| 1206 |
+
"logps/chosen": -6.649240493774414,
|
| 1207 |
+
"logps/rejected": -31.494646072387695,
|
| 1208 |
+
"loss": 0.8884384036064148,
|
| 1209 |
+
"memory(GiB)": 46.13,
|
| 1210 |
+
"nll_loss": 0.5308991074562073,
|
| 1211 |
+
"rewards/accuracies": 0.828125,
|
| 1212 |
+
"rewards/chosen": 0.10617589205503464,
|
| 1213 |
+
"rewards/margins": 1.5828511714935303,
|
| 1214 |
+
"rewards/rejected": -1.4766751527786255,
|
| 1215 |
+
"step": 67,
|
| 1216 |
+
"train_speed(iter/s)": 0.00563
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.4094843808806925,
|
| 1220 |
+
"grad_norm": 2.774994134902954,
|
| 1221 |
+
"learning_rate": 0.00018734082006171299,
|
| 1222 |
+
"logits/chosen": -0.25100839138031006,
|
| 1223 |
+
"logits/rejected": -0.4020751118659973,
|
| 1224 |
+
"logps/chosen": -7.144831657409668,
|
| 1225 |
+
"logps/rejected": -29.9566707611084,
|
| 1226 |
+
"loss": 0.8242477178573608,
|
| 1227 |
+
"memory(GiB)": 46.13,
|
| 1228 |
+
"nll_loss": 0.5139617919921875,
|
| 1229 |
+
"rewards/accuracies": 0.90625,
|
| 1230 |
+
"rewards/chosen": 0.13794061541557312,
|
| 1231 |
+
"rewards/margins": 1.6364552974700928,
|
| 1232 |
+
"rewards/rejected": -1.4985146522521973,
|
| 1233 |
+
"step": 68,
|
| 1234 |
+
"train_speed(iter/s)": 0.005631
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.41550621001129096,
|
| 1238 |
+
"grad_norm": 3.4358551502227783,
|
| 1239 |
+
"learning_rate": 0.0001868507953366989,
|
| 1240 |
+
"logits/chosen": -0.24417072534561157,
|
| 1241 |
+
"logits/rejected": -0.36953434348106384,
|
| 1242 |
+
"logps/chosen": -6.3338799476623535,
|
| 1243 |
+
"logps/rejected": -32.21156311035156,
|
| 1244 |
+
"loss": 0.7758980989456177,
|
| 1245 |
+
"memory(GiB)": 46.13,
|
| 1246 |
+
"nll_loss": 0.48221129179000854,
|
| 1247 |
+
"rewards/accuracies": 0.90625,
|
| 1248 |
+
"rewards/chosen": 0.14049482345581055,
|
| 1249 |
+
"rewards/margins": 1.806667447090149,
|
| 1250 |
+
"rewards/rejected": -1.666172742843628,
|
| 1251 |
+
"step": 69,
|
| 1252 |
+
"train_speed(iter/s)": 0.005631
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.42152803914188935,
|
| 1256 |
+
"grad_norm": 2.821110248565674,
|
| 1257 |
+
"learning_rate": 0.0001863521319028231,
|
| 1258 |
+
"logits/chosen": -0.2586533725261688,
|
| 1259 |
+
"logits/rejected": -0.36297786235809326,
|
| 1260 |
+
"logps/chosen": -7.11786413192749,
|
| 1261 |
+
"logps/rejected": -31.115772247314453,
|
| 1262 |
+
"loss": 0.8807559013366699,
|
| 1263 |
+
"memory(GiB)": 46.13,
|
| 1264 |
+
"nll_loss": 0.5770330429077148,
|
| 1265 |
+
"rewards/accuracies": 0.828125,
|
| 1266 |
+
"rewards/chosen": 0.07292432337999344,
|
| 1267 |
+
"rewards/margins": 2.026425838470459,
|
| 1268 |
+
"rewards/rejected": -1.9535014629364014,
|
| 1269 |
+
"step": 70,
|
| 1270 |
+
"train_speed(iter/s)": 0.005632
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.42754986827248775,
|
| 1274 |
+
"grad_norm": 2.6153159141540527,
|
| 1275 |
+
"learning_rate": 0.00018584487936018661,
|
| 1276 |
+
"logits/chosen": -0.305365651845932,
|
| 1277 |
+
"logits/rejected": -0.4076971709728241,
|
| 1278 |
+
"logps/chosen": -4.401376247406006,
|
| 1279 |
+
"logps/rejected": -25.76744842529297,
|
| 1280 |
+
"loss": 0.7574676275253296,
|
| 1281 |
+
"memory(GiB)": 46.13,
|
| 1282 |
+
"nll_loss": 0.3759956955909729,
|
| 1283 |
+
"rewards/accuracies": 0.796875,
|
| 1284 |
+
"rewards/chosen": -0.015199379995465279,
|
| 1285 |
+
"rewards/margins": 1.5401445627212524,
|
| 1286 |
+
"rewards/rejected": -1.5553438663482666,
|
| 1287 |
+
"step": 71,
|
| 1288 |
+
"train_speed(iter/s)": 0.005632
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 0.4335716974030862,
|
| 1292 |
+
"grad_norm": 4.708693027496338,
|
| 1293 |
+
"learning_rate": 0.00018532908816321558,
|
| 1294 |
+
"logits/chosen": -0.22257624566555023,
|
| 1295 |
+
"logits/rejected": -0.36776047945022583,
|
| 1296 |
+
"logps/chosen": -5.041079998016357,
|
| 1297 |
+
"logps/rejected": -31.822460174560547,
|
| 1298 |
+
"loss": 0.7604058980941772,
|
| 1299 |
+
"memory(GiB)": 46.13,
|
| 1300 |
+
"nll_loss": 0.3996947705745697,
|
| 1301 |
+
"rewards/accuracies": 0.84375,
|
| 1302 |
+
"rewards/chosen": 0.0631202757358551,
|
| 1303 |
+
"rewards/margins": 1.8200846910476685,
|
| 1304 |
+
"rewards/rejected": -1.7569644451141357,
|
| 1305 |
+
"step": 72,
|
| 1306 |
+
"train_speed(iter/s)": 0.005633
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.4395935265336846,
|
| 1310 |
+
"grad_norm": 4.16855001449585,
|
| 1311 |
+
"learning_rate": 0.0001848048096156426,
|
| 1312 |
+
"logits/chosen": -0.1833086460828781,
|
| 1313 |
+
"logits/rejected": -0.40944039821624756,
|
| 1314 |
+
"logps/chosen": -7.540615081787109,
|
| 1315 |
+
"logps/rejected": -41.107208251953125,
|
| 1316 |
+
"loss": 1.1337863206863403,
|
| 1317 |
+
"memory(GiB)": 46.13,
|
| 1318 |
+
"nll_loss": 0.7324954271316528,
|
| 1319 |
+
"rewards/accuracies": 0.828125,
|
| 1320 |
+
"rewards/chosen": -0.13253508508205414,
|
| 1321 |
+
"rewards/margins": 2.3366596698760986,
|
| 1322 |
+
"rewards/rejected": -2.4691946506500244,
|
| 1323 |
+
"step": 73,
|
| 1324 |
+
"train_speed(iter/s)": 0.005633
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.445615355664283,
|
| 1328 |
+
"grad_norm": 2.806788444519043,
|
| 1329 |
+
"learning_rate": 0.0001842720958654039,
|
| 1330 |
+
"logits/chosen": -0.3181215524673462,
|
| 1331 |
+
"logits/rejected": -0.39911144971847534,
|
| 1332 |
+
"logps/chosen": -5.607314586639404,
|
| 1333 |
+
"logps/rejected": -28.428564071655273,
|
| 1334 |
+
"loss": 0.6960507035255432,
|
| 1335 |
+
"memory(GiB)": 46.13,
|
| 1336 |
+
"nll_loss": 0.40211787819862366,
|
| 1337 |
+
"rewards/accuracies": 0.9375,
|
| 1338 |
+
"rewards/chosen": 0.15752148628234863,
|
| 1339 |
+
"rewards/margins": 1.9513351917266846,
|
| 1340 |
+
"rewards/rejected": -1.7938138246536255,
|
| 1341 |
+
"step": 74,
|
| 1342 |
+
"train_speed(iter/s)": 0.005634
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.45163718479488146,
|
| 1346 |
+
"grad_norm": 3.420902967453003,
|
| 1347 |
+
"learning_rate": 0.00018373099989945236,
|
| 1348 |
+
"logits/chosen": -0.22940213978290558,
|
| 1349 |
+
"logits/rejected": -0.4191284477710724,
|
| 1350 |
+
"logps/chosen": -6.713252067565918,
|
| 1351 |
+
"logps/rejected": -37.6962890625,
|
| 1352 |
+
"loss": 0.920997679233551,
|
| 1353 |
+
"memory(GiB)": 46.13,
|
| 1354 |
+
"nll_loss": 0.5668741464614868,
|
| 1355 |
+
"rewards/accuracies": 0.890625,
|
| 1356 |
+
"rewards/chosen": -0.0326458215713501,
|
| 1357 |
+
"rewards/margins": 2.1453044414520264,
|
| 1358 |
+
"rewards/rejected": -2.177950382232666,
|
| 1359 |
+
"step": 75,
|
| 1360 |
+
"train_speed(iter/s)": 0.005634
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.45765901392547986,
|
| 1364 |
+
"grad_norm": 3.6164190769195557,
|
| 1365 |
+
"learning_rate": 0.0001831815755384869,
|
| 1366 |
+
"logits/chosen": -0.20445629954338074,
|
| 1367 |
+
"logits/rejected": -0.3627314567565918,
|
| 1368 |
+
"logps/chosen": -5.958117485046387,
|
| 1369 |
+
"logps/rejected": -38.63805389404297,
|
| 1370 |
+
"loss": 0.884059727191925,
|
| 1371 |
+
"memory(GiB)": 46.13,
|
| 1372 |
+
"nll_loss": 0.5545799732208252,
|
| 1373 |
+
"rewards/accuracies": 0.828125,
|
| 1374 |
+
"rewards/chosen": -0.049575597047805786,
|
| 1375 |
+
"rewards/margins": 2.2317745685577393,
|
| 1376 |
+
"rewards/rejected": -2.2813501358032227,
|
| 1377 |
+
"step": 76,
|
| 1378 |
+
"train_speed(iter/s)": 0.005634
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.46368084305607826,
|
| 1382 |
+
"grad_norm": 2.9762485027313232,
|
| 1383 |
+
"learning_rate": 0.0001826238774315995,
|
| 1384 |
+
"logits/chosen": -0.23990976810455322,
|
| 1385 |
+
"logits/rejected": -0.38009804487228394,
|
| 1386 |
+
"logps/chosen": -5.436191082000732,
|
| 1387 |
+
"logps/rejected": -34.6729850769043,
|
| 1388 |
+
"loss": 0.7682054042816162,
|
| 1389 |
+
"memory(GiB)": 46.13,
|
| 1390 |
+
"nll_loss": 0.4302901029586792,
|
| 1391 |
+
"rewards/accuracies": 0.859375,
|
| 1392 |
+
"rewards/chosen": 0.1761879026889801,
|
| 1393 |
+
"rewards/margins": 2.175489902496338,
|
| 1394 |
+
"rewards/rejected": -1.9993020296096802,
