Training in progress, step 1050, checkpoint
Browse files- checkpoint-1050/README.md +202 -0
- checkpoint-1050/adapter_config.json +34 -0
- checkpoint-1050/adapter_model.safetensors +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- checkpoint-1050/global_step1050/mp_rank_00_model_states.pt +3 -0
- checkpoint-1050/latest +1 -0
- checkpoint-1050/rng_state_0.pth +3 -0
- checkpoint-1050/rng_state_1.pth +3 -0
- checkpoint-1050/rng_state_2.pth +3 -0
- checkpoint-1050/rng_state_3.pth +3 -0
- checkpoint-1050/rng_state_4.pth +3 -0
- checkpoint-1050/rng_state_5.pth +3 -0
- checkpoint-1050/rng_state_6.pth +3 -0
- checkpoint-1050/rng_state_7.pth +3 -0
- checkpoint-1050/scheduler.pt +3 -0
- checkpoint-1050/special_tokens_map.json +30 -0
- checkpoint-1050/tokenizer.json +0 -0
- checkpoint-1050/tokenizer_config.json +133 -0
- checkpoint-1050/trainer_state.json +1944 -0
- checkpoint-1050/training_args.bin +3 -0
- checkpoint-1050/zero_to_fp32.py +674 -0
checkpoint-1050/README.md
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: microsoft/Phi-3-mini-4k-instruct
|
| 3 |
+
library_name: peft
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Model Card for Model ID
|
| 7 |
+
|
| 8 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
## Model Details
|
| 13 |
+
|
| 14 |
+
### Model Description
|
| 15 |
+
|
| 16 |
+
<!-- Provide a longer summary of what this model is. -->
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
- **Developed by:** [More Information Needed]
|
| 21 |
+
- **Funded by [optional]:** [More Information Needed]
|
| 22 |
+
- **Shared by [optional]:** [More Information Needed]
|
| 23 |
+
- **Model type:** [More Information Needed]
|
| 24 |
+
- **Language(s) (NLP):** [More Information Needed]
|
| 25 |
+
- **License:** [More Information Needed]
|
| 26 |
+
- **Finetuned from model [optional]:** [More Information Needed]
|
| 27 |
+
|
| 28 |
+
### Model Sources [optional]
|
| 29 |
+
|
| 30 |
+
<!-- Provide the basic links for the model. -->
|
| 31 |
+
|
| 32 |
+
- **Repository:** [More Information Needed]
|
| 33 |
+
- **Paper [optional]:** [More Information Needed]
|
| 34 |
+
- **Demo [optional]:** [More Information Needed]
|
| 35 |
+
|
| 36 |
+
## Uses
|
| 37 |
+
|
| 38 |
+
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
| 39 |
+
|
| 40 |
+
### Direct Use
|
| 41 |
+
|
| 42 |
+
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
| 43 |
+
|
| 44 |
+
[More Information Needed]
|
| 45 |
+
|
| 46 |
+
### Downstream Use [optional]
|
| 47 |
+
|
| 48 |
+
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
| 49 |
+
|
| 50 |
+
[More Information Needed]
|
| 51 |
+
|
| 52 |
+
### Out-of-Scope Use
|
| 53 |
+
|
| 54 |
+
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
| 55 |
+
|
| 56 |
+
[More Information Needed]
|
| 57 |
+
|
| 58 |
+
## Bias, Risks, and Limitations
|
| 59 |
+
|
| 60 |
+
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
| 61 |
+
|
| 62 |
+
[More Information Needed]
|
| 63 |
+
|
| 64 |
+
### Recommendations
|
| 65 |
+
|
| 66 |
+
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
| 67 |
+
|
| 68 |
+
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
| 69 |
+
|
| 70 |
+
## How to Get Started with the Model
|
| 71 |
+
|
| 72 |
+
Use the code below to get started with the model.
|
| 73 |
+
|
| 74 |
+
[More Information Needed]
|
| 75 |
+
|
| 76 |
+
## Training Details
|
| 77 |
+
|
| 78 |
+
### Training Data
|
| 79 |
+
|
| 80 |
+
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 81 |
+
|
| 82 |
+
[More Information Needed]
|
| 83 |
+
|
| 84 |
+
### Training Procedure
|
| 85 |
+
|
| 86 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 87 |
+
|
| 88 |
+
#### Preprocessing [optional]
|
| 89 |
+
|
| 90 |
+
[More Information Needed]
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
#### Training Hyperparameters
|
| 94 |
+
|
| 95 |
+
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 96 |
+
|
| 97 |
+
#### Speeds, Sizes, Times [optional]
|
| 98 |
+
|
| 99 |
+
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 100 |
+
|
| 101 |
+
[More Information Needed]
|
| 102 |
+
|
| 103 |
+
## Evaluation
|
| 104 |
+
|
| 105 |
+
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 106 |
+
|
| 107 |
+
### Testing Data, Factors & Metrics
|
| 108 |
+
|
| 109 |
+
#### Testing Data
|
| 110 |
+
|
| 111 |
+
<!-- This should link to a Dataset Card if possible. -->
|
| 112 |
+
|
| 113 |
+
[More Information Needed]
|
| 114 |
+
|
| 115 |
+
#### Factors
|
| 116 |
+
|
| 117 |
+
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 118 |
+
|
| 119 |
+
[More Information Needed]
|
| 120 |
+
|
| 121 |
+
#### Metrics
|
| 122 |
+
|
| 123 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 124 |
+
|
| 125 |
+
[More Information Needed]
|
| 126 |
+
|
| 127 |
+
### Results
|
| 128 |
+
|
| 129 |
+
[More Information Needed]
|
| 130 |
+
|
| 131 |
+
#### Summary
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
|
| 135 |
+
## Model Examination [optional]
|
| 136 |
+
|
| 137 |
+
<!-- Relevant interpretability work for the model goes here -->
|
| 138 |
+
|
| 139 |
+
[More Information Needed]
|
| 140 |
+
|
| 141 |
+
## Environmental Impact
|
| 142 |
+
|
| 143 |
+
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
| 144 |
+
|
| 145 |
+
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
| 146 |
+
|
| 147 |
+
- **Hardware Type:** [More Information Needed]
|
| 148 |
+
- **Hours used:** [More Information Needed]
|
| 149 |
+
- **Cloud Provider:** [More Information Needed]
|
| 150 |
+
- **Compute Region:** [More Information Needed]
|
| 151 |
+
- **Carbon Emitted:** [More Information Needed]
|
| 152 |
+
|
| 153 |
+
## Technical Specifications [optional]
|
| 154 |
+
|
| 155 |
+
### Model Architecture and Objective
|
| 156 |
+
|
| 157 |
+
[More Information Needed]
|
| 158 |
+
|
| 159 |
+
### Compute Infrastructure
|
| 160 |
+
|
| 161 |
+
[More Information Needed]
|
| 162 |
+
|
| 163 |
+
#### Hardware
|
| 164 |
+
|
| 165 |
+
[More Information Needed]
|
| 166 |
+
|
| 167 |
+
#### Software
|
| 168 |
+
|
| 169 |
+
[More Information Needed]
|
| 170 |
+
|
| 171 |
+
## Citation [optional]
|
| 172 |
+
|
| 173 |
+
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
| 174 |
+
|
| 175 |
+
**BibTeX:**
|
| 176 |
+
|
| 177 |
+
[More Information Needed]
|
| 178 |
+
|
| 179 |
+
**APA:**
|
| 180 |
+
|
| 181 |
+
[More Information Needed]
|
| 182 |
+
|
| 183 |
+
## Glossary [optional]
|
| 184 |
+
|
| 185 |
+
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
| 186 |
+
|
| 187 |
+
[More Information Needed]
|
| 188 |
+
|
| 189 |
+
## More Information [optional]
|
| 190 |
+
|
| 191 |
+
[More Information Needed]
|
| 192 |
+
|
| 193 |
+
## Model Card Authors [optional]
|
| 194 |
+
|
| 195 |
+
[More Information Needed]
|
| 196 |
+
|
| 197 |
+
## Model Card Contact
|
| 198 |
+
|
| 199 |
+
[More Information Needed]
|
| 200 |
+
### Framework versions
|
| 201 |
+
|
| 202 |
+
- PEFT 0.14.0
|
checkpoint-1050/adapter_config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "microsoft/Phi-3-mini-4k-instruct",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"eva_config": null,
|
| 7 |
+
"exclude_modules": null,
|
| 8 |
+
"fan_in_fan_out": false,
|
| 9 |
+
"inference_mode": true,
|
| 10 |
+
"init_lora_weights": true,
|
| 11 |
+
"layer_replication": null,
|
| 12 |
+
"layers_pattern": null,
|
| 13 |
+
"layers_to_transform": null,
|
| 14 |
+
"loftq_config": {},
|
| 15 |
+
"lora_alpha": 16,
|
| 16 |
+
"lora_bias": false,
|
| 17 |
+
"lora_dropout": 0.0,
|
| 18 |
+
"megatron_config": null,
|
| 19 |
+
"megatron_core": "megatron.core",
|
| 20 |
+
"modules_to_save": null,
|
| 21 |
+
"peft_type": "LORA",
|
| 22 |
+
"r": 8,
|
| 23 |
+
"rank_pattern": {},
|
| 24 |
+
"revision": null,
|
| 25 |
+
"target_modules": [
|
| 26 |
+
"down_proj",
|
| 27 |
+
"gate_up_proj",
|
| 28 |
+
"o_proj",
|
| 29 |
+
"qkv_proj"
|
| 30 |
+
],
|
| 31 |
+
"task_type": "CAUSAL_LM",
|
| 32 |
+
"use_dora": false,
|
| 33 |
+
"use_rslora": false
|
| 34 |
+
}
|
checkpoint-1050/adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:707f69ec53b9e82651ba33bfd4ca7084f98cea387d2651fa7556d406e84d1845
|
| 3 |
+
size 25200088
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:090a660bbd269fab6fa6bc1fe8846098bdf757d103ebf0fdb0420e4cac250339
|
| 3 |
+
size 18881328
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc252a3730fa3d14d29dd85c87329a96b96c645a6a88c22e76997f3ed014f7c5
|
| 3 |
+
size 18881328
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe6f3f8f28341ec03eabae3da8eb0bbfed6b0efb86431c7b902a218a82c04031
|
| 3 |
+
size 18881328
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2b611885e22ae5a5ab16626073efcc635aedd82820865fe09b68dc7c3fb2801
|
| 3 |
+
size 18881392
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f57aa491452b8bea87bc8f878719cdf1300c8c657ab336c4de3482c781ff7098
|
| 3 |
+
size 18881392
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7dade57fa0cbd10e3e9c80390b132482c46604f7d6e7d49310da296312bc4a1c
|
| 3 |
+
size 18881392
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91c12fa400a96f3eceaa3c868cca352dda0eb4a5d69dfdf707929477463df63a
|
| 3 |
+
size 18881392
|
checkpoint-1050/global_step1050/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fc84c87f9b2b9bcffa317acecd30613ba3a04be41688926da6f890156b391b9
|
| 3 |
+
size 18881392
|
checkpoint-1050/global_step1050/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:32c2b67d00bccb9049fcfad4394b282403a689a5880e105afb69dfc9b21f8470
|
| 3 |
+
size 25379244
|
checkpoint-1050/latest
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
global_step1050
|
checkpoint-1050/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2509a0f61a0c413c809bd0b1cfe15a6a6c54b8866dbf91355c7feff001a85f29
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd1967f364c9e8a4f272ac615805effb60160d44b5f950f10ece423066d39b97
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1436be35db72cc7fe06713671c8fc91397641d49a2e6c8d9f5fe86ec7df69fa4
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:560a974f1bb0a75400e22c0bdde337d37191ae05522726074577068218c38fef
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1bd5c9b731e7462b56487a4584fff0a7ff50f156c4351dc695e65917e90ef3de
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cf3b425daa225f516388e50945878ebc7b4050c284ce140f112bea6b828acfd
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9cfd223fbcfaa02ba082595f3292bd5854d4d21cf372043c8587c359ca9233c4
|
| 3 |
+
size 15984
|
checkpoint-1050/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a272f8d79e0f30cdb3560db8006b56ed0e2a66e427b3739f82dd7001b8aa8d8f
|
| 3 |
+
size 15984
|