|
| 1395 |
+
"step": 77,
|
| 1396 |
+
"train_speed(iter/s)": 0.005634
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.4697026721866767,
|
| 1400 |
+
"grad_norm": 2.6883339881896973,
|
| 1401 |
+
"learning_rate": 0.00018205796105083915,
|
| 1402 |
+
"logits/chosen": -0.27416834235191345,
|
| 1403 |
+
"logits/rejected": -0.387048602104187,
|
| 1404 |
+
"logps/chosen": -6.189967155456543,
|
| 1405 |
+
"logps/rejected": -30.830598831176758,
|
| 1406 |
+
"loss": 0.8499253988265991,
|
| 1407 |
+
"memory(GiB)": 46.13,
|
| 1408 |
+
"nll_loss": 0.46602901816368103,
|
| 1409 |
+
"rewards/accuracies": 0.859375,
|
| 1410 |
+
"rewards/chosen": 0.06820189207792282,
|
| 1411 |
+
"rewards/margins": 1.8341140747070312,
|
| 1412 |
+
"rewards/rejected": -1.7659121751785278,
|
| 1413 |
+
"step": 78,
|
| 1414 |
+
"train_speed(iter/s)": 0.005635
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.4757245013172751,
|
| 1418 |
+
"grad_norm": 2.6955106258392334,
|
| 1419 |
+
"learning_rate": 0.00018148388268569453,
|
| 1420 |
+
"logits/chosen": -0.22744283080101013,
|
| 1421 |
+
"logits/rejected": -0.35057857632637024,
|
| 1422 |
+
"logps/chosen": -4.485665798187256,
|
| 1423 |
+
"logps/rejected": -29.74733543395996,
|
| 1424 |
+
"loss": 0.7549667954444885,
|
| 1425 |
+
"memory(GiB)": 46.13,
|
| 1426 |
+
"nll_loss": 0.4278162717819214,
|
| 1427 |
+
"rewards/accuracies": 0.828125,
|
| 1428 |
+
"rewards/chosen": 0.20732006430625916,
|
| 1429 |
+
"rewards/margins": 2.0227808952331543,
|
| 1430 |
+
"rewards/rejected": -1.8154606819152832,
|
| 1431 |
+
"step": 79,
|
| 1432 |
+
"train_speed(iter/s)": 0.005635
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 0.4817463304478735,
|
| 1436 |
+
"grad_norm": 2.907402515411377,
|
| 1437 |
+
"learning_rate": 0.00018090169943749476,
|
| 1438 |
+
"logits/chosen": -0.13104476034641266,
|
| 1439 |
+
"logits/rejected": -0.36177706718444824,
|
| 1440 |
+
"logps/chosen": -3.6902096271514893,
|
| 1441 |
+
"logps/rejected": -34.731773376464844,
|
| 1442 |
+
"loss": 0.6847038269042969,
|
| 1443 |
+
"memory(GiB)": 46.13,
|
| 1444 |
+
"nll_loss": 0.34349215030670166,
|
| 1445 |
+
"rewards/accuracies": 0.890625,
|
| 1446 |
+
"rewards/chosen": 0.11521060019731522,
|
| 1447 |
+
"rewards/margins": 1.8666913509368896,
|
| 1448 |
+
"rewards/rejected": -1.7514808177947998,
|
| 1449 |
+
"step": 80,
|
| 1450 |
+
"train_speed(iter/s)": 0.005635
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.48776815957847197,
|
| 1454 |
+
"grad_norm": 2.823246955871582,
|
| 1455 |
+
"learning_rate": 0.00018031146921373018,
|
| 1456 |
+
"logits/chosen": -0.22965751588344574,
|
| 1457 |
+
"logits/rejected": -0.3291424512863159,
|
| 1458 |
+
"logps/chosen": -5.761353969573975,
|
| 1459 |
+
"logps/rejected": -36.051082611083984,
|
| 1460 |
+
"loss": 0.7157370448112488,
|
| 1461 |
+
"memory(GiB)": 46.13,
|
| 1462 |
+
"nll_loss": 0.4026964008808136,
|
| 1463 |
+
"rewards/accuracies": 0.859375,
|
| 1464 |
+
"rewards/chosen": 0.07981520146131516,
|
| 1465 |
+
"rewards/margins": 2.387995958328247,
|
| 1466 |
+
"rewards/rejected": -2.308180809020996,
|
| 1467 |
+
"step": 81,
|
| 1468 |
+
"train_speed(iter/s)": 0.005636
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 0.49378998870907037,
|
| 1472 |
+
"grad_norm": 2.73394513130188,
|
| 1473 |
+
"learning_rate": 0.00017971325072229226,
|
| 1474 |
+
"logits/chosen": -0.19317063689231873,
|
| 1475 |
+
"logits/rejected": -0.40784093737602234,
|
| 1476 |
+
"logps/chosen": -5.166616916656494,
|
| 1477 |
+
"logps/rejected": -40.66303634643555,
|
| 1478 |
+
"loss": 0.6980624198913574,
|
| 1479 |
+
"memory(GiB)": 46.13,
|
| 1480 |
+
"nll_loss": 0.4219977855682373,
|
| 1481 |
+
"rewards/accuracies": 0.875,
|
| 1482 |
+
"rewards/chosen": 0.030578728765249252,
|
| 1483 |
+
"rewards/margins": 2.3669514656066895,
|
| 1484 |
+
"rewards/rejected": -2.3363726139068604,
|
| 1485 |
+
"step": 82,
|
| 1486 |
+
"train_speed(iter/s)": 0.005635
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.4998118178396688,
|
| 1490 |
+
"grad_norm": 5.509173393249512,
|
| 1491 |
+
"learning_rate": 0.00017910710346563416,
|
| 1492 |
+
"logits/chosen": -0.1950196623802185,
|
| 1493 |
+
"logits/rejected": -0.3517180383205414,
|
| 1494 |
+
"logps/chosen": -8.197553634643555,
|
| 1495 |
+
"logps/rejected": -42.4526252746582,
|
| 1496 |
+
"loss": 0.9209753274917603,
|
| 1497 |
+
"memory(GiB)": 46.13,
|
| 1498 |
+
"nll_loss": 0.608447790145874,
|
| 1499 |
+
"rewards/accuracies": 0.859375,
|
| 1500 |
+
"rewards/chosen": 0.012901969254016876,
|
| 1501 |
+
"rewards/margins": 2.6168861389160156,
|
| 1502 |
+
"rewards/rejected": -2.6039838790893555,
|
| 1503 |
+
"step": 83,
|
| 1504 |
+
"train_speed(iter/s)": 0.005636
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 0.5058336469702672,
|
| 1508 |
+
"grad_norm": 3.090348482131958,
|
| 1509 |
+
"learning_rate": 0.00017849308773485226,
|
| 1510 |
+
"logits/chosen": -0.2090476155281067,
|
| 1511 |
+
"logits/rejected": -0.3706829845905304,
|
| 1512 |
+
"logps/chosen": -10.569226264953613,
|
| 1513 |
+
"logps/rejected": -43.29594421386719,
|
| 1514 |
+
"loss": 1.0815891027450562,
|
| 1515 |
+
"memory(GiB)": 46.13,
|
| 1516 |
+
"nll_loss": 0.7449980974197388,
|
| 1517 |
+
"rewards/accuracies": 0.796875,
|
| 1518 |
+
"rewards/chosen": -0.03622851148247719,
|
| 1519 |
+
"rewards/margins": 2.7713966369628906,
|
| 1520 |
+
"rewards/rejected": -2.8076250553131104,
|
| 1521 |
+
"step": 84,
|
| 1522 |
+
"train_speed(iter/s)": 0.005636
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.5118554761008657,
|
| 1526 |
+
"grad_norm": 3.75370717048645,
|
| 1527 |
+
"learning_rate": 0.0001778712646036894,
|
| 1528 |
+
"logits/chosen": -0.2378547489643097,
|
| 1529 |
+
"logits/rejected": -0.3873806297779083,
|
| 1530 |
+
"logps/chosen": -8.168487548828125,
|
| 1531 |
+
"logps/rejected": -37.17033386230469,
|
| 1532 |
+
"loss": 1.0251762866973877,
|
| 1533 |
+
"memory(GiB)": 46.13,
|
| 1534 |
+
"nll_loss": 0.6340765953063965,
|
| 1535 |
+
"rewards/accuracies": 0.765625,
|
| 1536 |
+
"rewards/chosen": -0.11355286091566086,
|
| 1537 |
+
"rewards/margins": 2.2919938564300537,
|
| 1538 |
+
"rewards/rejected": -2.4055469036102295,
|
| 1539 |
+
"step": 85,
|
| 1540 |
+
"train_speed(iter/s)": 0.005636
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.517877305231464,
|
| 1544 |
+
"grad_norm": 3.9649507999420166,
|
| 1545 |
+
"learning_rate": 0.00017724169592245995,
|
| 1546 |
+
"logits/chosen": -0.20498239994049072,
|
| 1547 |
+
"logits/rejected": -0.369476854801178,
|
| 1548 |
+
"logps/chosen": -6.779036521911621,
|
| 1549 |
+
"logps/rejected": -37.98490905761719,
|
| 1550 |
+
"loss": 0.797798216342926,
|
| 1551 |
+
"memory(GiB)": 46.13,
|
| 1552 |
+
"nll_loss": 0.5137072801589966,
|
| 1553 |
+
"rewards/accuracies": 0.890625,
|
| 1554 |
+
"rewards/chosen": 0.07833351194858551,
|
| 1555 |
+
"rewards/margins": 2.144557476043701,
|
| 1556 |
+
"rewards/rejected": -2.0662238597869873,
|
| 1557 |
+
"step": 86,
|
| 1558 |
+
"train_speed(iter/s)": 0.005636
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.5238991343620625,
|
| 1562 |
+
"grad_norm": 3.4646809101104736,
|
| 1563 |
+
"learning_rate": 0.0001766044443118978,
|
| 1564 |
+
"logits/chosen": -0.22029685974121094,
|
| 1565 |
+
"logits/rejected": -0.33659541606903076,
|
| 1566 |
+
"logps/chosen": -5.358316421508789,
|
| 1567 |
+
"logps/rejected": -24.96926498413086,
|
| 1568 |
+
"loss": 0.809937596321106,
|
| 1569 |
+
"memory(GiB)": 46.13,
|
| 1570 |
+
"nll_loss": 0.3953189551830292,
|
| 1571 |
+
"rewards/accuracies": 0.796875,
|
| 1572 |
+
"rewards/chosen": 0.08655387908220291,
|
| 1573 |
+
"rewards/margins": 1.2862589359283447,
|
| 1574 |
+
"rewards/rejected": -1.1997051239013672,
|
| 1575 |
+
"step": 87,
|
| 1576 |
+
"train_speed(iter/s)": 0.005636
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 0.5299209634926609,
|
| 1580 |
+
"grad_norm": 2.3174376487731934,
|
| 1581 |
+
"learning_rate": 0.00017595957315692782,
|
| 1582 |
+
"logits/chosen": -0.23393434286117554,
|
| 1583 |
+
"logits/rejected": -0.37657639384269714,
|
| 1584 |
+
"logps/chosen": -5.694438934326172,
|
| 1585 |
+
"logps/rejected": -26.54452896118164,
|
| 1586 |
+
"loss": 0.8167567253112793,
|
| 1587 |
+
"memory(GiB)": 46.13,
|
| 1588 |
+
"nll_loss": 0.46906933188438416,
|
| 1589 |
+
"rewards/accuracies": 0.890625,
|
| 1590 |
+
"rewards/chosen": 0.12023281306028366,
|
| 1591 |
+
"rewards/margins": 1.344594120979309,
|
| 1592 |
+
"rewards/rejected": -1.2243614196777344,
|
| 1593 |
+
"step": 88,
|
| 1594 |
+
"train_speed(iter/s)": 0.005636
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.5359427926232593,
|
| 1598 |
+
"grad_norm": 2.29378342628479,
|
| 1599 |
+
"learning_rate": 0.00017530714660036112,
|
| 1600 |
+
"logits/chosen": -0.23060575127601624,
|
| 1601 |
+
"logits/rejected": -0.34717997908592224,
|
| 1602 |
+
"logps/chosen": -8.571242332458496,
|
| 1603 |
+
"logps/rejected": -27.7784423828125,
|
| 1604 |
+
"loss": 0.9238011837005615,
|
| 1605 |
+
"memory(GiB)": 46.13,
|
| 1606 |
+
"nll_loss": 0.5501088500022888,
|
| 1607 |
+
"rewards/accuracies": 0.8125,
|
| 1608 |
+
"rewards/chosen": 0.21566246449947357,
|
| 1609 |
+
"rewards/margins": 1.4910237789154053,
|
| 1610 |
+
"rewards/rejected": -1.2753612995147705,
|
| 1611 |
+
"step": 89,
|
| 1612 |
+
"train_speed(iter/s)": 0.005636
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.5419646217538577,
|
| 1616 |
+
"grad_norm": 4.33613395690918,
|
| 1617 |
+
"learning_rate": 0.00017464722953651504,
|
| 1618 |
+
"logits/chosen": -0.17152062058448792,
|
| 1619 |
+
"logits/rejected": -0.3561325669288635,
|
| 1620 |
+
"logps/chosen": -6.058459281921387,
|
| 1621 |
+
"logps/rejected": -32.55611801147461,
|
| 1622 |
+
"loss": 0.8752692341804504,
|
| 1623 |
+
"memory(GiB)": 46.13,
|
| 1624 |
+
"nll_loss": 0.5144660472869873,
|
| 1625 |
+
"rewards/accuracies": 0.84375,
|
| 1626 |
+
"rewards/chosen": 0.0842672809958458,
|
| 1627 |
+
"rewards/margins": 1.5552325248718262,
|
| 1628 |
+
"rewards/rejected": -1.4709652662277222,
|
| 1629 |
+
"step": 90,
|
| 1630 |
+
"train_speed(iter/s)": 0.005636
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 0.5479864508844562,
|
| 1634 |
+
"grad_norm": 2.2437572479248047,
|
| 1635 |
+
"learning_rate": 0.0001739798876047584,
|
| 1636 |
+
"logits/chosen": -0.23511910438537598,
|
| 1637 |
+
"logits/rejected": -0.3851989507675171,
|
| 1638 |
+
"logps/chosen": -4.1521100997924805,
|
| 1639 |
+
"logps/rejected": -28.886817932128906,
|
| 1640 |
+
"loss": 0.7264631390571594,
|
| 1641 |
+
"memory(GiB)": 46.13,
|
| 1642 |
+
"nll_loss": 0.36743801832199097,
|
| 1643 |
+
"rewards/accuracies": 0.859375,
|
| 1644 |
+
"rewards/chosen": 0.10910709202289581,
|
| 1645 |
+
"rewards/margins": 1.5735148191452026,
|
| 1646 |
+
"rewards/rejected": -1.4644078016281128,
|
| 1647 |
+
"step": 91,
|
| 1648 |
+
"train_speed(iter/s)": 0.005636
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.5540082800150545,
|
| 1652 |
+
"grad_norm": 2.482116460800171,
|
| 1653 |
+
"learning_rate": 0.00017330518718298264,
|
| 1654 |
+
"logits/chosen": -0.19554060697555542,
|
| 1655 |
+
"logits/rejected": -0.38464802503585815,
|
| 1656 |
+
"logps/chosen": -5.105670928955078,
|
| 1657 |
+
"logps/rejected": -42.76155471801758,
|
| 1658 |
+
"loss": 0.6233892440795898,
|
| 1659 |
+
"memory(GiB)": 46.13,
|
| 1660 |
+
"nll_loss": 0.36078184843063354,
|
| 1661 |
+
"rewards/accuracies": 0.9375,
|
| 1662 |
+
"rewards/chosen": 0.06755763292312622,
|
| 1663 |
+
"rewards/margins": 2.4900288581848145,
|
| 1664 |
+
"rewards/rejected": -2.422471523284912,
|
| 1665 |
+
"step": 92,
|
| 1666 |
+
"train_speed(iter/s)": 0.005635
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.560030109145653,
|
| 1670 |
+
"grad_norm": 3.629793167114258,
|
| 1671 |
+
"learning_rate": 0.0001726231953809993,
|
| 1672 |
+
"logits/chosen": -0.23075315356254578,
|
| 1673 |
+
"logits/rejected": -0.4017852246761322,
|
| 1674 |
+
"logps/chosen": -7.799711227416992,
|
| 1675 |
+
"logps/rejected": -38.58491516113281,
|
| 1676 |
+
"loss": 0.7807697057723999,
|
| 1677 |
+
"memory(GiB)": 46.13,
|
| 1678 |
+
"nll_loss": 0.5262271165847778,
|
| 1679 |
+
"rewards/accuracies": 0.875,
|
| 1680 |
+
"rewards/chosen": 0.15829423069953918,
|
| 1681 |
+
"rewards/margins": 2.4986987113952637,
|
| 1682 |
+
"rewards/rejected": -2.340404510498047,
|
| 1683 |
+
"step": 93,
|
| 1684 |
+
"train_speed(iter/s)": 0.005635
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.5660519382762514,
|
| 1688 |
+
"grad_norm": 10.366227149963379,
|
| 1689 |
+
"learning_rate": 0.0001719339800338651,
|
| 1690 |
+
"logits/chosen": -0.29187849164009094,
|
| 1691 |
+
"logits/rejected": -0.42478230595588684,
|
| 1692 |
+
"logps/chosen": -7.801548004150391,
|
| 1693 |
+
"logps/rejected": -34.29828643798828,
|
| 1694 |
+
"loss": 1.1007416248321533,
|
| 1695 |
+
"memory(GiB)": 46.13,
|
| 1696 |
+
"nll_loss": 0.7008671164512634,
|
| 1697 |
+
"rewards/accuracies": 0.828125,
|
| 1698 |
+
"rewards/chosen": -0.0017978232353925705,
|
| 1699 |
+
"rewards/margins": 1.9699221849441528,
|
| 1700 |
+
"rewards/rejected": -1.9717202186584473,
|
| 1701 |
+
"step": 94,
|
| 1702 |
+
"train_speed(iter/s)": 0.005635
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.5720737674068498,
|
| 1706 |
+
"grad_norm": 4.7859039306640625,
|
| 1707 |
+
"learning_rate": 0.0001712376096951345,
|
| 1708 |
+
"logits/chosen": -0.2881244421005249,
|
| 1709 |
+
"logits/rejected": -0.39795249700546265,
|
| 1710 |
+
"logps/chosen": -6.9338459968566895,
|
| 1711 |
+
"logps/rejected": -34.682186126708984,
|
| 1712 |
+
"loss": 0.8693129420280457,
|
| 1713 |
+
"memory(GiB)": 46.13,
|
| 1714 |
+
"nll_loss": 0.5744621157646179,
|
| 1715 |
+
"rewards/accuracies": 0.90625,
|
| 1716 |
+
"rewards/chosen": 0.12829959392547607,
|
| 1717 |
+
"rewards/margins": 2.2821671962738037,
|
| 1718 |
+
"rewards/rejected": -2.153867244720459,
|
| 1719 |
+
"step": 95,
|
| 1720 |
+
"train_speed(iter/s)": 0.005636
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 0.5780955965374482,
|
| 1724 |
+
"grad_norm": 2.521221160888672,
|
| 1725 |
+
"learning_rate": 0.0001705341536300409,
|
| 1726 |
+
"logits/chosen": -0.2661621868610382,
|
| 1727 |
+
"logits/rejected": -0.3923559784889221,
|
| 1728 |
+
"logps/chosen": -4.780788898468018,
|
| 1729 |
+
"logps/rejected": -28.810710906982422,
|
| 1730 |
+
"loss": 0.7346460819244385,
|
| 1731 |
+
"memory(GiB)": 46.13,
|
| 1732 |
+
"nll_loss": 0.40230822563171387,
|
| 1733 |
+
"rewards/accuracies": 0.796875,
|
| 1734 |
+
"rewards/chosen": 0.10683214664459229,
|
| 1735 |
+
"rewards/margins": 1.7658103704452515,
|
| 1736 |
+
"rewards/rejected": -1.6589782238006592,
|
| 1737 |
+
"step": 96,
|
| 1738 |
+
"train_speed(iter/s)": 0.005636
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.5841174256680467,
|
| 1742 |
+
"grad_norm": 5.023887634277344,
|
| 1743 |
+
"learning_rate": 0.00016982368180860728,
|
| 1744 |
+
"logits/chosen": -0.25105586647987366,
|
| 1745 |
+
"logits/rejected": -0.3866545855998993,
|
| 1746 |
+
"logps/chosen": -7.320672512054443,
|
| 1747 |
+
"logps/rejected": -34.95018005371094,
|
| 1748 |
+
"loss": 0.9987191557884216,
|
| 1749 |
+
"memory(GiB)": 46.13,
|
| 1750 |
+
"nll_loss": 0.6328321695327759,
|
| 1751 |
+
"rewards/accuracies": 0.8125,
|
| 1752 |
+
"rewards/chosen": -0.07585513591766357,
|
| 1753 |
+
"rewards/margins": 2.0516769886016846,
|
| 1754 |
+
"rewards/rejected": -2.127532482147217,
|
| 1755 |
+
"step": 97,
|
| 1756 |
+
"train_speed(iter/s)": 0.005636
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.590139254798645,
|
| 1760 |
+
"grad_norm": 1.8328591585159302,
|
| 1761 |
+
"learning_rate": 0.00016910626489868649,
|
| 1762 |
+
"logits/chosen": -0.2209641933441162,
|
| 1763 |
+
"logits/rejected": -0.398262083530426,
|
| 1764 |
+
"logps/chosen": -4.149725437164307,
|
| 1765 |
+
"logps/rejected": -35.193931579589844,
|
| 1766 |
+
"loss": 0.6401265263557434,
|
| 1767 |
+
"memory(GiB)": 46.13,
|
| 1768 |
+
"nll_loss": 0.31758564710617065,
|
| 1769 |
+
"rewards/accuracies": 0.84375,
|
| 1770 |
+
"rewards/chosen": 0.14480024576187134,
|
| 1771 |
+
"rewards/margins": 2.074404716491699,
|
| 1772 |
+
"rewards/rejected": -1.9296045303344727,
|
| 1773 |
+
"step": 98,
|
| 1774 |
+