checkpoint-1050/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0b74a81ec3c311e33804cf38f6e78705408f2eb4b88ea83dce3ccb641eecb37
|
| 3 |
+
size 1064
|
checkpoint-1050/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "<|end|>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "<|endoftext|>",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "<unk>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
checkpoint-1050/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1050/tokenizer_config.json
ADDED
|
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": true,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": false
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "<|endoftext|>",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
},
|
| 38 |
+
"32001": {
|
| 39 |
+
"content": "<|assistant|>",
|
| 40 |
+
"lstrip": false,
|
| 41 |
+
"normalized": false,
|
| 42 |
+
"rstrip": true,
|
| 43 |
+
"single_word": false,
|
| 44 |
+
"special": true
|
| 45 |
+
},
|
| 46 |
+
"32002": {
|
| 47 |
+
"content": "<|placeholder1|>",
|
| 48 |
+
"lstrip": false,
|
| 49 |
+
"normalized": false,
|
| 50 |
+
"rstrip": true,
|
| 51 |
+
"single_word": false,
|
| 52 |
+
"special": true
|
| 53 |
+
},
|
| 54 |
+
"32003": {
|
| 55 |
+
"content": "<|placeholder2|>",
|
| 56 |
+
"lstrip": false,
|
| 57 |
+
"normalized": false,
|
| 58 |
+
"rstrip": true,
|
| 59 |
+
"single_word": false,
|
| 60 |
+
"special": true
|
| 61 |
+
},
|
| 62 |
+
"32004": {
|
| 63 |
+
"content": "<|placeholder3|>",
|
| 64 |
+
"lstrip": false,
|
| 65 |
+
"normalized": false,
|
| 66 |
+
"rstrip": true,
|
| 67 |
+
"single_word": false,
|
| 68 |
+
"special": true
|
| 69 |
+
},
|
| 70 |
+
"32005": {
|
| 71 |
+
"content": "<|placeholder4|>",
|
| 72 |
+
"lstrip": false,
|
| 73 |
+
"normalized": false,
|
| 74 |
+
"rstrip": true,
|
| 75 |
+
"single_word": false,
|
| 76 |
+
"special": true
|
| 77 |
+
},
|
| 78 |
+
"32006": {
|
| 79 |
+
"content": "<|system|>",
|
| 80 |
+
"lstrip": false,
|
| 81 |
+
"normalized": false,
|
| 82 |
+
"rstrip": true,
|
| 83 |
+
"single_word": false,
|
| 84 |
+
"special": true
|
| 85 |
+
},
|
| 86 |
+
"32007": {
|
| 87 |
+
"content": "<|end|>",
|
| 88 |
+
"lstrip": false,
|
| 89 |
+
"normalized": false,
|
| 90 |
+
"rstrip": false,
|
| 91 |
+
"single_word": false,
|
| 92 |
+
"special": true
|
| 93 |
+
},
|
| 94 |
+
"32008": {
|
| 95 |
+
"content": "<|placeholder5|>",
|
| 96 |
+
"lstrip": false,
|
| 97 |
+
"normalized": false,
|
| 98 |
+
"rstrip": true,
|
| 99 |
+
"single_word": false,
|
| 100 |
+
"special": true
|
| 101 |
+
},
|
| 102 |
+
"32009": {
|
| 103 |
+
"content": "<|placeholder6|>",
|
| 104 |
+
"lstrip": false,
|
| 105 |
+
"normalized": false,
|
| 106 |
+
"rstrip": true,
|
| 107 |
+
"single_word": false,
|
| 108 |
+
"special": true
|
| 109 |
+
},
|
| 110 |
+
"32010": {
|
| 111 |
+
"content": "<|user|>",
|
| 112 |
+
"lstrip": false,
|
| 113 |
+
"normalized": false,
|
| 114 |
+
"rstrip": true,
|
| 115 |
+
"single_word": false,
|
| 116 |
+
"special": true
|
| 117 |
+
}
|
| 118 |
+
},
|
| 119 |
+
"bos_token": "<s>",
|
| 120 |
+
"chat_template": "{% set system_message = 'You are a helpful AI assistant.' %}{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '<s>' + '<|system|>\n' + system_message + '<|end|>\n' }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|user|>\n' + content + '<|end|>\n<|assistant|>\n' }}{% elif message['role'] == 'assistant' %}{{ content + '<|end|>' + '\n' }}{% endif %}{% endfor %}",
|
| 121 |
+
"clean_up_tokenization_spaces": false,
|
| 122 |
+
"eos_token": "<|end|>",
|
| 123 |
+
"extra_special_tokens": {},
|
| 124 |
+
"legacy": false,
|
| 125 |
+
"model_max_length": 4096,
|
| 126 |
+
"pad_token": "<|endoftext|>",
|
| 127 |
+
"padding_side": "right",
|
| 128 |
+
"sp_model_kwargs": {},
|
| 129 |
+
"split_special_tokens": false,
|
| 130 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 131 |
+
"unk_token": "<unk>",
|
| 132 |
+
"use_default_system_prompt": false
|
| 133 |
+
}
|
checkpoint-1050/trainer_state.json
ADDED
|
@@ -0,0 +1,1944 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8741804558226662,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 1050,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00832552815069206,
|
| 13 |
+
"grad_norm": 0.04514288529753685,
|
| 14 |
+
"learning_rate": 4.999451708687114e-06,
|
| 15 |
+
"logits/chosen": 14.412135124206543,
|
| 16 |
+
"logits/rejected": 14.867518424987793,
|
| 17 |
+
"logps/chosen": -0.29279541969299316,
|
| 18 |
+
"logps/rejected": -0.33705300092697144,
|
| 19 |
+
"loss": 0.9248,
|
| 20 |
+
"rewards/accuracies": 0.512499988079071,
|
| 21 |
+
"rewards/chosen": -0.43919315934181213,
|
| 22 |
+
"rewards/margins": 0.066386379301548,
|
| 23 |
+
"rewards/rejected": -0.5055795311927795,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.01665105630138412,
|
| 28 |
+
"grad_norm": 0.05052826926112175,
|
| 29 |
+
"learning_rate": 4.997807075247147e-06,
|
| 30 |
+
"logits/chosen": 14.956459045410156,
|
| 31 |
+
"logits/rejected": 15.363263130187988,
|
| 32 |
+
"logps/chosen": -0.3096744120121002,
|
| 33 |
+
"logps/rejected": -0.36214715242385864,
|
| 34 |
+
"loss": 0.9355,
|
| 35 |
+
"rewards/accuracies": 0.5,
|
| 36 |
+
"rewards/chosen": -0.46451157331466675,
|
| 37 |
+
"rewards/margins": 0.07870914041996002,
|
| 38 |
+
"rewards/rejected": -0.5432207584381104,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.024976584452076178,
|
| 43 |
+
"grad_norm": 0.04879612475633621,
|
| 44 |
+
"learning_rate": 4.9950668210706795e-06,
|
| 45 |
+
"logits/chosen": 14.485757827758789,
|
| 46 |
+
"logits/rejected": 15.057507514953613,
|
| 47 |
+
"logps/chosen": -0.27136802673339844,
|
| 48 |
+
"logps/rejected": -0.31497400999069214,
|
| 49 |
+
"loss": 0.9268,
|
| 50 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 51 |
+
"rewards/chosen": -0.4070519804954529,
|
| 52 |
+
"rewards/margins": 0.06540900468826294,
|
| 53 |
+
"rewards/rejected": -0.4724610447883606,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.03330211260276824,
|
| 58 |
+
"grad_norm": 0.05672155320644379,
|
| 59 |
+
"learning_rate": 4.9912321481237616e-06,
|
| 60 |
+
"logits/chosen": 14.529332160949707,
|
| 61 |
+
"logits/rejected": 14.814855575561523,
|
| 62 |
+
"logps/chosen": -0.29139184951782227,
|
| 63 |
+
"logps/rejected": -0.31259119510650635,
|
| 64 |
+
"loss": 0.9267,
|
| 65 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 66 |
+
"rewards/chosen": -0.4370877742767334,
|
| 67 |
+
"rewards/margins": 0.03179898113012314,
|
| 68 |
+
"rewards/rejected": -0.46888676285743713,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.041627640753460295,
|
| 73 |
+
"grad_norm": 0.065071240067482,
|
| 74 |
+
"learning_rate": 4.986304738420684e-06,
|
| 75 |
+
"logits/chosen": 14.174386978149414,
|
| 76 |
+
"logits/rejected": 15.223234176635742,
|
| 77 |
+
"logps/chosen": -0.2745029330253601,
|
| 78 |
+
"logps/rejected": -0.37693315744400024,
|
| 79 |
+
"loss": 0.9243,
|
| 80 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 81 |
+
"rewards/chosen": -0.41175442934036255,
|
| 82 |
+
"rewards/margins": 0.1536453813314438,
|
| 83 |
+
"rewards/rejected": -0.5653998255729675,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.041627640753460295,
|
| 88 |
+
"eval_logits/chosen": 14.56569766998291,
|
| 89 |
+
"eval_logits/rejected": 15.157320976257324,
|
| 90 |
+
"eval_logps/chosen": -0.27527979016304016,
|
| 91 |
+
"eval_logps/rejected": -0.3633999824523926,
|
| 92 |
+
"eval_loss": 0.9083622694015503,
|
| 93 |
+
"eval_rewards/accuracies": 0.5612244606018066,
|
| 94 |
+
"eval_rewards/chosen": -0.41291970014572144,
|
| 95 |
+
"eval_rewards/margins": 0.13218028843402863,
|
| 96 |
+
"eval_rewards/rejected": -0.5450999736785889,
|
| 97 |
+
"eval_runtime": 29.029,
|
| 98 |
+
"eval_samples_per_second": 26.766,
|
| 99 |
+
"eval_steps_per_second": 3.376,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.049953168904152356,
|
| 104 |
+
"grad_norm": 0.14002270996570587,
|
| 105 |
+
"learning_rate": 4.980286753286196e-06,
|
| 106 |
+
"logits/chosen": 14.408930778503418,
|
| 107 |
+
"logits/rejected": 14.791458129882812,
|
| 108 |
+
"logps/chosen": -0.285602867603302,
|
| 109 |
+
"logps/rejected": -0.3351826071739197,
|
| 110 |
+
"loss": 0.9177,
|
| 111 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 112 |
+
"rewards/chosen": -0.4284043312072754,
|
| 113 |
+
"rewards/margins": 0.07436960190534592,
|
| 114 |
+
"rewards/rejected": -0.5027738809585571,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.05827869705484442,
|
| 119 |
+
"grad_norm": 0.05595069006085396,
|
| 120 |
+
"learning_rate": 4.973180832407471e-06,
|
| 121 |
+
"logits/chosen": 14.41168212890625,
|
| 122 |
+
"logits/rejected": 14.865121841430664,
|
| 123 |
+
"logps/chosen": -0.25851207971572876,
|
| 124 |
+
"logps/rejected": -0.32240185141563416,
|
| 125 |
+
"loss": 0.9168,
|
| 126 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 127 |
+
"rewards/chosen": -0.3877681493759155,
|
| 128 |
+
"rewards/margins": 0.0958346277475357,
|
| 129 |
+
"rewards/rejected": -0.4836028218269348,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.06660422520553648,
|
| 134 |
+
"grad_norm": 0.058645494282245636,
|
| 135 |
+
"learning_rate": 4.964990092676263e-06,
|
| 136 |
+
"logits/chosen": 14.897825241088867,
|
| 137 |
+
"logits/rejected": 15.01073932647705,
|
| 138 |
+
"logps/chosen": -0.2668797969818115,
|
| 139 |
+
"logps/rejected": -0.3204379975795746,
|
| 140 |
+
"loss": 0.9242,
|
| 141 |
+
"rewards/accuracies": 0.5,
|
| 142 |
+
"rewards/chosen": -0.4003197252750397,
|
| 143 |
+
"rewards/margins": 0.08033724129199982,
|
| 144 |
+
"rewards/rejected": -0.4806569516658783,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.07492975335622853,
|
| 149 |
+
"grad_norm": 0.0597861111164093,
|
| 150 |
+
"learning_rate": 4.9557181268217225e-06,
|
| 151 |
+
"logits/chosen": 14.531021118164062,
|
| 152 |
+
"logits/rejected": 14.767858505249023,
|
| 153 |
+
"logps/chosen": -0.26787540316581726,
|
| 154 |
+
"logps/rejected": -0.32972821593284607,
|
| 155 |
+
"loss": 0.9077,
|
| 156 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 157 |
+
"rewards/chosen": -0.4018131196498871,
|
| 158 |
+
"rewards/margins": 0.09277921915054321,
|
| 159 |
+
"rewards/rejected": -0.4945923686027527,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.08325528150692059,
|
| 164 |
+
"grad_norm": 0.0863095372915268,
|
| 165 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 166 |
+
"logits/chosen": 14.179275512695312,
|
| 167 |
+
"logits/rejected": 14.909070014953613,
|
| 168 |
+
"logps/chosen": -0.2532978057861328,
|
| 169 |
+
"logps/rejected": -0.35474082827568054,
|
| 170 |
+
"loss": 0.903,