"train_speed(iter/s)": 0.005637
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.5961610839292435,
|
| 1778 |
+
"grad_norm": 1.696344017982483,
|
| 1779 |
+
"learning_rate": 0.00016838197425893202,
|
| 1780 |
+
"logits/chosen": -0.2796581983566284,
|
| 1781 |
+
"logits/rejected": -0.4003712832927704,
|
| 1782 |
+
"logps/chosen": -5.847466468811035,
|
| 1783 |
+
"logps/rejected": -30.844961166381836,
|
| 1784 |
+
"loss": 0.7925845384597778,
|
| 1785 |
+
"memory(GiB)": 46.13,
|
| 1786 |
+
"nll_loss": 0.4691714644432068,
|
| 1787 |
+
"rewards/accuracies": 0.875,
|
| 1788 |
+
"rewards/chosen": 0.15563875436782837,
|
| 1789 |
+
"rewards/margins": 1.9312589168548584,
|
| 1790 |
+
"rewards/rejected": -1.7756202220916748,
|
| 1791 |
+
"step": 99,
|
| 1792 |
+
"train_speed(iter/s)": 0.005637
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 0.602182913059842,
|
| 1796 |
+
"grad_norm": 2.3960113525390625,
|
| 1797 |
+
"learning_rate": 0.00016765088193170053,
|
| 1798 |
+
"logits/chosen": -0.2749744653701782,
|
| 1799 |
+
"logits/rejected": -0.39394593238830566,
|
| 1800 |
+
"logps/chosen": -6.603066444396973,
|
| 1801 |
+
"logps/rejected": -25.912214279174805,
|
| 1802 |
+
"loss": 0.9519745707511902,
|
| 1803 |
+
"memory(GiB)": 46.13,
|
| 1804 |
+
"nll_loss": 0.5554052591323853,
|
| 1805 |
+
"rewards/accuracies": 0.890625,
|
| 1806 |
+
"rewards/chosen": 0.11803217232227325,
|
| 1807 |
+
"rewards/margins": 1.5442473888397217,
|
| 1808 |
+
"rewards/rejected": -1.4262150526046753,
|
| 1809 |
+
"step": 100,
|
| 1810 |
+
"train_speed(iter/s)": 0.005637
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.6082047421904403,
|
| 1814 |
+
"grad_norm": 2.1803297996520996,
|
| 1815 |
+
"learning_rate": 0.00016691306063588583,
|
| 1816 |
+
"logits/chosen": -0.14278045296669006,
|
| 1817 |
+
"logits/rejected": -0.32641783356666565,
|
| 1818 |
+
"logps/chosen": -6.367352485656738,
|
| 1819 |
+
"logps/rejected": -36.09748077392578,
|
| 1820 |
+
"loss": 0.7241327166557312,
|
| 1821 |
+
"memory(GiB)": 46.13,
|
| 1822 |
+
"nll_loss": 0.4310902953147888,
|
| 1823 |
+
"rewards/accuracies": 0.90625,
|
| 1824 |
+
"rewards/chosen": 0.20738306641578674,
|
| 1825 |
+
"rewards/margins": 1.9153938293457031,
|
| 1826 |
+
"rewards/rejected": -1.7080106735229492,
|
| 1827 |
+
"step": 101,
|
| 1828 |
+
"train_speed(iter/s)": 0.005632
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 0.6142265713210388,
|
| 1832 |
+
"grad_norm": 2.27510142326355,
|
| 1833 |
+
"learning_rate": 0.00016616858375968595,
|
| 1834 |
+
"logits/chosen": -0.20140963792800903,
|
| 1835 |
+
"logits/rejected": -0.34892868995666504,
|
| 1836 |
+
"logps/chosen": -4.8850579261779785,
|
| 1837 |
+
"logps/rejected": -34.458396911621094,
|
| 1838 |
+
"loss": 0.6944241523742676,
|
| 1839 |
+
"memory(GiB)": 46.13,
|
| 1840 |
+
"nll_loss": 0.39014381170272827,
|
| 1841 |
+
"rewards/accuracies": 0.828125,
|
| 1842 |
+
"rewards/chosen": 0.05396907776594162,
|
| 1843 |
+
"rewards/margins": 2.1584324836730957,
|
| 1844 |
+
"rewards/rejected": -2.1044633388519287,
|
| 1845 |
+
"step": 102,
|
| 1846 |
+
"train_speed(iter/s)": 0.005633
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.6202484004516372,
|
| 1850 |
+
"grad_norm": 3.8538124561309814,
|
| 1851 |
+
"learning_rate": 0.00016541752535330345,
|
| 1852 |
+
"logits/chosen": -0.15393608808517456,
|
| 1853 |
+
"logits/rejected": -0.32232990860939026,
|
| 1854 |
+
"logps/chosen": -7.490941047668457,
|
| 1855 |
+
"logps/rejected": -38.019405364990234,
|
| 1856 |
+
"loss": 0.899553656578064,
|
| 1857 |
+
"memory(GiB)": 46.13,
|
| 1858 |
+
"nll_loss": 0.5592579245567322,
|
| 1859 |
+
"rewards/accuracies": 0.875,
|
| 1860 |
+
"rewards/chosen": -0.04624926298856735,
|
| 1861 |
+
"rewards/margins": 2.2129275798797607,
|
| 1862 |
+
"rewards/rejected": -2.259176731109619,
|
| 1863 |
+
"step": 103,
|
| 1864 |
+
"train_speed(iter/s)": 0.005633
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.6262702295822355,
|
| 1868 |
+
"grad_norm": 3.2243363857269287,
|
| 1869 |
+
"learning_rate": 0.00016465996012157995,
|
| 1870 |
+
"logits/chosen": -0.1763302981853485,
|
| 1871 |
+
"logits/rejected": -0.30002710223197937,
|
| 1872 |
+
"logps/chosen": -7.91542387008667,
|
| 1873 |
+
"logps/rejected": -35.163421630859375,
|
| 1874 |
+
"loss": 0.8102483153343201,
|
| 1875 |
+
"memory(GiB)": 46.13,
|
| 1876 |
+
"nll_loss": 0.5111361145973206,
|
| 1877 |
+
"rewards/accuracies": 0.859375,
|
| 1878 |
+
"rewards/chosen": 0.1345573365688324,
|
| 1879 |
+
"rewards/margins": 2.1800684928894043,
|
| 1880 |
+
"rewards/rejected": -2.045511245727539,
|
| 1881 |
+
"step": 104,
|
| 1882 |
+
"train_speed(iter/s)": 0.005633
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.632292058712834,
|
| 1886 |
+
"grad_norm": 2.4296176433563232,
|
| 1887 |
+
"learning_rate": 0.0001638959634165656,
|
| 1888 |
+
"logits/chosen": -0.2042100727558136,
|
| 1889 |
+
"logits/rejected": -0.3498837947845459,
|
| 1890 |
+
"logps/chosen": -5.70271635055542,
|
| 1891 |
+
"logps/rejected": -33.96530532836914,
|
| 1892 |
+
"loss": 0.7614078521728516,
|
| 1893 |
+
"memory(GiB)": 46.13,
|
| 1894 |
+
"nll_loss": 0.46085914969444275,
|
| 1895 |
+
"rewards/accuracies": 0.84375,
|
| 1896 |
+
"rewards/chosen": 0.10853511095046997,
|
| 1897 |
+
"rewards/margins": 2.407238483428955,
|
| 1898 |
+
"rewards/rejected": -2.298703670501709,
|
| 1899 |
+
"step": 105,
|
| 1900 |
+
"train_speed(iter/s)": 0.005633
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.6383138878434325,
|
| 1904 |
+
"grad_norm": 5.010378837585449,
|
| 1905 |
+
"learning_rate": 0.0001631256112300239,
|
| 1906 |
+
"logits/chosen": -0.17106319963932037,
|
| 1907 |
+
"logits/rejected": -0.3643282651901245,
|
| 1908 |
+
"logps/chosen": -5.012183666229248,
|
| 1909 |
+
"logps/rejected": -39.59567642211914,
|
| 1910 |
+
"loss": 0.7013410329818726,
|
| 1911 |
+
"memory(GiB)": 46.13,
|
| 1912 |
+
"nll_loss": 0.4339035153388977,
|
| 1913 |
+
"rewards/accuracies": 0.859375,
|
| 1914 |
+
"rewards/chosen": 0.07484348863363266,
|
| 1915 |
+
"rewards/margins": 2.594076156616211,
|
| 1916 |
+
"rewards/rejected": -2.5192322731018066,
|
| 1917 |
+
"step": 106,
|
| 1918 |
+
"train_speed(iter/s)": 0.005633
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.6443357169740309,
|
| 1922 |
+
"grad_norm": 3.9870245456695557,
|
| 1923 |
+
"learning_rate": 0.00016234898018587337,
|
| 1924 |
+
"logits/chosen": -0.13750208914279938,
|
| 1925 |
+
"logits/rejected": -0.3746242821216583,
|
| 1926 |
+
"logps/chosen": -5.721056938171387,
|
| 1927 |
+
"logps/rejected": -44.44032287597656,
|
| 1928 |
+
"loss": 0.7298431396484375,
|
| 1929 |
+
"memory(GiB)": 46.13,
|
| 1930 |
+
"nll_loss": 0.4898163080215454,
|
| 1931 |
+
"rewards/accuracies": 0.921875,
|
| 1932 |
+
"rewards/chosen": 0.036943770945072174,
|
| 1933 |
+
"rewards/margins": 2.8432068824768066,
|
| 1934 |
+
"rewards/rejected": -2.806262969970703,
|
| 1935 |
+
"step": 107,
|
| 1936 |
+
"train_speed(iter/s)": 0.005633
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 0.6503575461046293,
|
| 1940 |
+
"grad_norm": 2.7155423164367676,
|
| 1941 |
+
"learning_rate": 0.0001615661475325658,
|
| 1942 |
+
"logits/chosen": -0.1980631947517395,
|
| 1943 |
+
"logits/rejected": -0.36777210235595703,
|
| 1944 |
+
"logps/chosen": -5.860571384429932,
|
| 1945 |
+
"logps/rejected": -45.095069885253906,
|
| 1946 |
+
"loss": 0.6415181756019592,
|
| 1947 |
+
"memory(GiB)": 46.13,
|
| 1948 |
+
"nll_loss": 0.43772590160369873,
|
| 1949 |
+
"rewards/accuracies": 0.921875,
|
| 1950 |
+
"rewards/chosen": 0.029481690376996994,
|
| 1951 |
+
"rewards/margins": 2.904158115386963,
|
| 1952 |
+
"rewards/rejected": -2.874676465988159,
|
| 1953 |
+
"step": 108,
|
| 1954 |
+
"train_speed(iter/s)": 0.005633
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.6563793752352277,
|
| 1958 |
+
"grad_norm": 3.9366610050201416,
|
| 1959 |
+
"learning_rate": 0.00016077719113540302,
|
| 1960 |
+
"logits/chosen": -0.201637402176857,
|
| 1961 |
+
"logits/rejected": -0.36231857538223267,
|
| 1962 |
+
"logps/chosen": -8.20331859588623,
|
| 1963 |
+
"logps/rejected": -47.63356018066406,
|
| 1964 |
+