|
| 171 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 172 |
+
"rewards/chosen": -0.3799467086791992,
|
| 173 |
+
"rewards/margins": 0.1521645337343216,
|
| 174 |
+
"rewards/rejected": -0.5321112275123596,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.08325528150692059,
|
| 179 |
+
"eval_logits/chosen": 14.326024055480957,
|
| 180 |
+
"eval_logits/rejected": 14.979863166809082,
|
| 181 |
+
"eval_logps/chosen": -0.2673422694206238,
|
| 182 |
+
"eval_logps/rejected": -0.3668619990348816,
|
| 183 |
+
"eval_loss": 0.8989922404289246,
|
| 184 |
+
"eval_rewards/accuracies": 0.6020408272743225,
|
| 185 |
+
"eval_rewards/chosen": -0.4010133445262909,
|
| 186 |
+
"eval_rewards/margins": 0.1492796391248703,
|
| 187 |
+
"eval_rewards/rejected": -0.5502930283546448,
|
| 188 |
+
"eval_runtime": 29.0209,
|
| 189 |
+
"eval_samples_per_second": 26.774,
|
| 190 |
+
"eval_steps_per_second": 3.377,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.09158080965761266,
|
| 195 |
+
"grad_norm": 0.07181967049837112,
|
| 196 |
+
"learning_rate": 4.933947257182901e-06,
|
| 197 |
+
"logits/chosen": 14.118756294250488,
|
| 198 |
+
"logits/rejected": 14.755918502807617,
|
| 199 |
+
"logps/chosen": -0.27995947003364563,
|
| 200 |
+
"logps/rejected": -0.3749552369117737,
|
| 201 |
+
"loss": 0.9097,
|
| 202 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 203 |
+
"rewards/chosen": -0.41993919014930725,
|
| 204 |
+
"rewards/margins": 0.14249366521835327,
|
| 205 |
+
"rewards/rejected": -0.5624328255653381,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.09990633780830471,
|
| 210 |
+
"grad_norm": 0.08269819617271423,
|
| 211 |
+
"learning_rate": 4.921457902821578e-06,
|
| 212 |
+
"logits/chosen": 13.764413833618164,
|
| 213 |
+
"logits/rejected": 14.43315315246582,
|
| 214 |
+
"logps/chosen": -0.28177163004875183,
|
| 215 |
+
"logps/rejected": -0.3637630343437195,
|
| 216 |
+
"loss": 0.9075,
|
| 217 |
+
"rewards/accuracies": 0.625,
|
| 218 |
+
"rewards/chosen": -0.42265743017196655,
|
| 219 |
+
"rewards/margins": 0.12298711389303207,
|
| 220 |
+
"rewards/rejected": -0.5456445813179016,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.10823186595899677,
|
| 225 |
+
"grad_norm": 1.9071497917175293,
|
| 226 |
+
"learning_rate": 4.907906416994146e-06,
|
| 227 |
+
"logits/chosen": 14.103793144226074,
|
| 228 |
+
"logits/rejected": 14.727777481079102,
|
| 229 |
+
"logps/chosen": -0.2665451765060425,
|
| 230 |
+
"logps/rejected": -0.3827117085456848,
|
| 231 |
+
"loss": 0.9217,
|
| 232 |
+
"rewards/accuracies": 0.574999988079071,
|
| 233 |
+
"rewards/chosen": -0.3998177647590637,
|
| 234 |
+
"rewards/margins": 0.1742497682571411,
|
| 235 |
+
"rewards/rejected": -0.5740675926208496,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.11655739410968884,
|
| 240 |
+
"grad_norm": 0.12107716500759125,
|
| 241 |
+
"learning_rate": 4.893298743830168e-06,
|
| 242 |
+
"logits/chosen": 13.517863273620605,
|
| 243 |
+
"logits/rejected": 14.42052173614502,
|
| 244 |
+
"logps/chosen": -0.26627904176712036,
|
| 245 |
+
"logps/rejected": -0.3745174705982208,
|
| 246 |
+
"loss": 0.904,
|
| 247 |
+
"rewards/accuracies": 0.5625,
|
| 248 |
+
"rewards/chosen": -0.39941853284835815,
|
| 249 |
+
"rewards/margins": 0.16235767304897308,
|
| 250 |
+
"rewards/rejected": -0.5617762207984924,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.12488292226038089,
|
| 255 |
+
"grad_norm": 0.1638205647468567,
|
| 256 |
+
"learning_rate": 4.8776412907378845e-06,
|
| 257 |
+
"logits/chosen": 12.83032512664795,
|
| 258 |
+
"logits/rejected": 13.673515319824219,
|
| 259 |
+
"logps/chosen": -0.24289576709270477,
|
| 260 |
+
"logps/rejected": -0.37163227796554565,
|
| 261 |
+
"loss": 0.8779,
|
| 262 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 263 |
+
"rewards/chosen": -0.36434367299079895,
|
| 264 |
+
"rewards/margins": 0.19310477375984192,
|
| 265 |
+
"rewards/rejected": -0.5574483871459961,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.12488292226038089,
|
| 270 |
+
"eval_logits/chosen": 12.317696571350098,
|
| 271 |
+
"eval_logits/rejected": 13.164616584777832,
|
| 272 |
+
"eval_logps/chosen": -0.266156405210495,
|
| 273 |
+
"eval_logps/rejected": -0.4009220004081726,
|
| 274 |
+
"eval_loss": 0.8768696784973145,
|
| 275 |
+
"eval_rewards/accuracies": 0.6224489808082581,
|
| 276 |
+
"eval_rewards/chosen": -0.3992346227169037,
|
| 277 |
+
"eval_rewards/margins": 0.20214837789535522,
|
| 278 |
+
"eval_rewards/rejected": -0.6013829708099365,
|
| 279 |
+
"eval_runtime": 29.0257,
|
| 280 |
+
"eval_samples_per_second": 26.769,
|
| 281 |
+
"eval_steps_per_second": 3.376,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.13320845041107296,
|
| 286 |
+
"grad_norm": 0.1479438841342926,
|
| 287 |
+
"learning_rate": 4.860940925593703e-06,
|
| 288 |
+
"logits/chosen": 12.736433029174805,
|
| 289 |
+
"logits/rejected": 13.475964546203613,
|
| 290 |
+
"logps/chosen": -0.2913517355918884,
|
| 291 |
+
"logps/rejected": -0.36094629764556885,
|
| 292 |
+
"loss": 0.8756,
|
| 293 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 294 |
+
"rewards/chosen": -0.43702763319015503,
|
| 295 |
+
"rewards/margins": 0.10439182817935944,
|
| 296 |
+
"rewards/rejected": -0.5414193868637085,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.141533978561765,
|
| 301 |
+
"grad_norm": 0.17609630525112152,
|
| 302 |
+
"learning_rate": 4.84320497372973e-06,
|
| 303 |
+
"logits/chosen": 10.606362342834473,
|
| 304 |
+
"logits/rejected": 11.537567138671875,
|
| 305 |
+
"logps/chosen": -0.2560296952724457,
|
| 306 |
+
"logps/rejected": -0.4312233328819275,
|
| 307 |
+
"loss": 0.8489,
|
| 308 |
+
"rewards/accuracies": 0.699999988079071,
|
| 309 |
+
"rewards/chosen": -0.38404449820518494,
|
| 310 |
+
"rewards/margins": 0.2627905011177063,
|
| 311 |
+
"rewards/rejected": -0.6468349695205688,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.14985950671245707,
|
| 316 |
+
"grad_norm": 0.18054936826229095,
|
| 317 |
+
"learning_rate": 4.824441214720629e-06,
|
| 318 |
+
"logits/chosen": 10.13754653930664,
|
| 319 |
+
"logits/rejected": 10.914222717285156,
|
| 320 |
+
"logps/chosen": -0.29278701543807983,
|
| 321 |
+
"logps/rejected": -0.43448886275291443,
|
| 322 |
+
"loss": 0.8715,
|
| 323 |
+
"rewards/accuracies": 0.625,
|
| 324 |
+
"rewards/chosen": -0.43918052315711975,
|
| 325 |
+
"rewards/margins": 0.21255281567573547,
|
| 326 |
+
"rewards/rejected": -0.6517333388328552,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.15818503486314914,
|
| 331 |
+
"grad_norm": 0.19739146530628204,
|
| 332 |
+
"learning_rate": 4.804657878971252e-06,
|
| 333 |
+
"logits/chosen": 8.077766418457031,
|
| 334 |
+
"logits/rejected": 9.669368743896484,
|
| 335 |
+
"logps/chosen": -0.2844889760017395,
|
| 336 |
+
"logps/rejected": -0.5050357580184937,
|
| 337 |
+
"loss": 0.8582,
|
| 338 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 339 |
+
"rewards/chosen": -0.42673349380493164,
|
| 340 |
+
"rewards/margins": 0.3308201730251312,
|
| 341 |
+
"rewards/rejected": -0.7575536966323853,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.16651056301384118,
|
| 346 |
+
"grad_norm": 0.2397814244031906,
|
| 347 |
+
"learning_rate": 4.783863644106502e-06,
|
| 348 |
+
"logits/chosen": 6.790783882141113,
|
| 349 |
+
"logits/rejected": 7.849525451660156,
|
| 350 |
+
"logps/chosen": -0.2940555512905121,
|
| 351 |
+
"logps/rejected": -0.5699166059494019,
|
| 352 |
+
"loss": 0.8196,
|
| 353 |
+
"rewards/accuracies": 0.75,
|
| 354 |
+
"rewards/chosen": -0.4410833418369293,
|
| 355 |
+
"rewards/margins": 0.41379159688949585,
|
| 356 |
+
"rewards/rejected": -0.8548749089241028,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.16651056301384118,
|
| 361 |
+
"eval_logits/chosen": 6.290835857391357,
|
| 362 |
+
"eval_logits/rejected": 6.757873058319092,
|
| 363 |
+
"eval_logps/chosen": -0.317629337310791,
|
| 364 |
+
"eval_logps/rejected": -0.581989586353302,
|
| 365 |
+
"eval_loss": 0.8032433986663818,
|
| 366 |
+
"eval_rewards/accuracies": 0.6734693646430969,
|
| 367 |
+
"eval_rewards/chosen": -0.47644397616386414,
|
| 368 |
+
"eval_rewards/margins": 0.39654040336608887,
|
| 369 |
+
"eval_rewards/rejected": -0.8729843497276306,
|
| 370 |
+
"eval_runtime": 29.025,
|
| 371 |
+
"eval_samples_per_second": 26.77,
|
| 372 |
+
"eval_steps_per_second": 3.376,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.17483609116453325,
|
| 377 |
+
"grad_norm": 0.2858545184135437,
|
| 378 |
+
"learning_rate": 4.762067631165049e-06,
|
| 379 |
+
"logits/chosen": 6.875879764556885,
|
| 380 |
+
"logits/rejected": 6.691536903381348,
|
| 381 |
+
"logps/chosen": -0.37194910645484924,
|
| 382 |
+
"logps/rejected": -0.5639354586601257,
|
| 383 |
+
"loss": 0.8129,
|
| 384 |
+
"rewards/accuracies": 0.5625,
|
| 385 |
+
"rewards/chosen": -0.5579236745834351,
|
| 386 |
+
"rewards/margins": 0.2879795432090759,
|
| 387 |
+
"rewards/rejected": -0.8459032773971558,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.18316161931522532,
|
| 392 |
+
"grad_norm": 0.30206382274627686,
|
| 393 |
+
"learning_rate": 4.7392794005985324e-06,
|
| 394 |
+
"logits/chosen": 4.656112194061279,
|
| 395 |
+
"logits/rejected": 4.483086585998535,
|
| 396 |
+
"logps/chosen": -0.360150009393692,
|
| 397 |
+
"logps/rejected": -0.6204283833503723,
|
| 398 |
+
"loss": 0.7954,
|
| 399 |
+
"rewards/accuracies": 0.612500011920929,
|
| 400 |
+
"rewards/chosen": -0.5402250289916992,
|
| 401 |
+
"rewards/margins": 0.39041754603385925,
|
| 402 |
+
"rewards/rejected": -0.9306427240371704,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.19148714746591736,
|
| 407 |
+
"grad_norm": 0.40204310417175293,
|
| 408 |
+
"learning_rate": 4.715508948078037e-06,
|
| 409 |
+
"logits/chosen": 3.9398162364959717,
|
| 410 |
+
"logits/rejected": 3.38537859916687,
|
| 411 |
+
"logps/chosen": -0.39010342955589294,
|
| 412 |
+
"logps/rejected": -0.7167688608169556,
|
| 413 |
+
"loss": 0.7664,
|
| 414 |
+
"rewards/accuracies": 0.6875,
|
| 415 |
+
"rewards/chosen": -0.5851551294326782,
|
| 416 |
+
"rewards/margins": 0.4899981617927551,
|
| 417 |
+
"rewards/rejected": -1.0751533508300781,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.19981267561660943,
|
| 422 |
+
"grad_norm": 0.48389795422554016,
|
| 423 |
+
"learning_rate": 4.690766700109659e-06,
|
| 424 |
+
"logits/chosen": 2.925476551055908,
|
| 425 |
+
"logits/rejected": 2.824068069458008,
|
| 426 |
+
"logps/chosen": -0.41053348779678345,
|
| 427 |
+
"logps/rejected": -0.8508625030517578,
|
| 428 |
+
"loss": 0.7606,
|
| 429 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 430 |
+
"rewards/chosen": -0.6158002018928528,
|
| 431 |
+
"rewards/margins": 0.6604936718940735,
|
| 432 |
+
"rewards/rejected": -1.2762939929962158,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.2081382037673015,
|
| 437 |
+
"grad_norm": 0.6687452793121338,
|
| 438 |
+
"learning_rate": 4.665063509461098e-06,
|
| 439 |
+
"logits/chosen": 2.751737594604492,
|
| 440 |
+
"logits/rejected": 2.2424545288085938,
|
| 441 |
+
"logps/chosen": -0.4365699291229248,
|
| 442 |
+
"logps/rejected": -0.8550359606742859,
|
| 443 |
+
"loss": 0.7234,
|
| 444 |
+
"rewards/accuracies": 0.637499988079071,
|
| 445 |
+
"rewards/chosen": -0.6548548936843872,
|
| 446 |
+
"rewards/margins": 0.6276990175247192,
|
| 447 |
+
"rewards/rejected": -1.2825539112091064,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.2081382037673015,
|
| 452 |
+
"eval_logits/chosen": 2.1380228996276855,
|
| 453 |
+
"eval_logits/rejected": 1.3922746181488037,
|
| 454 |
+
"eval_logps/chosen": -0.48307570815086365,
|
| 455 |
+
"eval_logps/rejected": -1.0382359027862549,
|
| 456 |
+
"eval_loss": 0.668463945388794,
|
| 457 |
+
"eval_rewards/accuracies": 0.6938775777816772,
|
| 458 |
+
"eval_rewards/chosen": -0.7246134877204895,
|
| 459 |
+
"eval_rewards/margins": 0.8327403664588928,
|
| 460 |
+
"eval_rewards/rejected": -1.5573538541793823,
|
| 461 |
+
"eval_runtime": 29.0228,
|
| 462 |
+
"eval_samples_per_second": 26.772,
|
| 463 |
+
"eval_steps_per_second": 3.377,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.21646373191799353,
|
| 468 |
+
"grad_norm": 0.7085956335067749,
|
| 469 |
+
"learning_rate": 4.638410650401267e-06,
|
| 470 |
+
"logits/chosen": 1.7889283895492554,
|
| 471 |
+
"logits/rejected": 0.9420136213302612,
|
| 472 |
+
"logps/chosen": -0.5195389986038208,
|
| 473 |
+
"logps/rejected": -1.0534025430679321,
|
| 474 |
+
"loss": 0.6863,
|
| 475 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 476 |
+
"rewards/chosen": -0.7793084979057312,
|
| 477 |
+
"rewards/margins": 0.8007953763008118,
|
| 478 |
+
"rewards/rejected": -1.580103874206543,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2247892600686856,
|
| 483 |
+
"grad_norm": 0.4416671097278595,
|
| 484 |
+
"learning_rate": 4.610819813755038e-06,
|
| 485 |
+
"logits/chosen": 1.582745909690857,
|
| 486 |
+
"logits/rejected": 0.3820720911026001,
|
| 487 |
+
"logps/chosen": -0.5181297063827515,
|
| 488 |
+
"logps/rejected": -1.2198141813278198,
|
| 489 |
+
"loss": 0.5809,
|
| 490 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 491 |
+
"rewards/chosen": -0.7771945595741272,
|
| 492 |
+
"rewards/margins": 1.0525267124176025,
|
| 493 |
+
"rewards/rejected": -1.8297210931777954,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.23311478821937767,
|
| 498 |
+
"grad_norm": 2.7746617794036865,
|
| 499 |
+
"learning_rate": 4.582303101775249e-06,
|
| 500 |
+
"logits/chosen": 1.2947760820388794,
|
| 501 |
+
"logits/rejected": 0.27237796783447266,
|
| 502 |
+
"logps/chosen": -0.643541693687439,
|
| 503 |
+
"logps/rejected": -1.7467323541641235,
|
| 504 |
+
"loss": 0.5775,
|
| 505 |
+
"rewards/accuracies": 0.6875,
|
| 506 |
+
"rewards/chosen": -0.9653124809265137,
|
| 507 |
+
"rewards/margins": 1.6547861099243164,
|
| 508 |
+
"rewards/rejected": -2.62009859085083,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.2414403163700697,
|
| 513 |
+
"grad_norm": 0.6444702744483948,
|
| 514 |
+
"learning_rate": 4.55287302283426e-06,
|
| 515 |
+
"logits/chosen": 1.2399464845657349,
|
| 516 |
+
"logits/rejected": 0.22667090594768524,
|
| 517 |
+
"logps/chosen": -0.7517040967941284,
|
| 518 |
+
"logps/rejected": -1.9010766744613647,
|
| 519 |
+
"loss": 0.5314,
|
| 520 |
+
"rewards/accuracies": 0.625,
|
| 521 |
+
"rewards/chosen": -1.1275560855865479,
|
| 522 |
+
"rewards/margins": 1.724058747291565,
|
| 523 |
+
"rewards/rejected": -2.8516147136688232,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.24976584452076178,
|
| 528 |
+
"grad_norm": 0.5103917717933655,
|
| 529 |
+
"learning_rate": 4.522542485937369e-06,
|
| 530 |
+
"logits/chosen": 1.438954472541809,
|
| 531 |
+
"logits/rejected": 0.5288833379745483,
|
| 532 |
+
"logps/chosen": -0.7871009707450867,
|
| 533 |
+
"logps/rejected": -2.0329811573028564,
|
| 534 |
+
"loss": 0.5271,
|
| 535 |
+
"rewards/accuracies": 0.699999988079071,
|
| 536 |
+
"rewards/chosen": -1.1806514263153076,
|
| 537 |
+
"rewards/margins": 1.8688204288482666,
|
| 538 |
+
"rewards/rejected": -3.049471616744995,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.24976584452076178,
|
| 543 |
+
"eval_logits/chosen": 1.3706706762313843,
|
| 544 |
+
"eval_logits/rejected": 0.8007871508598328,
|
| 545 |
+
"eval_logps/chosen": -0.7460500001907349,
|
| 546 |
+
"eval_logps/rejected": -2.209245443344116,
|
| 547 |
+
"eval_loss": 0.5008835792541504,
|
| 548 |
+
"eval_rewards/accuracies": 0.7244898080825806,
|
| 549 |
+
"eval_rewards/chosen": -1.1190749406814575,
|
| 550 |
+
"eval_rewards/margins": 2.194793224334717,
|
| 551 |
+
"eval_rewards/rejected": -3.313868284225464,
|
| 552 |
+
"eval_runtime": 29.0227,
|
| 553 |
+
"eval_samples_per_second": 26.772,
|
| 554 |
+
"eval_steps_per_second": 3.377,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.2580913726714538,
|
| 559 |
+
"grad_norm": 0.7984316945075989,
|
| 560 |
+
"learning_rate": 4.491324795060491e-06,
|
| 561 |
+
"logits/chosen": 0.9250973463058472,
|
| 562 |
+
"logits/rejected": 0.1887839138507843,
|
| 563 |
+
"logps/chosen": -0.8511486053466797,
|
| 564 |
+
"logps/rejected": -2.447072982788086,
|
| 565 |
+
"loss": 0.5506,
|
| 566 |
+
"rewards/accuracies": 0.699999988079071,
|
| 567 |
+
"rewards/chosen": -1.2767229080200195,
|
| 568 |
+
"rewards/margins": 2.3938865661621094,
|
| 569 |
+
"rewards/rejected": -3.670609712600708,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.2664169008221459,
|
| 574 |
+
"grad_norm": 0.5243161916732788,
|
| 575 |
+
"learning_rate": 4.4592336433146e-06,
|
| 576 |
+
"logits/chosen": 2.437886953353882,
|
| 577 |
+
"logits/rejected": 1.6011940240859985,
|
| 578 |
+
"logps/chosen": -0.7107629776000977,
|
| 579 |
+
"logps/rejected": -2.132263422012329,
|
| 580 |
+
"loss": 0.5423,
|
| 581 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 582 |
+
"rewards/chosen": -1.0661444664001465,
|
| 583 |
+
"rewards/margins": 2.1322507858276367,
|
| 584 |
+
"rewards/rejected": -3.198395013809204,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.27474242897283796,
|
| 589 |
+
"grad_norm": 0.4742359220981598,
|
| 590 |
+
"learning_rate": 4.426283106939474e-06,
|
| 591 |
+
"logits/chosen": 1.8433977365493774,
|
| 592 |
+
"logits/rejected": 1.199568748474121,
|
| 593 |
+
"logps/chosen": -0.8737133145332336,
|
| 594 |
+
"logps/rejected": -2.1652615070343018,
|
| 595 |
+
"loss": 0.5015,
|
| 596 |
+
"rewards/accuracies": 0.612500011920929,
|
| 597 |
+
"rewards/chosen": -1.3105700016021729,
|
| 598 |
+
"rewards/margins": 1.9373222589492798,
|
| 599 |
+
"rewards/rejected": -3.247892379760742,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.28306795712353,
|
| 604 |
+
"grad_norm": 0.5529736280441284,
|
| 605 |
+
"learning_rate": 4.3924876391293915e-06,
|
| 606 |
+
"logits/chosen": 2.0044589042663574,
|
| 607 |
+
"logits/rejected": 0.9263212084770203,
|
| 608 |
+
"logps/chosen": -0.9175036549568176,
|
| 609 |
+
"logps/rejected": -2.6408374309539795,
|
| 610 |
+
"loss": 0.4921,
|
| 611 |
+
"rewards/accuracies": 0.699999988079071,
|
| 612 |
+
"rewards/chosen": -1.3762553930282593,
|
| 613 |
+
"rewards/margins": 2.585000991821289,
|
| 614 |
+
"rewards/rejected": -3.961256504058838,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.2913934852742221,
|
| 619 |
+
"grad_norm": 0.7060612440109253,
|
| 620 |
+
"learning_rate": 4.357862063693486e-06,
|
| 621 |
+
"logits/chosen": 2.243232250213623,
|
| 622 |
+
"logits/rejected": 1.6251205205917358,
|
| 623 |
+
"logps/chosen": -0.9481338262557983,
|
| 624 |
+
"logps/rejected": -2.9519124031066895,
|
| 625 |
+
"loss": 0.4753,
|
| 626 |
+
"rewards/accuracies": 0.675000011920929,
|
| 627 |
+
"rewards/chosen": -1.4222007989883423,
|
| 628 |
+
"rewards/margins": 3.0056674480438232,
|
| 629 |
+
"rewards/rejected": -4.427868366241455,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.2913934852742221,
|
| 634 |
+
"eval_logits/chosen": 1.7781500816345215,
|
| 635 |
+
"eval_logits/rejected": 1.412752628326416,
|
| 636 |
+
"eval_logps/chosen": -0.9692521095275879,
|
| 637 |
+
"eval_logps/rejected": -2.8247811794281006,
|
| 638 |
+
"eval_loss": 0.4446474015712738,
|
| 639 |
+
"eval_rewards/accuracies": 0.7346938848495483,
|
| 640 |
+
"eval_rewards/chosen": -1.4538781642913818,
|
| 641 |
+
"eval_rewards/margins": 2.7832937240600586,
|
| 642 |
+
"eval_rewards/rejected": -4.2371721267700195,
|
| 643 |
+
"eval_runtime": 29.0245,
|
| 644 |
+
"eval_samples_per_second": 26.77,
|
| 645 |
+
"eval_steps_per_second": 3.376,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.29971901342491414,
|
| 650 |
+
"grad_norm": 0.9664792418479919,
|
| 651 |
+
"learning_rate": 4.322421568553529e-06,
|
| 652 |
+
"logits/chosen": 1.7094570398330688,
|
| 653 |
+
"logits/rejected": 1.1617993116378784,
|
| 654 |
+
"logps/chosen": -0.992924690246582,
|
| 655 |
+
"logps/rejected": -2.7834811210632324,
|
| 656 |
+
"loss": 0.4972,
|
| 657 |
+
"rewards/accuracies": 0.675000011920929,
|
| 658 |
+
"rewards/chosen": -1.4893869161605835,
|
| 659 |
+
"rewards/margins": 2.6858346462249756,
|
| 660 |
+
"rewards/rejected": -4.1752214431762695,
|
| 661 |
+
"step": 360
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.3080445415756062,
|
| 665 |
+
"grad_norm": 0.7800536155700684,
|
| 666 |
+
"learning_rate": 4.286181699082008e-06,
|
| 667 |
+
"logits/chosen": 2.9170143604278564,
|
| 668 |
+
"logits/rejected": 2.384690523147583,
|
| 669 |
+
"logps/chosen": -1.0323909521102905,
|
| 670 |
+
"logps/rejected": -2.726369857788086,
|
| 671 |
+
"loss": 0.4689,
|
| 672 |
+
"rewards/accuracies": 0.625,
|
| 673 |
+
"rewards/chosen": -1.548586368560791,
|
| 674 |
+
"rewards/margins": 2.540968418121338,
|
| 675 |
+
"rewards/rejected": -4.089555263519287,
|
| 676 |
+
"step": 370
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.3163700697262983,
|
| 680 |
+
"grad_norm": 1.3163660764694214,
|
| 681 |
+
"learning_rate": 4.249158351283414e-06,
|
| 682 |
+
"logits/chosen": 2.780831813812256,
|
| 683 |
+
"logits/rejected": 1.753291130065918,
|
| 684 |
+
"logps/chosen": -1.0468894243240356,
|
| 685 |
+
"logps/rejected": -2.7425389289855957,
|
| 686 |
+
"loss": 0.4835,
|
| 687 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 688 |
+
"rewards/chosen": -1.5703339576721191,
|
| 689 |
+
"rewards/margins": 2.5434746742248535,
|
| 690 |
+