"loss": 0.8109980821609497,
|
| 1965 |
+
"memory(GiB)": 46.13,
|
| 1966 |
+
"nll_loss": 0.588591456413269,
|
| 1967 |
+
"rewards/accuracies": 0.90625,
|
| 1968 |
+
"rewards/chosen": -0.11406072974205017,
|
| 1969 |
+
"rewards/margins": 3.15336275100708,
|
| 1970 |
+
"rewards/rejected": -3.267423629760742,
|
| 1971 |
+
"step": 109,
|
| 1972 |
+
"train_speed(iter/s)": 0.005633
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 0.6624012043658262,
|
| 1976 |
+
"grad_norm": 1.8952491283416748,
|
| 1977 |
+
"learning_rate": 0.00015998218946879138,
|
| 1978 |
+
"logits/chosen": -0.20580148696899414,
|
| 1979 |
+
"logits/rejected": -0.37684956192970276,
|
| 1980 |
+
"logps/chosen": -4.04914665222168,
|
| 1981 |
+
"logps/rejected": -40.22568130493164,
|
| 1982 |
+
"loss": 0.6388105154037476,
|
| 1983 |
+
"memory(GiB)": 46.13,
|
| 1984 |
+
"nll_loss": 0.39933112263679504,
|
| 1985 |
+
"rewards/accuracies": 0.921875,
|
| 1986 |
+
"rewards/chosen": 0.07669900357723236,
|
| 1987 |
+
"rewards/margins": 2.701504945755005,
|
| 1988 |
+
"rewards/rejected": -2.6248061656951904,
|
| 1989 |
+
"step": 110,
|
| 1990 |
+
"train_speed(iter/s)": 0.005634
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.6684230334964245,
|
| 1994 |
+
"grad_norm": 2.1146159172058105,
|
| 1995 |
+
"learning_rate": 0.00015918122160843678,
|
| 1996 |
+
"logits/chosen": -0.21242693066596985,
|
| 1997 |
+
"logits/rejected": -0.3542603850364685,
|
| 1998 |
+
"logps/chosen": -7.400607109069824,
|
| 1999 |
+
"logps/rejected": -41.83246612548828,
|
| 2000 |
+
"loss": 0.9009187817573547,
|
| 2001 |
+
"memory(GiB)": 46.13,
|
| 2002 |
+
"nll_loss": 0.6562238931655884,
|
| 2003 |
+
"rewards/accuracies": 0.90625,
|
| 2004 |
+
"rewards/chosen": 0.07808566093444824,
|
| 2005 |
+
"rewards/margins": 2.6408753395080566,
|
| 2006 |
+
"rewards/rejected": -2.5627896785736084,
|
| 2007 |
+
"step": 111,
|
| 2008 |
+
"train_speed(iter/s)": 0.005634
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.674444862627023,
|
| 2012 |
+
"grad_norm": 3.891860008239746,
|
| 2013 |
+
"learning_rate": 0.000158374367223479,
|
| 2014 |
+
"logits/chosen": -0.2456948161125183,
|
| 2015 |
+
"logits/rejected": -0.3476937413215637,
|
| 2016 |
+
"logps/chosen": -6.91848087310791,
|
| 2017 |
+
"logps/rejected": -35.34187698364258,
|
| 2018 |
+
"loss": 0.9391785860061646,
|
| 2019 |
+
"memory(GiB)": 46.13,
|
| 2020 |
+
"nll_loss": 0.618550181388855,
|
| 2021 |
+
"rewards/accuracies": 0.8125,
|
| 2022 |
+
"rewards/chosen": -0.011886654421687126,
|
| 2023 |
+
"rewards/margins": 2.1392247676849365,
|
| 2024 |
+
"rewards/rejected": -2.151111602783203,
|
| 2025 |
+
"step": 112,
|
| 2026 |
+
"train_speed(iter/s)": 0.005635
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.6804666917576214,
|
| 2030 |
+
"grad_norm": 4.234856128692627,
|
| 2031 |
+
"learning_rate": 0.00015756170656856737,
|
| 2032 |
+
"logits/chosen": -0.11636840552091599,
|
| 2033 |
+
"logits/rejected": -0.23480768501758575,
|
| 2034 |
+
"logps/chosen": -7.3483710289001465,
|
| 2035 |
+
"logps/rejected": -33.948822021484375,
|
| 2036 |
+
"loss": 0.8160063624382019,
|
| 2037 |
+
"memory(GiB)": 46.13,
|
| 2038 |
+
"nll_loss": 0.5110282301902771,
|
| 2039 |
+
"rewards/accuracies": 0.953125,
|
| 2040 |
+
"rewards/chosen": 0.08705638349056244,
|
| 2041 |
+
"rewards/margins": 2.111095905303955,
|
| 2042 |
+
"rewards/rejected": -2.0240395069122314,
|
| 2043 |
+
"step": 113,
|
| 2044 |
+
"train_speed(iter/s)": 0.005634
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.6864885208882198,
|
| 2048 |
+
"grad_norm": 1.5901049375534058,
|
| 2049 |
+
"learning_rate": 0.0001567433204758782,
|
| 2050 |
+
"logits/chosen": -0.15935438871383667,
|
| 2051 |
+
"logits/rejected": -0.3056758642196655,
|
| 2052 |
+
"logps/chosen": -5.887468338012695,
|
| 2053 |
+
"logps/rejected": -39.561500549316406,
|
| 2054 |
+
"loss": 0.7292969226837158,
|
| 2055 |
+
"memory(GiB)": 46.13,
|
| 2056 |
+
"nll_loss": 0.4773525893688202,
|
| 2057 |
+
"rewards/accuracies": 0.921875,
|
| 2058 |
+
"rewards/chosen": 0.141588032245636,
|
| 2059 |
+
"rewards/margins": 2.442899703979492,
|
| 2060 |
+
"rewards/rejected": -2.301311731338501,
|
| 2061 |
+
"step": 114,
|
| 2062 |
+
"train_speed(iter/s)": 0.005634
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.6925103500188182,
|
| 2066 |
+
"grad_norm": 2.1147372722625732,
|
| 2067 |
+
"learning_rate": 0.0001559192903470747,
|
| 2068 |
+
"logits/chosen": -0.1537177711725235,
|
| 2069 |
+
"logits/rejected": -0.2976364493370056,
|
| 2070 |
+
"logps/chosen": -7.735870361328125,
|
| 2071 |
+
"logps/rejected": -36.14551544189453,
|
| 2072 |
+
"loss": 0.822877049446106,
|
| 2073 |
+
"memory(GiB)": 46.13,
|
| 2074 |
+
"nll_loss": 0.5838936567306519,
|
| 2075 |
+
"rewards/accuracies": 0.90625,
|
| 2076 |
+
"rewards/chosen": 0.16609230637550354,
|
| 2077 |
+
"rewards/margins": 2.184666633605957,
|
| 2078 |
+
"rewards/rejected": -2.0185742378234863,
|
| 2079 |
+
"step": 115,
|
| 2080 |
+
"train_speed(iter/s)": 0.005634
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 0.6985321791494167,
|
| 2084 |
+
"grad_norm": 3.8271219730377197,
|
| 2085 |
+
"learning_rate": 0.00015508969814521025,
|
| 2086 |
+
"logits/chosen": -0.16476985812187195,
|
| 2087 |
+
"logits/rejected": -0.2944994568824768,
|
| 2088 |
+
"logps/chosen": -4.954969882965088,
|
| 2089 |
+
"logps/rejected": -31.42076301574707,
|
| 2090 |
+
"loss": 0.8119629621505737,
|
| 2091 |
+
"memory(GiB)": 46.13,
|
| 2092 |
+
"nll_loss": 0.4658912420272827,
|
| 2093 |
+
"rewards/accuracies": 0.828125,
|
| 2094 |
+
"rewards/chosen": 0.09785570204257965,
|
| 2095 |
+
"rewards/margins": 1.8094596862792969,
|
| 2096 |
+
"rewards/rejected": -1.711604118347168,
|
| 2097 |
+
"step": 116,
|
| 2098 |
+
"train_speed(iter/s)": 0.005634
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.704554008280015,
|
| 2102 |
+
"grad_norm": 1.9744106531143188,
|
| 2103 |
+
"learning_rate": 0.00015425462638657595,
|
| 2104 |
+
"logits/chosen": -0.2048090547323227,
|
| 2105 |
+
"logits/rejected": -0.2695329189300537,
|
| 2106 |
+
"logps/chosen": -5.02174711227417,
|
| 2107 |
+
"logps/rejected": -31.327930450439453,
|
| 2108 |
+
"loss": 0.6611427068710327,
|
| 2109 |
+
"memory(GiB)": 46.13,
|
| 2110 |
+
"nll_loss": 0.3884551525115967,
|
| 2111 |
+
"rewards/accuracies": 0.875,
|
| 2112 |
+
"rewards/chosen": 0.18278352916240692,
|
| 2113 |
+
"rewards/margins": 2.1040430068969727,
|
| 2114 |
+
"rewards/rejected": -1.9212596416473389,
|
| 2115 |
+
"step": 117,
|
| 2116 |
+
"train_speed(iter/s)": 0.005634
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.7105758374106135,
|
| 2120 |
+
"grad_norm": 2.7671103477478027,
|
| 2121 |
+
"learning_rate": 0.00015341415813249288,
|
| 2122 |
+
"logits/chosen": -0.18153494596481323,
|
| 2123 |
+
"logits/rejected": -0.31533682346343994,
|
| 2124 |
+
"logps/chosen": -3.5208582878112793,
|
| 2125 |
+
"logps/rejected": -28.974889755249023,
|
| 2126 |
+
"loss": 0.6663014888763428,
|
| 2127 |
+
"memory(GiB)": 46.13,
|
| 2128 |
+
"nll_loss": 0.31734898686408997,
|
| 2129 |
+
"rewards/accuracies": 0.859375,
|
| 2130 |
+
"rewards/chosen": 0.08336202800273895,
|
| 2131 |
+
"rewards/margins": 1.5573171377182007,
|
| 2132 |
+
"rewards/rejected": -1.4739550352096558,
|
| 2133 |
+
"step": 118,
|
| 2134 |
+
"train_speed(iter/s)": 0.005635
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.7165976665412119,
|
| 2138 |
+
"grad_norm": 2.2519025802612305,
|
| 2139 |
+
"learning_rate": 0.00015256837698105047,
|
| 2140 |
+
"logits/chosen": -0.18909907341003418,
|
| 2141 |
+
"logits/rejected": -0.32073426246643066,
|
| 2142 |
+
"logps/chosen": -5.691015243530273,
|
| 2143 |
+
"logps/rejected": -38.229949951171875,
|
| 2144 |
+
"loss": 0.6945815086364746,
|
| 2145 |
+
"memory(GiB)": 46.13,
|
| 2146 |
+
"nll_loss": 0.4534215033054352,
|
| 2147 |
+
"rewards/accuracies": 0.9375,
|
| 2148 |
+
"rewards/chosen": -0.016410622745752335,
|
| 2149 |
+
"rewards/margins": 2.294341564178467,
|
| 2150 |
+
"rewards/rejected": -2.3107523918151855,
|
| 2151 |
+
"step": 119,
|
| 2152 |
+
"train_speed(iter/s)": 0.005635
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.7226194956718103,
|
| 2156 |
+
"grad_norm": 2.6865475177764893,
|
| 2157 |
+
"learning_rate": 0.00015171736705879126,
|
| 2158 |
+
"logits/chosen": -0.11694711446762085,
|
| 2159 |
+
"logits/rejected": -0.30947864055633545,
|
| 2160 |
+
"logps/chosen": -4.004083156585693,
|
| 2161 |
+
"logps/rejected": -42.21112060546875,
|
| 2162 |
+
"loss": 0.5099095702171326,
|
| 2163 |
+
"memory(GiB)": 46.13,
|
| 2164 |
+
"nll_loss": 0.2805330157279968,
|
| 2165 |
+
"rewards/accuracies": 0.90625,
|
| 2166 |
+
"rewards/chosen": 0.08305128663778305,
|
| 2167 |
+
"rewards/margins": 2.5462100505828857,
|
| 2168 |
+
"rewards/rejected": -2.46315860748291,
|
| 2169 |
+
"step": 120,
|
| 2170 |
+
"train_speed(iter/s)": 0.005635
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.7286413248024087,
|
| 2174 |
+
"grad_norm": 5.433821201324463,
|
| 2175 |
+
"learning_rate": 0.00015086121301234316,
|
| 2176 |
+
"logits/chosen": -0.15977299213409424,
|
| 2177 |
+
"logits/rejected": -0.30839866399765015,
|
| 2178 |
+
"logps/chosen": -7.34042501449585,
|
| 2179 |
+
"logps/rejected": -38.30808639526367,
|
| 2180 |
+
"loss": 0.9920263886451721,
|
| 2181 |
+
"memory(GiB)": 46.13,
|
| 2182 |
+
"nll_loss": 0.6664140224456787,
|
| 2183 |
+
"rewards/accuracies": 0.84375,
|
| 2184 |
+
"rewards/chosen": -0.023471834138035774,
|
| 2185 |
+
"rewards/margins": 2.068791151046753,
|
| 2186 |
+
"rewards/rejected": -2.0922627449035645,
|
| 2187 |
+
"step": 121,
|
| 2188 |
+
"train_speed(iter/s)": 0.005635
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 0.7346631539330072,
|
| 2192 |
+
"grad_norm": 1.8786602020263672,
|
| 2193 |
+
"learning_rate": 0.00015000000000000001,
|
| 2194 |
+
"logits/chosen": -0.12786100804805756,
|
| 2195 |
+
"logits/rejected": -0.2967092990875244,
|
| 2196 |
+
"logps/chosen": -4.349850654602051,
|
| 2197 |
+
"logps/rejected": -32.289390563964844,
|
| 2198 |
+
"loss": 0.6910618543624878,
|
| 2199 |
+
"memory(GiB)": 46.13,
|
| 2200 |
+
"nll_loss": 0.37633225321769714,
|
| 2201 |
+
"rewards/accuracies": 0.84375,
|
| 2202 |
+
"rewards/chosen": 0.09151757508516312,
|
| 2203 |
+
"rewards/margins": 1.6974818706512451,
|
| 2204 |
+
"rewards/rejected": -1.6059644222259521,
|
| 2205 |
+
"step": 122,
|
| 2206 |
+
"train_speed(iter/s)": 0.005634
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.7406849830636055,
|
| 2210 |
+
"grad_norm": 2.327937364578247,
|
| 2211 |
+
"learning_rate": 0.00014913381368325115,
|
| 2212 |
+
"logits/chosen": -0.12772715091705322,
|
| 2213 |
+
"logits/rejected": -0.3204698860645294,
|
| 2214 |
+
"logps/chosen": -3.4514901638031006,
|
| 2215 |
+
"logps/rejected": -40.17311096191406,
|
| 2216 |
+
"loss": 0.5442243814468384,
|
| 2217 |
+
"memory(GiB)": 46.13,
|
| 2218 |
+
"nll_loss": 0.2678568661212921,
|
| 2219 |
+
"rewards/accuracies": 0.84375,
|
| 2220 |
+
"rewards/chosen": 0.06198100745677948,
|
| 2221 |
+
"rewards/margins": 2.348788022994995,
|
| 2222 |
+
"rewards/rejected": -2.286807060241699,
|
| 2223 |
+
"step": 123,
|
| 2224 |
+
"train_speed(iter/s)": 0.005634
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.746706812194204,
|
| 2228 |
+
"grad_norm": 3.345496892929077,
|
| 2229 |
+
"learning_rate": 0.0001482627402182611,
|
| 2230 |
+
"logits/chosen": -0.23605135083198547,
|
| 2231 |
+
"logits/rejected": -0.3248905539512634,
|
| 2232 |
+
"logps/chosen": -6.426857948303223,
|
| 2233 |
+
"logps/rejected": -27.78033447265625,
|
| 2234 |
+
"loss": 0.8108398914337158,
|
| 2235 |
+
"memory(GiB)": 46.13,
|
| 2236 |
+
"nll_loss": 0.4387803077697754,
|
| 2237 |
+
"rewards/accuracies": 0.828125,
|
| 2238 |
+
"rewards/chosen": 0.12587979435920715,
|
| 2239 |
+
"rewards/margins": 1.6669539213180542,
|
| 2240 |
+
"rewards/rejected": -1.5410741567611694,
|
| 2241 |
+
"step": 124,
|
| 2242 |
+
"train_speed(iter/s)": 0.005634
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.7527286413248024,
|
| 2246 |
+
"grad_norm": 2.567275285720825,
|
| 2247 |
+
"learning_rate": 0.00014738686624729986,
|
| 2248 |
+
"logits/chosen": -0.19267147779464722,
|
| 2249 |
+
"logits/rejected": -0.3316629230976105,
|
| 2250 |
+
"logps/chosen": -5.075592994689941,
|
| 2251 |
+
"logps/rejected": -33.312400817871094,
|
| 2252 |
+
"loss": 0.7099701166152954,
|
| 2253 |
+
"memory(GiB)": 46.13,
|
| 2254 |
+
"nll_loss": 0.3936714828014374,
|
| 2255 |
+
"rewards/accuracies": 0.875,
|
| 2256 |
+
"rewards/chosen": 0.13089658319950104,
|
| 2257 |
+
"rewards/margins": 1.870408535003662,
|
| 2258 |
+
"rewards/rejected": -1.7395117282867432,
|
| 2259 |
+
"step": 125,
|
| 2260 |
+
"train_speed(iter/s)": 0.005635
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.7587504704554008,
|
| 2264 |
+
"grad_norm": 3.0261592864990234,
|
| 2265 |
+
"learning_rate": 0.00014650627889012507,
|
| 2266 |
+
"logits/chosen": -0.2155275195837021,
|
| 2267 |
+
"logits/rejected": -0.37678274512290955,
|
| 2268 |
+
"logps/chosen": -4.682234764099121,
|
| 2269 |
+
"logps/rejected": -28.974559783935547,
|
| 2270 |
+
"loss": 0.7498282790184021,
|
| 2271 |
+
"memory(GiB)": 46.13,
|
| 2272 |
+
"nll_loss": 0.4256855249404907,
|
| 2273 |
+
"rewards/accuracies": 0.890625,
|
| 2274 |
+
"rewards/chosen": 0.15469788014888763,
|
| 2275 |
+
"rewards/margins": 1.657031774520874,
|
| 2276 |
+
"rewards/rejected": -1.502333641052246,
|
| 2277 |
+
"step": 126,
|
| 2278 |
+
"train_speed(iter/s)": 0.005635
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.7647722995859992,
|
| 2282 |
+
"grad_norm": 2.6022913455963135,
|
| 2283 |
+
"learning_rate": 0.0001456210657353163,
|
| 2284 |
+
"logits/chosen": -0.19964055716991425,
|
| 2285 |
+
"logits/rejected": -0.35465607047080994,
|
| 2286 |
+
"logps/chosen": -3.941540479660034,
|
| 2287 |
+
"logps/rejected": -32.9716682434082,
|
| 2288 |
+
"loss": 0.6142888069152832,
|
| 2289 |
+
"memory(GiB)": 46.13,
|
| 2290 |
+
"nll_loss": 0.3339819610118866,
|
| 2291 |
+
"rewards/accuracies": 0.953125,
|
| 2292 |
+
"rewards/chosen": 0.15736861526966095,
|
| 2293 |
+
"rewards/margins": 2.048983335494995,
|
| 2294 |
+
"rewards/rejected": -1.8916147947311401,
|
| 2295 |
+
"step": 127,
|
| 2296 |
+
"train_speed(iter/s)": 0.005635
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 0.7707941287165977,
|
| 2300 |
+
"grad_norm": 2.2070999145507812,
|
| 2301 |
+
"learning_rate": 0.00014473131483156327,
|
| 2302 |
+
"logits/chosen": -0.24945035576820374,
|
| 2303 |
+
"logits/rejected": -0.32592612504959106,
|
| 2304 |
+
"logps/chosen": -5.566514015197754,
|
| 2305 |
+
"logps/rejected": -27.138111114501953,
|
| 2306 |
+
"loss": 0.7188521027565002,
|
| 2307 |
+
"memory(GiB)": 46.13,
|
| 2308 |
+
"nll_loss": 0.4044472575187683,
|
| 2309 |
+
"rewards/accuracies": 0.921875,
|
| 2310 |
+
"rewards/chosen": 0.16954132914543152,
|
| 2311 |
+
"rewards/margins": 1.7891260385513306,
|
| 2312 |
+
"rewards/rejected": -1.6195846796035767,
|
| 2313 |
+
"step": 128,
|
| 2314 |
+
"train_speed(iter/s)": 0.005635
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.776815957847196,
|
| 2318 |
+
"grad_norm": 1.9873756170272827,
|
| 2319 |
+
"learning_rate": 0.00014383711467890774,
|
| 2320 |
+
"logits/chosen": -0.18453390896320343,
|
| 2321 |
+
"logits/rejected": -0.3942897319793701,
|
| 2322 |
+
"logps/chosen": -4.547736167907715,
|
| 2323 |
+
"logps/rejected": -43.979976654052734,
|
| 2324 |
+
"loss": 0.5406314730644226,
|
| 2325 |
+
"memory(GiB)": 46.13,
|
| 2326 |
+
"nll_loss": 0.3523382544517517,
|
| 2327 |
+
"rewards/accuracies": 0.953125,
|
| 2328 |
+
"rewards/chosen": 0.08807562291622162,
|
| 2329 |
+
"rewards/margins": 2.9467825889587402,
|
| 2330 |
+
"rewards/rejected": -2.8587071895599365,
|
| 2331 |
+
"step": 129,
|
| 2332 |
+
"train_speed(iter/s)": 0.005635
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 0.7828377869777945,
|
| 2336 |
+
"grad_norm": 3.6719274520874023,
|
| 2337 |
+
"learning_rate": 0.00014293855421994094,
|
| 2338 |
+
"logits/chosen": -0.20188507437705994,
|
| 2339 |
+
"logits/rejected": -0.34430867433547974,
|
| 2340 |
+
"logps/chosen": -5.333648204803467,
|
| 2341 |
+
"logps/rejected": -41.819515228271484,
|
| 2342 |
+
"loss": 0.7074599862098694,
|
| 2343 |
+