"rewards/rejected": -4.113808631896973,
|
| 691 |
+
"step": 380
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.3246955978769903,
|
| 695 |
+
"grad_norm": 0.6381780505180359,
|
| 696 |
+
"learning_rate": 4.211367764821722e-06,
|
| 697 |
+
"logits/chosen": 2.585071086883545,
|
| 698 |
+
"logits/rejected": 1.9254558086395264,
|
| 699 |
+
"logps/chosen": -1.2089946269989014,
|
| 700 |
+
"logps/rejected": -3.615030288696289,
|
| 701 |
+
"loss": 0.4518,
|
| 702 |
+
"rewards/accuracies": 0.737500011920929,
|
| 703 |
+
"rewards/chosen": -1.8134920597076416,
|
| 704 |
+
"rewards/margins": 3.6090526580810547,
|
| 705 |
+
"rewards/rejected": -5.422544956207275,
|
| 706 |
+
"step": 390
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.33302112602768236,
|
| 710 |
+
"grad_norm": 0.9214782118797302,
|
| 711 |
+
"learning_rate": 4.172826515897146e-06,
|
| 712 |
+
"logits/chosen": 1.9765586853027344,
|
| 713 |
+
"logits/rejected": 1.1926987171173096,
|
| 714 |
+
"logps/chosen": -1.2852815389633179,
|
| 715 |
+
"logps/rejected": -3.786972761154175,
|
| 716 |
+
"loss": 0.4165,
|
| 717 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 718 |
+
"rewards/chosen": -1.9279224872589111,
|
| 719 |
+
"rewards/margins": 3.7525367736816406,
|
| 720 |
+
"rewards/rejected": -5.680459022521973,
|
| 721 |
+
"step": 400
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.33302112602768236,
|
| 725 |
+
"eval_logits/chosen": 2.6366844177246094,
|
| 726 |
+
"eval_logits/rejected": 2.394319534301758,
|
| 727 |
+
"eval_logps/chosen": -1.322396993637085,
|
| 728 |
+
"eval_logps/rejected": -3.686817169189453,
|
| 729 |
+
"eval_loss": 0.4065541923046112,
|
| 730 |
+
"eval_rewards/accuracies": 0.7551020383834839,
|
| 731 |
+
"eval_rewards/chosen": -1.9835957288742065,
|
| 732 |
+
"eval_rewards/margins": 3.5466296672821045,
|
| 733 |
+
"eval_rewards/rejected": -5.5302252769470215,
|
| 734 |
+
"eval_runtime": 29.025,
|
| 735 |
+
"eval_samples_per_second": 26.77,
|
| 736 |
+
"eval_steps_per_second": 3.376,
|
| 737 |
+
"step": 400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.34134665417837445,
|
| 741 |
+
"grad_norm": 1.5113208293914795,
|
| 742 |
+
"learning_rate": 4.133551509975264e-06,
|
| 743 |
+
"logits/chosen": 2.0068416595458984,
|
| 744 |
+
"logits/rejected": 1.5152744054794312,
|
| 745 |
+
"logps/chosen": -1.5090525150299072,
|
| 746 |
+
"logps/rejected": -3.9272122383117676,
|
| 747 |
+
"loss": 0.4004,
|
| 748 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 749 |
+
"rewards/chosen": -2.2635788917541504,
|
| 750 |
+
"rewards/margins": 3.627239227294922,
|
| 751 |
+
"rewards/rejected": -5.890818119049072,
|
| 752 |
+
"step": 410
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.3496721823290665,
|
| 756 |
+
"grad_norm": 11.516369819641113,
|
| 757 |
+
"learning_rate": 4.093559974371725e-06,
|
| 758 |
+
"logits/chosen": 3.343449115753174,
|
| 759 |
+
"logits/rejected": 2.920070171356201,
|
| 760 |
+
"logps/chosen": -1.8312532901763916,
|
| 761 |
+
"logps/rejected": -4.115124702453613,
|
| 762 |
+
"loss": 0.4045,
|
| 763 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 764 |
+
"rewards/chosen": -2.746879816055298,
|
| 765 |
+
"rewards/margins": 3.425807476043701,
|
| 766 |
+
"rewards/rejected": -6.17268705368042,
|
| 767 |
+
"step": 420
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.35799771047975854,
|
| 771 |
+
"grad_norm": 3.0497395992279053,
|
| 772 |
+
"learning_rate": 4.052869450695776e-06,
|
| 773 |
+
"logits/chosen": 2.5527279376983643,
|
| 774 |
+
"logits/rejected": 2.2495744228363037,
|
| 775 |
+
"logps/chosen": -2.2998366355895996,
|
| 776 |
+
"logps/rejected": -4.966278076171875,
|
| 777 |
+
"loss": 0.3758,
|
| 778 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 779 |
+
"rewards/chosen": -3.4497551918029785,
|
| 780 |
+
"rewards/margins": 3.9996612071990967,
|
| 781 |
+
"rewards/rejected": -7.4494171142578125,
|
| 782 |
+
"step": 430
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.36632323863045063,
|
| 786 |
+
"grad_norm": 3.900503158569336,
|
| 787 |
+
"learning_rate": 4.011497787155938e-06,
|
| 788 |
+
"logits/chosen": 2.4560112953186035,
|
| 789 |
+
"logits/rejected": 2.3936328887939453,
|
| 790 |
+
"logps/chosen": -2.563218593597412,
|
| 791 |
+
"logps/rejected": -5.063398838043213,
|
| 792 |
+
"loss": 0.3739,
|
| 793 |
+
"rewards/accuracies": 0.8125,
|
| 794 |
+
"rewards/chosen": -3.8448281288146973,
|
| 795 |
+
"rewards/margins": 3.750270366668701,
|
| 796 |
+
"rewards/rejected": -7.595097541809082,
|
| 797 |
+
"step": 440
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.3746487667811427,
|
| 801 |
+
"grad_norm": 2.8846070766448975,
|
| 802 |
+
"learning_rate": 3.969463130731183e-06,
|
| 803 |
+
"logits/chosen": 2.5467796325683594,
|
| 804 |
+
"logits/rejected": 2.4370405673980713,
|
| 805 |
+
"logps/chosen": -2.4494822025299072,
|
| 806 |
+
"logps/rejected": -5.12601900100708,
|
| 807 |
+
"loss": 0.2905,
|
| 808 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 809 |
+
"rewards/chosen": -3.6742234230041504,
|
| 810 |
+
"rewards/margins": 4.014804840087891,
|
| 811 |
+
"rewards/rejected": -7.689028263092041,
|
| 812 |
+
"step": 450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.3746487667811427,
|
| 816 |
+
"eval_logits/chosen": 2.922081232070923,
|
| 817 |
+
"eval_logits/rejected": 2.879075050354004,
|
| 818 |
+
"eval_logps/chosen": -2.352473020553589,
|
| 819 |
+
"eval_logps/rejected": -5.1224799156188965,
|
| 820 |
+
"eval_loss": 0.3302614390850067,
|
| 821 |
+
"eval_rewards/accuracies": 0.8673469424247742,
|
| 822 |
+
"eval_rewards/chosen": -3.5287091732025146,
|
| 823 |
+
"eval_rewards/margins": 4.155009746551514,
|
| 824 |
+
"eval_rewards/rejected": -7.683719635009766,
|
| 825 |
+
"eval_runtime": 29.0235,
|
| 826 |
+
"eval_samples_per_second": 26.771,
|
| 827 |
+
"eval_steps_per_second": 3.377,
|
| 828 |
+
"step": 450
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.3829742949318347,
|
| 832 |
+
"grad_norm": 4.662614345550537,
|
| 833 |
+
"learning_rate": 3.92678391921108e-06,
|
| 834 |
+
"logits/chosen": 2.428154468536377,
|
| 835 |
+
"logits/rejected": 2.2403202056884766,
|
| 836 |
+
"logps/chosen": -2.5936172008514404,
|
| 837 |
+
"logps/rejected": -5.356133460998535,
|
| 838 |
+
"loss": 0.2881,
|
| 839 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 840 |
+
"rewards/chosen": -3.89042592048645,
|
| 841 |
+
"rewards/margins": 4.143774509429932,
|
| 842 |
+
"rewards/rejected": -8.034199714660645,
|
| 843 |
+
"step": 460
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.3912998230825268,
|
| 847 |
+
"grad_norm": 2.716899871826172,
|
| 848 |
+
"learning_rate": 3.88347887310836e-06,
|
| 849 |
+
"logits/chosen": 2.437295436859131,
|
| 850 |
+
"logits/rejected": 2.271914005279541,
|
| 851 |
+
"logps/chosen": -2.470245361328125,
|
| 852 |
+
"logps/rejected": -5.719494819641113,
|
| 853 |
+
"loss": 0.31,
|
| 854 |
+
"rewards/accuracies": 0.887499988079071,
|
| 855 |
+
"rewards/chosen": -3.70536732673645,
|
| 856 |
+
"rewards/margins": 4.873874187469482,
|
| 857 |
+
"rewards/rejected": -8.579241752624512,
|
| 858 |
+
"step": 470
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.39962535123321885,
|
| 862 |
+
"grad_norm": 3.343271255493164,
|
| 863 |
+
"learning_rate": 3.839566987447492e-06,
|
| 864 |
+
"logits/chosen": 2.144461154937744,
|
| 865 |
+
"logits/rejected": 2.0314810276031494,
|
| 866 |
+
"logps/chosen": -2.5805585384368896,
|
| 867 |
+
"logps/rejected": -5.418456077575684,
|
| 868 |
+
"loss": 0.3194,
|
| 869 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 870 |
+
"rewards/chosen": -3.870838165283203,
|
| 871 |
+
"rewards/margins": 4.256844997406006,
|
| 872 |
+
"rewards/rejected": -8.12768268585205,
|
| 873 |
+
"step": 480
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.4079508793839109,
|
| 877 |
+
"grad_norm": 6.411283493041992,
|
| 878 |
+
"learning_rate": 3.795067523432826e-06,
|
| 879 |
+
"logits/chosen": 2.408092498779297,
|
| 880 |
+
"logits/rejected": 2.2996156215667725,
|
| 881 |
+
"logps/chosen": -2.8846375942230225,
|
| 882 |
+
"logps/rejected": -5.957771301269531,
|
| 883 |
+
"loss": 0.3353,
|
| 884 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 885 |
+
"rewards/chosen": -4.326956748962402,
|
| 886 |
+
"rewards/margins": 4.6097002029418945,
|
| 887 |
+
"rewards/rejected": -8.936657905578613,
|
| 888 |
+
"step": 490
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.416276407534603,
|
| 892 |
+
"grad_norm": 3.2472238540649414,
|
| 893 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 894 |
+
"logits/chosen": 3.0815653800964355,
|
| 895 |
+
"logits/rejected": 2.8496975898742676,
|
| 896 |
+
"logps/chosen": -3.061626434326172,
|
| 897 |
+
"logps/rejected": -5.966124534606934,
|
| 898 |
+
"loss": 0.3018,
|
| 899 |
+
"rewards/accuracies": 0.925000011920929,
|
| 900 |
+
"rewards/chosen": -4.592440128326416,
|
| 901 |
+
"rewards/margins": 4.356747627258301,
|
| 902 |
+
"rewards/rejected": -8.949186325073242,
|
| 903 |
+
"step": 500
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.416276407534603,
|
| 907 |
+
"eval_logits/chosen": 2.7115373611450195,
|
| 908 |
+
"eval_logits/rejected": 2.763493061065674,
|
| 909 |
+
"eval_logps/chosen": -2.85333251953125,
|
| 910 |
+
"eval_logps/rejected": -5.915884017944336,
|
| 911 |
+
"eval_loss": 0.3079966604709625,
|
| 912 |
+
"eval_rewards/accuracies": 0.8979591727256775,
|
| 913 |
+
"eval_rewards/chosen": -4.279998302459717,
|
| 914 |
+
"eval_rewards/margins": 4.593828201293945,
|
| 915 |
+
"eval_rewards/rejected": -8.873826026916504,
|
| 916 |
+
"eval_runtime": 29.0268,
|
| 917 |
+
"eval_samples_per_second": 26.768,
|
| 918 |
+
"eval_steps_per_second": 3.376,
|
| 919 |
+
"step": 500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.42460193568529503,
|
| 923 |
+
"grad_norm": 10.017457962036133,
|
| 924 |
+
"learning_rate": 3.7043841852542884e-06,
|
| 925 |
+
"logits/chosen": 2.775202989578247,
|
| 926 |
+
"logits/rejected": 2.6122496128082275,
|
| 927 |
+
"logps/chosen": -3.0054879188537598,
|
| 928 |
+
"logps/rejected": -6.258307456970215,
|
| 929 |
+
"loss": 0.3101,
|
| 930 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 931 |
+
"rewards/chosen": -4.5082316398620605,
|
| 932 |
+
"rewards/margins": 4.879229545593262,
|
| 933 |
+
"rewards/rejected": -9.387460708618164,
|
| 934 |
+
"step": 510
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.43292746383598707,
|
| 938 |
+
"grad_norm": 4.494226932525635,
|
| 939 |
+
"learning_rate": 3.658240087799655e-06,
|
| 940 |
+
"logits/chosen": 2.816701889038086,
|
| 941 |
+
"logits/rejected": 2.4107789993286133,
|
| 942 |
+
"logps/chosen": -3.2932097911834717,
|
| 943 |
+
"logps/rejected": -6.099677562713623,
|
| 944 |
+
"loss": 0.2925,
|
| 945 |
+
"rewards/accuracies": 0.875,
|
| 946 |
+
"rewards/chosen": -4.939814567565918,
|
| 947 |
+
"rewards/margins": 4.209702014923096,
|
| 948 |
+
"rewards/rejected": -9.149517059326172,
|
| 949 |
+
"step": 520
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.44125299198667917,
|
| 953 |
+
"grad_norm": 2.957486391067505,
|
| 954 |
+
"learning_rate": 3.611587947962319e-06,
|
| 955 |
+
"logits/chosen": 2.3626818656921387,
|
| 956 |
+
"logits/rejected": 2.4196550846099854,
|
| 957 |
+
"logps/chosen": -3.085209608078003,
|
| 958 |
+
"logps/rejected": -6.118277072906494,
|
| 959 |
+
"loss": 0.3169,
|
| 960 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 961 |
+
"rewards/chosen": -4.627814292907715,
|
| 962 |
+
"rewards/margins": 4.549601078033447,
|
| 963 |
+
"rewards/rejected": -9.17741584777832,
|
| 964 |
+
"step": 530
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.4495785201373712,
|
| 968 |
+
"grad_norm": 3.429408550262451,
|
| 969 |
+
"learning_rate": 3.564448228912682e-06,
|
| 970 |
+
"logits/chosen": 2.559816360473633,
|
| 971 |
+
"logits/rejected": 2.598250150680542,
|
| 972 |
+
"logps/chosen": -3.3060078620910645,
|
| 973 |
+
"logps/rejected": -6.124637126922607,
|
| 974 |
+
"loss": 0.3271,
|
| 975 |
+
"rewards/accuracies": 0.800000011920929,
|
| 976 |
+
"rewards/chosen": -4.959012031555176,
|
| 977 |
+
"rewards/margins": 4.227944850921631,
|
| 978 |
+
"rewards/rejected": -9.186956405639648,
|
| 979 |
+
"step": 540
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 0.45790404828806325,
|
| 983 |
+
"grad_norm": 2.110722780227661,
|
| 984 |
+
"learning_rate": 3.516841607689501e-06,
|
| 985 |
+
"logits/chosen": 2.4487693309783936,
|
| 986 |
+
"logits/rejected": 2.0568625926971436,
|
| 987 |
+
"logps/chosen": -3.396770477294922,
|
| 988 |
+
"logps/rejected": -6.35222864151001,
|
| 989 |
+
"loss": 0.3172,
|
| 990 |
+
"rewards/accuracies": 0.824999988079071,
|
| 991 |
+
"rewards/chosen": -5.095156192779541,
|
| 992 |
+
"rewards/margins": 4.4331865310668945,
|
| 993 |
+
"rewards/rejected": -9.528343200683594,
|
| 994 |
+
"step": 550
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.45790404828806325,
|
| 998 |
+
"eval_logits/chosen": 2.5644595623016357,
|
| 999 |
+
"eval_logits/rejected": 2.6437506675720215,
|
| 1000 |
+
"eval_logps/chosen": -3.1958370208740234,
|
| 1001 |
+
"eval_logps/rejected": -6.542325496673584,
|
| 1002 |
+
"eval_loss": 0.28538385033607483,
|
| 1003 |
+
"eval_rewards/accuracies": 0.918367326259613,
|
| 1004 |
+
"eval_rewards/chosen": -4.793755054473877,
|
| 1005 |
+
"eval_rewards/margins": 5.0197319984436035,
|
| 1006 |
+
"eval_rewards/rejected": -9.813486099243164,
|
| 1007 |
+
"eval_runtime": 29.0252,
|
| 1008 |
+
"eval_samples_per_second": 26.77,
|
| 1009 |
+
"eval_steps_per_second": 3.376,
|
| 1010 |
+
"step": 550
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.46622957643875534,
|
| 1014 |
+
"grad_norm": 2.0929551124572754,
|
| 1015 |
+
"learning_rate": 3.4687889661302577e-06,
|
| 1016 |
+
"logits/chosen": 2.497122287750244,
|
| 1017 |
+
"logits/rejected": 2.1119792461395264,
|
| 1018 |
+
"logps/chosen": -3.586158037185669,
|
| 1019 |
+
"logps/rejected": -6.939994812011719,
|
| 1020 |
+
"loss": 0.2826,
|
| 1021 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1022 |
+
"rewards/chosen": -5.379237174987793,
|
| 1023 |
+
"rewards/margins": 5.030755043029785,
|
| 1024 |
+
"rewards/rejected": -10.409992218017578,
|
| 1025 |
+
"step": 560
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.4745551045894474,
|
| 1029 |
+
"grad_norm": 3.344160556793213,
|
| 1030 |
+
"learning_rate": 3.4203113817116955e-06,
|
| 1031 |
+
"logits/chosen": 3.181488275527954,
|
| 1032 |
+
"logits/rejected": 2.8188672065734863,
|
| 1033 |
+
"logps/chosen": -3.465902328491211,
|
| 1034 |
+
"logps/rejected": -6.737443447113037,
|
| 1035 |
+
"loss": 0.3027,
|
| 1036 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1037 |
+
"rewards/chosen": -5.198853492736816,
|
| 1038 |
+
"rewards/margins": 4.90731143951416,
|
| 1039 |
+
"rewards/rejected": -10.106164932250977,
|
| 1040 |
+
"step": 570
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.4828806327401394,
|
| 1044 |
+
"grad_norm": 6.381539344787598,
|
| 1045 |
+
"learning_rate": 3.3714301183045382e-06,
|
| 1046 |
+
"logits/chosen": 3.8848679065704346,
|
| 1047 |
+
"logits/rejected": 3.54484224319458,
|
| 1048 |
+
"logps/chosen": -3.321965456008911,
|
| 1049 |
+
"logps/rejected": -6.796433448791504,
|
| 1050 |
+
"loss": 0.2619,
|
| 1051 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1052 |
+
"rewards/chosen": -4.982948303222656,
|
| 1053 |
+
"rewards/margins": 5.211700916290283,
|
| 1054 |
+
"rewards/rejected": -10.194650650024414,
|
| 1055 |
+
"step": 580
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 0.4912061608908315,
|
| 1059 |
+
"grad_norm": 3.058936834335327,
|
| 1060 |
+
"learning_rate": 3.3221666168464584e-06,
|
| 1061 |
+
"logits/chosen": 2.9645297527313232,
|
| 1062 |
+
"logits/rejected": 2.7630581855773926,
|
| 1063 |
+
"logps/chosen": -3.2019195556640625,
|
| 1064 |
+
"logps/rejected": -6.635239601135254,
|
| 1065 |
+
"loss": 0.2573,
|
| 1066 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1067 |
+
"rewards/chosen": -4.802879810333252,
|
| 1068 |
+
"rewards/margins": 5.149979114532471,
|
| 1069 |
+
"rewards/rejected": -9.952859878540039,
|
| 1070 |
+
"step": 590
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.49953168904152356,
|
| 1074 |
+
"grad_norm": 4.1828155517578125,
|
| 1075 |
+
"learning_rate": 3.272542485937369e-06,
|
| 1076 |
+
"logits/chosen": 2.696993350982666,
|
| 1077 |
+
"logits/rejected": 2.7842001914978027,
|
| 1078 |
+
"logps/chosen": -3.3624558448791504,
|
| 1079 |
+
"logps/rejected": -6.4542059898376465,
|
| 1080 |
+
"loss": 0.2598,
|
| 1081 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1082 |
+
"rewards/chosen": -5.043683052062988,
|
| 1083 |
+
"rewards/margins": 4.637625217437744,
|
| 1084 |
+
"rewards/rejected": -9.68130874633789,
|
| 1085 |
+
"step": 600
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.49953168904152356,
|
| 1089 |
+
"eval_logits/chosen": 2.9141366481781006,
|
| 1090 |
+
"eval_logits/rejected": 2.9971513748168945,
|
| 1091 |
+
"eval_logps/chosen": -3.1258208751678467,
|
| 1092 |
+
"eval_logps/rejected": -6.787447452545166,
|
| 1093 |
+
"eval_loss": 0.27035781741142273,
|
| 1094 |
+
"eval_rewards/accuracies": 0.918367326259613,
|
| 1095 |
+
"eval_rewards/chosen": -4.688731670379639,
|
| 1096 |
+
"eval_rewards/margins": 5.492439270019531,
|
| 1097 |
+
"eval_rewards/rejected": -10.181171417236328,
|
| 1098 |
+
"eval_runtime": 29.0227,
|
| 1099 |
+
"eval_samples_per_second": 26.772,
|
| 1100 |
+
"eval_steps_per_second": 3.377,
|
| 1101 |
+
"step": 600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.5078572171922157,
|
| 1105 |
+
"grad_norm": 3.1104886531829834,
|
| 1106 |
+
"learning_rate": 3.222579492361179e-06,
|
| 1107 |
+
"logits/chosen": 2.582984447479248,
|
| 1108 |
+
"logits/rejected": 2.424341917037964,
|
| 1109 |
+
"logps/chosen": -3.0132031440734863,
|
| 1110 |
+
"logps/rejected": -6.317469596862793,
|
| 1111 |
+
"loss": 0.2598,
|
| 1112 |
+
"rewards/accuracies": 0.875,
|
| 1113 |
+
"rewards/chosen": -4.519804954528809,
|
| 1114 |
+
"rewards/margins": 4.956398963928223,
|
| 1115 |
+
"rewards/rejected": -9.476203918457031,
|
| 1116 |
+
"step": 610
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.5161827453429076,
|
| 1120 |
+
"grad_norm": 12.320380210876465,
|
| 1121 |
+
"learning_rate": 3.1722995515381644e-06,
|
| 1122 |
+
"logits/chosen": 2.1016178131103516,
|
| 1123 |
+
"logits/rejected": 2.345324754714966,
|
| 1124 |
+
"logps/chosen": -3.1399683952331543,
|
| 1125 |
+
"logps/rejected": -7.096994876861572,
|
| 1126 |
+
"loss": 0.2601,
|
| 1127 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1128 |
+
"rewards/chosen": -4.709952354431152,
|
| 1129 |
+
"rewards/margins": 5.935539722442627,
|
| 1130 |
+
"rewards/rejected": -10.645492553710938,
|
| 1131 |
+
"step": 620
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.5245082734935997,
|
| 1135 |
+
"grad_norm": 2.704423189163208,
|
| 1136 |
+
"learning_rate": 3.121724717912138e-06,
|
| 1137 |
+
"logits/chosen": 2.108675718307495,
|
| 1138 |
+
"logits/rejected": 2.369410991668701,
|
| 1139 |
+
"logps/chosen": -3.6519737243652344,
|
| 1140 |
+
"logps/rejected": -6.964946746826172,
|
| 1141 |
+
"loss": 0.2351,
|
| 1142 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1143 |
+
"rewards/chosen": -5.477960586547852,
|
| 1144 |
+
"rewards/margins": 4.96945858001709,
|
| 1145 |
+
"rewards/rejected": -10.447419166564941,
|
| 1146 |
+
"step": 630
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.5328338016442918,
|
| 1150 |
+
"grad_norm": 4.401206970214844,
|
| 1151 |
+
"learning_rate": 3.0708771752766397e-06,
|
| 1152 |
+
"logits/chosen": 2.3692595958709717,
|
| 1153 |
+
"logits/rejected": 2.5313620567321777,
|
| 1154 |
+
"logps/chosen": -4.0485663414001465,
|
| 1155 |
+
"logps/rejected": -7.747661590576172,
|
| 1156 |
+
"loss": 0.2265,
|
| 1157 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1158 |
+
"rewards/chosen": -6.072849750518799,
|
| 1159 |
+
"rewards/margins": 5.548642635345459,
|
| 1160 |
+
"rewards/rejected": -11.621491432189941,
|
| 1161 |
+
"step": 640
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.5411593297949838,
|
| 1165 |
+
"grad_norm": 4.68662166595459,
|
| 1166 |
+
"learning_rate": 3.019779227044398e-06,
|
| 1167 |
+
"logits/chosen": 2.4383034706115723,
|
| 1168 |
+
"logits/rejected": 2.4655585289001465,
|
| 1169 |
+
"logps/chosen": -3.8650074005126953,
|
| 1170 |
+
"logps/rejected": -7.987051963806152,
|
| 1171 |
+
"loss": 0.263,
|
| 1172 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1173 |
+
"rewards/chosen": -5.797511100769043,
|
| 1174 |
+
"rewards/margins": 6.183066368103027,
|
| 1175 |
+
"rewards/rejected": -11.98057746887207,
|
| 1176 |
+
"step": 650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.5411593297949838,
|
| 1180 |
+
"eval_logits/chosen": 2.7321341037750244,
|
| 1181 |
+
"eval_logits/rejected": 2.906801700592041,
|
| 1182 |
+
"eval_logps/chosen": -3.7255136966705322,
|
| 1183 |
+
"eval_logps/rejected": -7.620375633239746,
|
| 1184 |
+
"eval_loss": 0.26394686102867126,
|
| 1185 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1186 |
+
"eval_rewards/chosen": -5.5882697105407715,
|
| 1187 |
+
"eval_rewards/margins": 5.8422932624816895,
|
| 1188 |
+