"memory(GiB)": 46.13,
|
| 2344 |
+
"nll_loss": 0.44100311398506165,
|
| 2345 |
+
"rewards/accuracies": 0.921875,
|
| 2346 |
+
"rewards/chosen": 0.027486218139529228,
|
| 2347 |
+
"rewards/margins": 2.8748302459716797,
|
| 2348 |
+
"rewards/rejected": -2.847343921661377,
|
| 2349 |
+
"step": 130,
|
| 2350 |
+
"train_speed(iter/s)": 0.005636
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 0.788859616108393,
|
| 2354 |
+
"grad_norm": 3.5544822216033936,
|
| 2355 |
+
"learning_rate": 0.00014203572283095657,
|
| 2356 |
+
"logits/chosen": -0.1257411241531372,
|
| 2357 |
+
"logits/rejected": -0.35228949785232544,
|
| 2358 |
+
"logps/chosen": -5.203995704650879,
|
| 2359 |
+
"logps/rejected": -53.26233673095703,
|
| 2360 |
+
"loss": 0.763848602771759,
|
| 2361 |
+
"memory(GiB)": 46.13,
|
| 2362 |
+
"nll_loss": 0.5348649024963379,
|
| 2363 |
+
"rewards/accuracies": 0.90625,
|
| 2364 |
+
"rewards/chosen": 0.0426684133708477,
|
| 2365 |
+
"rewards/margins": 3.6239006519317627,
|
| 2366 |
+
"rewards/rejected": -3.5812320709228516,
|
| 2367 |
+
"step": 131,
|
| 2368 |
+
"train_speed(iter/s)": 0.005636
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.7948814452389913,
|
| 2372 |
+
"grad_norm": 2.9171648025512695,
|
| 2373 |
+
"learning_rate": 0.00014112871031306119,
|
| 2374 |
+
"logits/chosen": -0.19912837445735931,
|
| 2375 |
+
"logits/rejected": -0.3755126893520355,
|
| 2376 |
+
"logps/chosen": -7.589679718017578,
|
| 2377 |
+
"logps/rejected": -46.79793930053711,
|
| 2378 |
+
"loss": 0.7475419640541077,
|
| 2379 |
+
"memory(GiB)": 46.13,
|
| 2380 |
+
"nll_loss": 0.5176687240600586,
|
| 2381 |
+
"rewards/accuracies": 0.875,
|
| 2382 |
+
"rewards/chosen": 0.19966475665569305,
|
| 2383 |
+
"rewards/margins": 3.243659496307373,
|
| 2384 |
+
"rewards/rejected": -3.043994426727295,
|
| 2385 |
+
"step": 132,
|
| 2386 |
+
"train_speed(iter/s)": 0.005635
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.8009032743695897,
|
| 2390 |
+
"grad_norm": 3.5576326847076416,
|
| 2391 |
+
"learning_rate": 0.00014021760688324176,
|
| 2392 |
+
"logits/chosen": -0.17760571837425232,
|
| 2393 |
+
"logits/rejected": -0.34351277351379395,
|
| 2394 |
+
"logps/chosen": -6.067855358123779,
|
| 2395 |
+
"logps/rejected": -45.17057800292969,
|
| 2396 |
+
"loss": 0.7542024850845337,
|
| 2397 |
+
"memory(GiB)": 46.13,
|
| 2398 |
+
"nll_loss": 0.46581241488456726,
|
| 2399 |
+
"rewards/accuracies": 0.859375,
|
| 2400 |
+
"rewards/chosen": -0.06834451109170914,
|
| 2401 |
+
"rewards/margins": 2.9498496055603027,
|
| 2402 |
+
"rewards/rejected": -3.0181939601898193,
|
| 2403 |
+
"step": 133,
|
| 2404 |
+
"train_speed(iter/s)": 0.005635
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.8069251035001882,
|
| 2408 |
+
"grad_norm": 3.7524359226226807,
|
| 2409 |
+
"learning_rate": 0.00013930250316539238,
|
| 2410 |
+
"logits/chosen": -0.24070125818252563,
|
| 2411 |
+
"logits/rejected": -0.36900123953819275,
|
| 2412 |
+
"logps/chosen": -6.3118109703063965,
|
| 2413 |
+
"logps/rejected": -41.5582389831543,
|
| 2414 |
+
"loss": 0.8333712220191956,
|
| 2415 |
+
"memory(GiB)": 46.13,
|
| 2416 |
+
"nll_loss": 0.5454068183898926,
|
| 2417 |
+
"rewards/accuracies": 0.90625,
|
| 2418 |
+
"rewards/chosen": 0.0363120511174202,
|
| 2419 |
+
"rewards/margins": 2.832289695739746,
|
| 2420 |
+
"rewards/rejected": -2.7959775924682617,
|
| 2421 |
+
"step": 134,
|
| 2422 |
+
"train_speed(iter/s)": 0.005636
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.8129469326307867,
|
| 2426 |
+
"grad_norm": 4.076293468475342,
|
| 2427 |
+
"learning_rate": 0.00013838349018130007,
|
| 2428 |
+
"logits/chosen": -0.22921916842460632,
|
| 2429 |
+
"logits/rejected": -0.3980764150619507,
|
| 2430 |
+
"logps/chosen": -6.020735263824463,
|
| 2431 |
+
"logps/rejected": -36.49159622192383,
|
| 2432 |
+
"loss": 0.7660678029060364,
|
| 2433 |
+
"memory(GiB)": 46.13,
|
| 2434 |
+
"nll_loss": 0.5027486085891724,
|
| 2435 |
+
"rewards/accuracies": 0.84375,
|
| 2436 |
+
"rewards/chosen": 0.11397107690572739,
|
| 2437 |
+
"rewards/margins": 2.2893691062927246,
|
| 2438 |
+
"rewards/rejected": -2.175398349761963,
|
| 2439 |
+
"step": 135,
|
| 2440 |
+
"train_speed(iter/s)": 0.005636
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 0.818968761761385,
|
| 2444 |
+
"grad_norm": 3.8375930786132812,
|
| 2445 |
+
"learning_rate": 0.00013746065934159123,
|
| 2446 |
+
"logits/chosen": -0.21597175300121307,
|
| 2447 |
+
"logits/rejected": -0.3748128414154053,
|
| 2448 |
+
"logps/chosen": -5.917651176452637,
|
| 2449 |
+
"logps/rejected": -34.7623176574707,
|
| 2450 |
+
"loss": 0.7133091688156128,
|
| 2451 |
+
"memory(GiB)": 46.13,
|
| 2452 |
+
"nll_loss": 0.43554598093032837,
|
| 2453 |
+
"rewards/accuracies": 0.90625,
|
| 2454 |
+
"rewards/chosen": 0.14565804600715637,
|
| 2455 |
+
"rewards/margins": 2.124354124069214,
|
| 2456 |
+
"rewards/rejected": -1.9786962270736694,
|
| 2457 |
+
"step": 136,
|
| 2458 |
+
"train_speed(iter/s)": 0.005636
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.8249905908919835,
|
| 2462 |
+
"grad_norm": 3.4924089908599854,
|
| 2463 |
+
"learning_rate": 0.00013653410243663952,
|
| 2464 |
+
"logits/chosen": -0.18882140517234802,
|
| 2465 |
+
"logits/rejected": -0.39621540904045105,
|
| 2466 |
+
"logps/chosen": -3.5247278213500977,
|
| 2467 |
+
"logps/rejected": -39.52235794067383,
|
| 2468 |
+
"loss": 0.5699490308761597,
|
| 2469 |
+
"memory(GiB)": 46.13,
|
| 2470 |
+
"nll_loss": 0.3287566304206848,
|
| 2471 |
+
"rewards/accuracies": 0.9375,
|
| 2472 |
+
"rewards/chosen": 0.16140440106391907,
|
| 2473 |
+
"rewards/margins": 2.0828282833099365,
|
| 2474 |
+
"rewards/rejected": -1.9214237928390503,
|
| 2475 |
+
"step": 137,
|
| 2476 |
+
"train_speed(iter/s)": 0.005635
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.8310124200225819,
|
| 2480 |
+
"grad_norm": 2.952988386154175,
|
| 2481 |
+
"learning_rate": 0.00013560391162743569,
|
| 2482 |
+
"logits/chosen": -0.1859571784734726,
|
| 2483 |
+
"logits/rejected": -0.381632536649704,
|
| 2484 |
+
"logps/chosen": -4.896010398864746,
|
| 2485 |
+
"logps/rejected": -35.494384765625,
|
| 2486 |
+
"loss": 0.7929538488388062,
|
| 2487 |
+
"memory(GiB)": 46.13,
|
| 2488 |
+
"nll_loss": 0.4508986473083496,
|
| 2489 |
+
"rewards/accuracies": 0.890625,
|
| 2490 |
+
"rewards/chosen": 0.021028656512498856,
|
| 2491 |
+
"rewards/margins": 1.990783452987671,
|
| 2492 |
+
"rewards/rejected": -1.969754695892334,
|
| 2493 |
+
"step": 138,
|
| 2494 |
+
"train_speed(iter/s)": 0.005635
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.8370342491531803,
|
| 2498 |
+
"grad_norm": 2.0521535873413086,
|
| 2499 |
+
"learning_rate": 0.00013467017943642073,
|
| 2500 |
+
"logits/chosen": -0.24490389227867126,
|
| 2501 |
+
"logits/rejected": -0.42894116044044495,
|
| 2502 |
+
"logps/chosen": -4.2235565185546875,
|
| 2503 |
+
"logps/rejected": -38.84959411621094,
|
| 2504 |
+
"loss": 0.6944494247436523,
|
| 2505 |
+
"memory(GiB)": 46.13,
|
| 2506 |
+
"nll_loss": 0.43725934624671936,
|
| 2507 |
+
"rewards/accuracies": 0.90625,
|
| 2508 |
+
"rewards/chosen": 0.12278101593255997,
|
| 2509 |
+
"rewards/margins": 2.2289106845855713,
|
| 2510 |
+
"rewards/rejected": -2.1061296463012695,
|
| 2511 |
+
"step": 139,
|
| 2512 |
+
"train_speed(iter/s)": 0.005635
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.8430560782837787,
|
| 2516 |
+
"grad_norm": 3.664316415786743,
|
| 2517 |
+
"learning_rate": 0.00013373299873828303,
|
| 2518 |
+
"logits/chosen": -0.29347366094589233,
|
| 2519 |
+
"logits/rejected": -0.41366317868232727,
|
| 2520 |
+
"logps/chosen": -4.828797340393066,
|
| 2521 |
+
"logps/rejected": -31.14942169189453,
|
| 2522 |
+
"loss": 0.6968463659286499,
|
| 2523 |
+
"memory(GiB)": 46.13,
|
| 2524 |
+
"nll_loss": 0.41966789960861206,
|
| 2525 |
+
"rewards/accuracies": 0.875,
|
| 2526 |
+
"rewards/chosen": 0.13118553161621094,
|
| 2527 |
+
"rewards/margins": 1.9206351041793823,
|
| 2528 |
+
"rewards/rejected": -1.7894494533538818,
|
| 2529 |
+
"step": 140,
|
| 2530 |
+
"train_speed(iter/s)": 0.005635
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.8490779074143772,