"eval_rewards/rejected": -11.430564880371094,
|
| 1189 |
+
"eval_runtime": 29.0258,
|
| 1190 |
+
"eval_samples_per_second": 26.769,
|
| 1191 |
+
"eval_steps_per_second": 3.376,
|
| 1192 |
+
"step": 650
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.5494848579456759,
|
| 1196 |
+
"grad_norm": 4.704371929168701,
|
| 1197 |
+
"learning_rate": 2.9684532864643123e-06,
|
| 1198 |
+
"logits/chosen": 2.7277207374572754,
|
| 1199 |
+
"logits/rejected": 2.7106287479400635,
|
| 1200 |
+
"logps/chosen": -3.979590654373169,
|
| 1201 |
+
"logps/rejected": -6.88008975982666,
|
| 1202 |
+
"loss": 0.2933,
|
| 1203 |
+
"rewards/accuracies": 0.875,
|
| 1204 |
+
"rewards/chosen": -5.969386100769043,
|
| 1205 |
+
"rewards/margins": 4.350748062133789,
|
| 1206 |
+
"rewards/rejected": -10.320135116577148,
|
| 1207 |
+
"step": 660
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.557810386096368,
|
| 1211 |
+
"grad_norm": 3.2897160053253174,
|
| 1212 |
+
"learning_rate": 2.9169218667902562e-06,
|
| 1213 |
+
"logits/chosen": 2.207106113433838,
|
| 1214 |
+
"logits/rejected": 2.454056978225708,
|
| 1215 |
+
"logps/chosen": -3.760200023651123,
|
| 1216 |
+
"logps/rejected": -7.504108428955078,
|
| 1217 |
+
"loss": 0.2262,
|
| 1218 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1219 |
+
"rewards/chosen": -5.6402997970581055,
|
| 1220 |
+
"rewards/margins": 5.615862846374512,
|
| 1221 |
+
"rewards/rejected": -11.256162643432617,
|
| 1222 |
+
"step": 670
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.56613591424706,
|
| 1226 |
+
"grad_norm": 3.6699540615081787,
|
| 1227 |
+
"learning_rate": 2.8652075714060296e-06,
|
| 1228 |
+
"logits/chosen": 2.5904622077941895,
|
| 1229 |
+
"logits/rejected": 2.693467617034912,
|
| 1230 |
+
"logps/chosen": -3.2713139057159424,
|
| 1231 |
+
"logps/rejected": -7.3422722816467285,
|
| 1232 |
+
"loss": 0.2721,
|
| 1233 |
+
"rewards/accuracies": 0.9375,
|
| 1234 |
+
"rewards/chosen": -4.906970500946045,
|
| 1235 |
+
"rewards/margins": 6.106438636779785,
|
| 1236 |
+
"rewards/rejected": -11.013408660888672,
|
| 1237 |
+
"step": 680
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.5744614423977521,
|
| 1241 |
+
"grad_norm": 3.054532289505005,
|
| 1242 |
+
"learning_rate": 2.813333083910761e-06,
|
| 1243 |
+
"logits/chosen": 2.9145145416259766,
|
| 1244 |
+
"logits/rejected": 2.7135214805603027,
|
| 1245 |
+
"logps/chosen": -3.5082690715789795,
|
| 1246 |
+
"logps/rejected": -7.293328762054443,
|
| 1247 |
+
"loss": 0.271,
|
| 1248 |
+
"rewards/accuracies": 0.9375,
|
| 1249 |
+
"rewards/chosen": -5.26240348815918,
|
| 1250 |
+
"rewards/margins": 5.677589416503906,
|
| 1251 |
+
"rewards/rejected": -10.939992904663086,
|
| 1252 |
+
"step": 690
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 0.5827869705484442,
|
| 1256 |
+
"grad_norm": 3.5161256790161133,
|
| 1257 |
+
"learning_rate": 2.761321158169134e-06,
|
| 1258 |
+
"logits/chosen": 2.915343761444092,
|
| 1259 |
+
"logits/rejected": 2.731520891189575,
|
| 1260 |
+
"logps/chosen": -3.4292550086975098,
|
| 1261 |
+
"logps/rejected": -8.124921798706055,
|
| 1262 |
+
"loss": 0.1985,
|
| 1263 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1264 |
+
"rewards/chosen": -5.143881797790527,
|
| 1265 |
+
"rewards/margins": 7.043501377105713,
|
| 1266 |
+
"rewards/rejected": -12.187383651733398,
|
| 1267 |
+
"step": 700
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 0.5827869705484442,
|
| 1271 |
+
"eval_logits/chosen": 2.5902156829833984,
|
| 1272 |
+
"eval_logits/rejected": 2.774846315383911,
|
| 1273 |
+
"eval_logps/chosen": -3.5158140659332275,
|
| 1274 |
+
"eval_logps/rejected": -7.544556140899658,
|
| 1275 |
+
"eval_loss": 0.24698135256767273,
|
| 1276 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1277 |
+
"eval_rewards/chosen": -5.273721694946289,
|
| 1278 |
+
"eval_rewards/margins": 6.043112754821777,
|
| 1279 |
+
"eval_rewards/rejected": -11.31683349609375,
|
| 1280 |
+
"eval_runtime": 29.0187,
|
| 1281 |
+
"eval_samples_per_second": 26.776,
|
| 1282 |
+
"eval_steps_per_second": 3.377,
|
| 1283 |
+
"step": 700
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.5911124986991362,
|
| 1287 |
+
"grad_norm": 3.2246947288513184,
|
| 1288 |
+
"learning_rate": 2.70919460833079e-06,
|
| 1289 |
+
"logits/chosen": 2.9566922187805176,
|
| 1290 |
+
"logits/rejected": 2.874277353286743,
|
| 1291 |
+
"logps/chosen": -3.772322177886963,
|
| 1292 |
+
"logps/rejected": -7.461319923400879,
|
| 1293 |
+
"loss": 0.2565,
|
| 1294 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1295 |
+
"rewards/chosen": -5.658483028411865,
|
| 1296 |
+
"rewards/margins": 5.533496856689453,
|
| 1297 |
+
"rewards/rejected": -11.191980361938477,
|
| 1298 |
+
"step": 710
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.5994380268498283,
|
| 1302 |
+
"grad_norm": 4.457447052001953,
|
| 1303 |
+
"learning_rate": 2.6569762988232838e-06,
|
| 1304 |
+
"logits/chosen": 2.653148889541626,
|
| 1305 |
+
"logits/rejected": 2.646437168121338,
|
| 1306 |
+
"logps/chosen": -3.8250937461853027,
|
| 1307 |
+
"logps/rejected": -7.855221748352051,
|
| 1308 |
+
"loss": 0.2244,
|
| 1309 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1310 |
+
"rewards/chosen": -5.737640857696533,
|
| 1311 |
+
"rewards/margins": 6.045191287994385,
|
| 1312 |
+
"rewards/rejected": -11.782832145690918,
|
| 1313 |
+
"step": 720
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.6077635550005204,
|
| 1317 |
+
"grad_norm": 3.477293014526367,
|
| 1318 |
+
"learning_rate": 2.604689134322999e-06,
|
| 1319 |
+
"logits/chosen": 2.2635607719421387,
|
| 1320 |
+
"logits/rejected": 2.2247064113616943,
|
| 1321 |
+
"logps/chosen": -3.974703550338745,
|
| 1322 |
+
"logps/rejected": -8.289571762084961,
|
| 1323 |
+
"loss": 0.2294,
|
| 1324 |
+
"rewards/accuracies": 0.9375,
|
| 1325 |
+
"rewards/chosen": -5.962055206298828,
|
| 1326 |
+
"rewards/margins": 6.4723029136657715,
|
| 1327 |
+
"rewards/rejected": -12.434357643127441,
|
| 1328 |
+
"step": 730
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.6160890831512124,
|
| 1332 |
+
"grad_norm": 1.6821621656417847,
|
| 1333 |
+
"learning_rate": 2.5523560497083927e-06,
|
| 1334 |
+
"logits/chosen": 1.8432185649871826,
|
| 1335 |
+
"logits/rejected": 1.9002739191055298,
|
| 1336 |
+
"logps/chosen": -3.8650963306427,
|
| 1337 |
+
"logps/rejected": -7.553779602050781,
|
| 1338 |
+
"loss": 0.2221,
|
| 1339 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1340 |
+
"rewards/chosen": -5.79764461517334,
|
| 1341 |
+
"rewards/margins": 5.533024787902832,
|
| 1342 |
+
"rewards/rejected": -11.330669403076172,
|
| 1343 |
+
"step": 740
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.6244146113019045,
|
| 1347 |
+
"grad_norm": 24.729644775390625,
|
| 1348 |
+
"learning_rate": 2.5e-06,
|
| 1349 |
+
"logits/chosen": 2.5135562419891357,
|
| 1350 |
+
"logits/rejected": 2.6035869121551514,
|
| 1351 |
+
"logps/chosen": -3.6619372367858887,
|
| 1352 |
+
"logps/rejected": -7.801999568939209,
|
| 1353 |
+
"loss": 0.2724,
|
| 1354 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1355 |
+
"rewards/chosen": -5.492905616760254,
|
| 1356 |
+
"rewards/margins": 6.2100934982299805,
|
| 1357 |
+
"rewards/rejected": -11.702998161315918,
|
| 1358 |
+
"step": 750
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.6244146113019045,
|
| 1362 |
+
"eval_logits/chosen": 2.876950979232788,
|
| 1363 |
+
"eval_logits/rejected": 3.0243964195251465,
|
| 1364 |
+
"eval_logps/chosen": -3.517216682434082,
|
| 1365 |
+
"eval_logps/rejected": -7.607268810272217,
|
| 1366 |
+
"eval_loss": 0.24484822154045105,
|
| 1367 |
+
"eval_rewards/accuracies": 0.9387755393981934,
|
| 1368 |
+
"eval_rewards/chosen": -5.275824546813965,
|
| 1369 |
+
"eval_rewards/margins": 6.135078430175781,
|
| 1370 |
+
"eval_rewards/rejected": -11.410903930664062,
|
| 1371 |
+
"eval_runtime": 28.9129,
|
| 1372 |
+
"eval_samples_per_second": 26.874,
|
| 1373 |
+
"eval_steps_per_second": 3.389,
|
| 1374 |
+
"step": 750
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.6327401394525966,
|
| 1378 |
+
"grad_norm": 9.702905654907227,
|
| 1379 |
+
"learning_rate": 2.447643950291608e-06,
|
| 1380 |
+
"logits/chosen": 2.693587064743042,
|
| 1381 |
+
"logits/rejected": 2.6106948852539062,
|
| 1382 |
+
"logps/chosen": -3.7441153526306152,
|
| 1383 |
+
"logps/rejected": -7.564157009124756,
|
| 1384 |
+
"loss": 0.2506,
|
| 1385 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1386 |
+
"rewards/chosen": -5.61617374420166,
|
| 1387 |
+
"rewards/margins": 5.730062961578369,
|
| 1388 |
+
"rewards/rejected": -11.346236228942871,
|
| 1389 |
+
"step": 760
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.6410656676032885,
|
| 1393 |
+
"grad_norm": 8.551860809326172,
|
| 1394 |
+
"learning_rate": 2.3953108656770018e-06,
|
| 1395 |
+
"logits/chosen": 2.894711971282959,
|
| 1396 |
+
"logits/rejected": 3.036170482635498,
|
| 1397 |
+
"logps/chosen": -3.972269058227539,
|
| 1398 |
+
"logps/rejected": -8.38014030456543,
|
| 1399 |
+
"loss": 0.2107,
|
| 1400 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1401 |
+
"rewards/chosen": -5.958403587341309,
|
| 1402 |
+
"rewards/margins": 6.6118059158325195,
|
| 1403 |
+
"rewards/rejected": -12.570208549499512,
|
| 1404 |
+
"step": 770
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.6493911957539806,
|
| 1408 |
+
"grad_norm": 2.4394350051879883,
|
| 1409 |
+
"learning_rate": 2.3430237011767166e-06,
|
| 1410 |
+
"logits/chosen": 3.1415820121765137,
|
| 1411 |
+
"logits/rejected": 3.1218018531799316,
|
| 1412 |
+
"logps/chosen": -4.007376194000244,
|
| 1413 |
+
"logps/rejected": -8.103262901306152,
|
| 1414 |
+
"loss": 0.1886,
|
| 1415 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1416 |
+
"rewards/chosen": -6.011064052581787,
|
| 1417 |
+
"rewards/margins": 6.1438307762146,
|
| 1418 |
+
"rewards/rejected": -12.154894828796387,
|
| 1419 |
+
"step": 780
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.6577167239046727,
|
| 1423 |
+
"grad_norm": 3.69184947013855,
|
| 1424 |
+
"learning_rate": 2.290805391669212e-06,
|
| 1425 |
+
"logits/chosen": 3.3487350940704346,
|
| 1426 |
+
"logits/rejected": 3.5375237464904785,
|
| 1427 |
+
"logps/chosen": -3.7646141052246094,
|
| 1428 |
+
"logps/rejected": -7.569940090179443,
|
| 1429 |
+
"loss": 0.2106,
|
| 1430 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1431 |
+
"rewards/chosen": -5.646921634674072,
|
| 1432 |
+
"rewards/margins": 5.707989692687988,
|
| 1433 |
+
"rewards/rejected": -11.354910850524902,
|
| 1434 |
+
"step": 790
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.6660422520553647,
|
| 1438 |
+