|
| 2534 |
+
"grad_norm": 3.614262580871582,
|
| 2535 |
+
"learning_rate": 0.00013279246275072046,
|
| 2536 |
+
"logits/chosen": -0.31295719742774963,
|
| 2537 |
+
"logits/rejected": -0.4447307288646698,
|
| 2538 |
+
"logps/chosen": -8.66226577758789,
|
| 2539 |
+
"logps/rejected": -31.42186737060547,
|
| 2540 |
+
"loss": 1.0331419706344604,
|
| 2541 |
+
"memory(GiB)": 46.13,
|
| 2542 |
+
"nll_loss": 0.6634517312049866,
|
| 2543 |
+
"rewards/accuracies": 0.78125,
|
| 2544 |
+
"rewards/chosen": 0.08492518961429596,
|
| 2545 |
+
"rewards/margins": 1.7986928224563599,
|
| 2546 |
+
"rewards/rejected": -1.713767647743225,
|
| 2547 |
+
"step": 141,
|
| 2548 |
+
"train_speed(iter/s)": 0.005636
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.8550997365449755,
|
| 2552 |
+
"grad_norm": 1.7650597095489502,
|
| 2553 |
+
"learning_rate": 0.00013184866502516845,
|
| 2554 |
+
"logits/chosen": -0.26669561862945557,
|
| 2555 |
+
"logits/rejected": -0.4871166944503784,
|
| 2556 |
+
"logps/chosen": -3.3100223541259766,
|
| 2557 |
+
"logps/rejected": -39.79899215698242,
|
| 2558 |
+
"loss": 0.5506975650787354,
|
| 2559 |
+
"memory(GiB)": 46.13,
|
| 2560 |
+
"nll_loss": 0.29588890075683594,
|
| 2561 |
+
"rewards/accuracies": 0.890625,
|
| 2562 |
+
"rewards/chosen": 0.10198746621608734,
|
| 2563 |
+
"rewards/margins": 2.259812831878662,
|
| 2564 |
+
"rewards/rejected": -2.157825231552124,
|
| 2565 |
+
"step": 142,
|
| 2566 |
+
"train_speed(iter/s)": 0.005636
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 0.861121565675574,
|
| 2570 |
+
"grad_norm": 3.4034411907196045,
|
| 2571 |
+
"learning_rate": 0.00013090169943749476,
|
| 2572 |
+
"logits/chosen": -0.3030458390712738,
|
| 2573 |
+
"logits/rejected": -0.45216381549835205,
|
| 2574 |
+
"logps/chosen": -5.469812393188477,
|
| 2575 |
+
"logps/rejected": -35.740806579589844,
|
| 2576 |
+
"loss": 0.6628592610359192,
|
| 2577 |
+
"memory(GiB)": 46.13,
|
| 2578 |
+
"nll_loss": 0.34965524077415466,
|
| 2579 |
+
"rewards/accuracies": 0.84375,
|
| 2580 |
+
"rewards/chosen": 0.07742348313331604,
|
| 2581 |
+
"rewards/margins": 2.012543201446533,
|
| 2582 |
+
"rewards/rejected": -1.935119867324829,
|
| 2583 |
+
"step": 143,
|
| 2584 |
+
"train_speed(iter/s)": 0.005636
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 0.8671433948061724,
|
| 2588 |
+
"grad_norm": 2.271817922592163,
|
| 2589 |
+
"learning_rate": 0.00012995166017866193,
|
| 2590 |
+
"logits/chosen": -0.35303038358688354,
|
| 2591 |
+
"logits/rejected": -0.4534931778907776,
|
| 2592 |
+
"logps/chosen": -7.304016590118408,
|
| 2593 |
+
"logps/rejected": -30.363569259643555,
|
| 2594 |
+
"loss": 0.8833200931549072,
|
| 2595 |
+
"memory(GiB)": 46.13,
|
| 2596 |
+
"nll_loss": 0.5442834496498108,
|
| 2597 |
+
"rewards/accuracies": 0.765625,
|
| 2598 |
+
"rewards/chosen": 0.18977457284927368,
|
| 2599 |
+
"rewards/margins": 2.04044508934021,
|
| 2600 |
+
"rewards/rejected": -1.850670576095581,
|
| 2601 |
+
"step": 144,
|
| 2602 |
+
"train_speed(iter/s)": 0.005636
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.8731652239367708,
|
| 2606 |
+
"grad_norm": 1.6546629667282104,
|
| 2607 |
+
"learning_rate": 0.00012899864174535864,
|
| 2608 |
+
"logits/chosen": -0.30710387229919434,
|
| 2609 |
+
"logits/rejected": -0.47071051597595215,
|
| 2610 |
+
"logps/chosen": -4.921065330505371,
|
| 2611 |
+
"logps/rejected": -37.83810043334961,
|
| 2612 |
+
"loss": 0.611689567565918,
|
| 2613 |
+
"memory(GiB)": 46.13,
|
| 2614 |
+
"nll_loss": 0.36449986696243286,
|
| 2615 |
+
"rewards/accuracies": 0.90625,
|
| 2616 |
+
"rewards/chosen": 0.14039897918701172,
|
| 2617 |
+
"rewards/margins": 2.364004373550415,
|
| 2618 |
+
"rewards/rejected": -2.2236056327819824,
|
| 2619 |
+
"step": 145,
|
| 2620 |
+
"train_speed(iter/s)": 0.005636
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.8791870530673692,
|
| 2624 |
+
"grad_norm": 1.4133797883987427,
|
| 2625 |
+
"learning_rate": 0.00012804273893060028,
|
| 2626 |
+
"logits/chosen": -0.35278186202049255,
|
| 2627 |
+
"logits/rejected": -0.4680350422859192,
|
| 2628 |
+
"logps/chosen": -4.455732345581055,
|
| 2629 |
+
"logps/rejected": -32.905887603759766,
|
| 2630 |
+
"loss": 0.5695258975028992,
|
| 2631 |
+
"memory(GiB)": 46.13,
|
| 2632 |
+
"nll_loss": 0.3188043534755707,
|
| 2633 |
+
"rewards/accuracies": 0.9375,
|
| 2634 |
+
"rewards/chosen": 0.18160954117774963,
|
| 2635 |
+
"rewards/margins": 2.25382137298584,
|
| 2636 |
+
"rewards/rejected": -2.072211742401123,
|
| 2637 |
+
"step": 146,
|
| 2638 |
+
"train_speed(iter/s)": 0.005636
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.8852088821979677,
|
| 2642 |
+
"grad_norm": 1.8506555557250977,
|
| 2643 |
+
"learning_rate": 0.00012708404681430053,
|
| 2644 |
+
"logits/chosen": -0.3219718635082245,
|
| 2645 |
+
"logits/rejected": -0.4345773756504059,
|
| 2646 |
+
"logps/chosen": -5.016592979431152,
|
| 2647 |
+
"logps/rejected": -34.69657516479492,
|
| 2648 |
+
"loss": 0.6405315399169922,
|
| 2649 |
+
"memory(GiB)": 46.13,
|
| 2650 |
+
"nll_loss": 0.35303816199302673,
|
| 2651 |
+
"rewards/accuracies": 0.921875,
|
| 2652 |
+
"rewards/chosen": 0.13450342416763306,
|
| 2653 |
+
"rewards/margins": 2.0750272274017334,
|
| 2654 |
+
"rewards/rejected": -1.9405235052108765,
|
| 2655 |
+
"step": 147,
|
| 2656 |
+
"train_speed(iter/s)": 0.005636
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.891230711328566,
|
| 2660 |
+
"grad_norm": 2.6582088470458984,
|
| 2661 |
+
"learning_rate": 0.00012612266075381386,
|
| 2662 |
+
"logits/chosen": -0.2835950255393982,
|
| 2663 |
+
"logits/rejected": -0.4708768129348755,
|
| 2664 |
+
"logps/chosen": -6.048505783081055,
|
| 2665 |
+
"logps/rejected": -42.608131408691406,
|
| 2666 |
+
"loss": 0.676105260848999,
|
| 2667 |
+
"memory(GiB)": 46.13,
|
| 2668 |
+
"nll_loss": 0.4015389084815979,
|
| 2669 |
+
"rewards/accuracies": 0.90625,
|
| 2670 |
+
"rewards/chosen": 0.033125825226306915,
|
| 2671 |
+
"rewards/margins": 2.4881484508514404,
|
| 2672 |
+
"rewards/rejected": -2.4550228118896484,
|
| 2673 |
+
"step": 148,
|
| 2674 |
+
"train_speed(iter/s)": 0.005636
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.8972525404591645,
|
| 2678 |
+
"grad_norm": 3.3612794876098633,
|
| 2679 |
+
"learning_rate": 0.00012515867637445086,
|
| 2680 |
+
"logits/chosen": -0.3777806758880615,
|
| 2681 |
+
"logits/rejected": -0.477069228887558,
|
| 2682 |
+
"logps/chosen": -6.0037922859191895,
|
| 2683 |
+
"logps/rejected": -32.48824691772461,
|
| 2684 |
+
"loss": 0.825937032699585,
|
| 2685 |
+
"memory(GiB)": 46.13,
|
| 2686 |
+
"nll_loss": 0.5271134972572327,
|
| 2687 |
+
"rewards/accuracies": 0.90625,
|
| 2688 |
+
"rewards/chosen": 0.07130320370197296,
|
| 2689 |
+
"rewards/margins": 2.3660011291503906,
|
| 2690 |
+
"rewards/rejected": -2.2946979999542236,
|
| 2691 |
+
"step": 149,
|
| 2692 |
+
"train_speed(iter/s)": 0.005636
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 0.9032743695897629,
|
| 2696 |
+
"grad_norm": 1.5847569704055786,
|
| 2697 |
+
"learning_rate": 0.00012419218955996676,
|
| 2698 |
+
"logits/chosen": -0.31665414571762085,
|
| 2699 |
+
"logits/rejected": -0.4995279312133789,
|
| 2700 |
+
"logps/chosen": -6.320664882659912,
|
| 2701 |
+
"logps/rejected": -52.651451110839844,
|
| 2702 |
+
"loss": 0.5624623894691467,
|
| 2703 |
+
"memory(GiB)": 46.13,
|
| 2704 |
+
"nll_loss": 0.368425577878952,
|
| 2705 |
+
"rewards/accuracies": 0.9375,
|
| 2706 |
+
"rewards/chosen": 0.14041581749916077,
|
| 2707 |
+
"rewards/margins": 3.4388084411621094,
|
| 2708 |
+
"rewards/rejected": -3.2983925342559814,
|
| 2709 |
+
"step": 150,
|
| 2710 |
+
"train_speed(iter/s)": 0.005636
|
| 2711 |
+
}
|
| 2712 |
+
],
|
| 2713 |
+
"logging_steps": 1,
|
| 2714 |
+
"max_steps": 332,
|
| 2715 |
+
"num_input_tokens_seen": 0,
|
| 2716 |
+
"num_train_epochs": 2,
|
| 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": 1.725942609131602e+17,
|
| 2731 |
+
"train_batch_size": 4,
|
| 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 sha256:5dbc87f4c3f0b72f72547a727cff5881c057678dd04a991ca554096f5933aab8
|
| 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)
|