"grad_norm": 4.604506015777588,
|
| 1439 |
+
"learning_rate": 2.238678841830867e-06,
|
| 1440 |
+
"logits/chosen": 3.159898519515991,
|
| 1441 |
+
"logits/rejected": 3.09334135055542,
|
| 1442 |
+
"logps/chosen": -4.009636878967285,
|
| 1443 |
+
"logps/rejected": -7.4454545974731445,
|
| 1444 |
+
"loss": 0.2379,
|
| 1445 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1446 |
+
"rewards/chosen": -6.014455318450928,
|
| 1447 |
+
"rewards/margins": 5.1537251472473145,
|
| 1448 |
+
"rewards/rejected": -11.168180465698242,
|
| 1449 |
+
"step": 800
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.6660422520553647,
|
| 1453 |
+
"eval_logits/chosen": 2.748328924179077,
|
| 1454 |
+
"eval_logits/rejected": 2.9500906467437744,
|
| 1455 |
+
"eval_logps/chosen": -3.652164936065674,
|
| 1456 |
+
"eval_logps/rejected": -7.951470375061035,
|
| 1457 |
+
"eval_loss": 0.23568958044052124,
|
| 1458 |
+
"eval_rewards/accuracies": 0.9387755393981934,
|
| 1459 |
+
"eval_rewards/chosen": -5.478247165679932,
|
| 1460 |
+
"eval_rewards/margins": 6.448958396911621,
|
| 1461 |
+
"eval_rewards/rejected": -11.927205085754395,
|
| 1462 |
+
"eval_runtime": 29.021,
|
| 1463 |
+
"eval_samples_per_second": 26.774,
|
| 1464 |
+
"eval_steps_per_second": 3.377,
|
| 1465 |
+
"step": 800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.6743677802060568,
|
| 1469 |
+
"grad_norm": 3.968970537185669,
|
| 1470 |
+
"learning_rate": 2.186666916089239e-06,
|
| 1471 |
+
"logits/chosen": 2.384208917617798,
|
| 1472 |
+
"logits/rejected": 2.3336739540100098,
|
| 1473 |
+
"logps/chosen": -3.8832621574401855,
|
| 1474 |
+
"logps/rejected": -7.72598123550415,
|
| 1475 |
+
"loss": 0.2706,
|
| 1476 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1477 |
+
"rewards/chosen": -5.824892520904541,
|
| 1478 |
+
"rewards/margins": 5.764077663421631,
|
| 1479 |
+
"rewards/rejected": -11.588971138000488,
|
| 1480 |
+
"step": 810
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.6826933083567489,
|
| 1484 |
+
"grad_norm": 3.6892929077148438,
|
| 1485 |
+
"learning_rate": 2.134792428593971e-06,
|
| 1486 |
+
"logits/chosen": 3.5869107246398926,
|
| 1487 |
+
"logits/rejected": 3.517749786376953,
|
| 1488 |
+
"logps/chosen": -3.306342363357544,
|
| 1489 |
+
"logps/rejected": -7.020272254943848,
|
| 1490 |
+
"loss": 0.2398,
|
| 1491 |
+
"rewards/accuracies": 0.875,
|
| 1492 |
+
"rewards/chosen": -4.9595136642456055,
|
| 1493 |
+
"rewards/margins": 5.570894718170166,
|
| 1494 |
+
"rewards/rejected": -10.530407905578613,
|
| 1495 |
+
"step": 820
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.6910188365074409,
|
| 1499 |
+
"grad_norm": 4.89448881149292,
|
| 1500 |
+
"learning_rate": 2.0830781332097446e-06,
|
| 1501 |
+
"logits/chosen": 2.5076346397399902,
|
| 1502 |
+
"logits/rejected": 2.3836727142333984,
|
| 1503 |
+
"logps/chosen": -3.843027114868164,
|
| 1504 |
+
"logps/rejected": -7.852384090423584,
|
| 1505 |
+
"loss": 0.2116,
|
| 1506 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1507 |
+
"rewards/chosen": -5.764540672302246,
|
| 1508 |
+
"rewards/margins": 6.014035701751709,
|
| 1509 |
+
"rewards/rejected": -11.77857494354248,
|
| 1510 |
+
"step": 830
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 0.699344364658133,
|
| 1514 |
+
"grad_norm": 8.198432922363281,
|
| 1515 |
+
"learning_rate": 2.031546713535688e-06,
|
| 1516 |
+
"logits/chosen": 2.5533287525177,
|
| 1517 |
+
"logits/rejected": 2.407637357711792,
|
| 1518 |
+
"logps/chosen": -3.574105739593506,
|
| 1519 |
+
"logps/rejected": -8.23727798461914,
|
| 1520 |
+
"loss": 0.2415,
|
| 1521 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1522 |
+
"rewards/chosen": -5.361158847808838,
|
| 1523 |
+
"rewards/margins": 6.994758605957031,
|
| 1524 |
+
"rewards/rejected": -12.355916976928711,
|
| 1525 |
+
"step": 840
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.7076698928088251,
|
| 1529 |
+
"grad_norm": 4.123171329498291,
|
| 1530 |
+
"learning_rate": 1.9802207729556023e-06,
|
| 1531 |
+
"logits/chosen": 2.4909422397613525,
|
| 1532 |
+
"logits/rejected": 2.3119165897369385,
|
| 1533 |
+
"logps/chosen": -3.927218198776245,
|
| 1534 |
+
"logps/rejected": -7.961021423339844,
|
| 1535 |
+
"loss": 0.2217,
|
| 1536 |
+
"rewards/accuracies": 0.9375,
|
| 1537 |
+
"rewards/chosen": -5.89082670211792,
|
| 1538 |
+
"rewards/margins": 6.050704002380371,
|
| 1539 |
+
"rewards/rejected": -11.94153118133545,
|
| 1540 |
+
"step": 850
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 0.7076698928088251,
|
| 1544 |
+
"eval_logits/chosen": 2.858954668045044,
|
| 1545 |
+
"eval_logits/rejected": 3.012629270553589,
|
| 1546 |
+
"eval_logps/chosen": -3.577458381652832,
|
| 1547 |
+
"eval_logps/rejected": -7.837220668792725,
|
| 1548 |
+
"eval_loss": 0.23848077654838562,
|
| 1549 |
+
"eval_rewards/accuracies": 0.9387755393981934,
|
| 1550 |
+
"eval_rewards/chosen": -5.36618709564209,
|
| 1551 |
+
"eval_rewards/margins": 6.389642715454102,
|
| 1552 |
+
"eval_rewards/rejected": -11.755829811096191,
|
| 1553 |
+
"eval_runtime": 29.02,
|
| 1554 |
+
"eval_samples_per_second": 26.775,
|
| 1555 |
+
"eval_steps_per_second": 3.377,
|
| 1556 |
+
"step": 850
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.7159954209595171,
|
| 1560 |
+
"grad_norm": 3.4179177284240723,
|
| 1561 |
+
"learning_rate": 1.9291228247233607e-06,
|
| 1562 |
+
"logits/chosen": 2.535378932952881,
|
| 1563 |
+
"logits/rejected": 2.5335640907287598,
|
| 1564 |
+
"logps/chosen": -3.541815996170044,
|
| 1565 |
+
"logps/rejected": -7.519083499908447,
|
| 1566 |
+
"loss": 0.2167,
|
| 1567 |
+
"rewards/accuracies": 0.9375,
|
| 1568 |
+
"rewards/chosen": -5.312723159790039,
|
| 1569 |
+
"rewards/margins": 5.965902328491211,
|
| 1570 |
+
"rewards/rejected": -11.27862548828125,
|
| 1571 |
+
"step": 860
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.7243209491102092,
|
| 1575 |
+
"grad_norm": 1.8562341928482056,
|
| 1576 |
+
"learning_rate": 1.8782752820878636e-06,
|
| 1577 |
+
"logits/chosen": 3.0650887489318848,
|
| 1578 |
+
"logits/rejected": 2.7918925285339355,
|
| 1579 |
+
"logps/chosen": -3.791342258453369,
|
| 1580 |
+
"logps/rejected": -7.656645774841309,
|
| 1581 |
+
"loss": 0.1925,
|
| 1582 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1583 |
+
"rewards/chosen": -5.687013149261475,
|
| 1584 |
+
"rewards/margins": 5.7979559898376465,
|
| 1585 |
+
"rewards/rejected": -11.484968185424805,
|
| 1586 |
+
"step": 870
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.7326464772609013,
|
| 1590 |
+
"grad_norm": 9.719799995422363,
|
| 1591 |
+
"learning_rate": 1.827700448461836e-06,
|
| 1592 |
+
"logits/chosen": 2.4594621658325195,
|
| 1593 |
+
"logits/rejected": 2.4324564933776855,
|
| 1594 |
+
"logps/chosen": -3.6558470726013184,
|
| 1595 |
+
"logps/rejected": -8.101290702819824,
|
| 1596 |
+
"loss": 0.1975,
|
| 1597 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1598 |
+
"rewards/chosen": -5.483770847320557,
|
| 1599 |
+
"rewards/margins": 6.6681647300720215,
|
| 1600 |
+
"rewards/rejected": -12.151935577392578,
|
| 1601 |
+
"step": 880
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.7409720054115932,
|
| 1605 |
+
"grad_norm": 3.240176200866699,
|
| 1606 |
+
"learning_rate": 1.7774205076388207e-06,
|
| 1607 |
+
"logits/chosen": 2.689762592315674,
|
| 1608 |
+
"logits/rejected": 2.553614616394043,
|
| 1609 |
+
"logps/chosen": -3.3837451934814453,
|
| 1610 |
+
"logps/rejected": -7.7740020751953125,
|
| 1611 |
+
"loss": 0.177,
|
| 1612 |
+
"rewards/accuracies": 1.0,
|
| 1613 |
+
"rewards/chosen": -5.075617790222168,
|
| 1614 |
+
"rewards/margins": 6.585384368896484,
|
| 1615 |
+
"rewards/rejected": -11.661002159118652,
|
| 1616 |
+
"step": 890
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.7492975335622853,
|
| 1620 |
+
"grad_norm": 3.8752946853637695,
|
| 1621 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1622 |
+
"logits/chosen": 3.2561440467834473,
|
| 1623 |
+
"logits/rejected": 3.13822603225708,
|
| 1624 |
+
"logps/chosen": -3.69258451461792,
|
| 1625 |
+
"logps/rejected": -7.472433567047119,
|
| 1626 |
+
"loss": 0.213,
|
| 1627 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1628 |
+
"rewards/chosen": -5.538876533508301,
|
| 1629 |
+
"rewards/margins": 5.669772624969482,
|
| 1630 |
+
"rewards/rejected": -11.208650588989258,
|
| 1631 |
+
"step": 900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 0.7492975335622853,
|
| 1635 |
+
"eval_logits/chosen": 2.8268215656280518,
|
| 1636 |
+
"eval_logits/rejected": 3.031662702560425,
|
| 1637 |
+
"eval_logps/chosen": -3.6311440467834473,
|
| 1638 |
+
"eval_logps/rejected": -8.067394256591797,
|
| 1639 |
+
"eval_loss": 0.23127013444900513,
|
| 1640 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1641 |
+
"eval_rewards/chosen": -5.44671630859375,
|
| 1642 |
+
"eval_rewards/margins": 6.654376029968262,
|
| 1643 |
+
"eval_rewards/rejected": -12.101091384887695,
|
| 1644 |
+
"eval_runtime": 29.022,
|
| 1645 |
+
"eval_samples_per_second": 26.773,
|
| 1646 |
+
"eval_steps_per_second": 3.377,
|
| 1647 |
+
"step": 900
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.7576230617129774,
|
| 1651 |
+
"grad_norm": 3.4024012088775635,
|
| 1652 |
+
"learning_rate": 1.677833383153542e-06,
|
| 1653 |
+
"logits/chosen": 2.0691773891448975,
|
| 1654 |
+
"logits/rejected": 2.190563201904297,
|
| 1655 |
+
"logps/chosen": -3.483668565750122,
|
| 1656 |
+
"logps/rejected": -8.020956039428711,
|
| 1657 |
+
"loss": 0.198,
|
| 1658 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1659 |
+
"rewards/chosen": -5.225502967834473,
|
| 1660 |
+
"rewards/margins": 6.805932521820068,
|
| 1661 |
+
"rewards/rejected": -12.0314359664917,
|
| 1662 |
+
"step": 910
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.7659485898636694,
|
| 1666 |
+
"grad_norm": 4.999133586883545,
|
| 1667 |
+
"learning_rate": 1.6285698816954626e-06,
|
| 1668 |
+
"logits/chosen": 2.4453094005584717,
|
| 1669 |
+
"logits/rejected": 2.440931558609009,
|
| 1670 |
+
"logps/chosen": -4.1138916015625,
|
| 1671 |
+
"logps/rejected": -8.617280960083008,
|
| 1672 |
+
"loss": 0.253,
|
| 1673 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1674 |
+
"rewards/chosen": -6.170836925506592,
|
| 1675 |
+
"rewards/margins": 6.7550835609436035,
|
| 1676 |
+
"rewards/rejected": -12.925920486450195,
|
| 1677 |
+
"step": 920
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.7742741180143615,
|
| 1681 |
+
"grad_norm": 3.1391687393188477,
|
| 1682 |
+
"learning_rate": 1.5796886182883053e-06,
|
| 1683 |
+
"logits/chosen": 2.892235517501831,
|
| 1684 |
+
"logits/rejected": 2.8754334449768066,
|
| 1685 |
+
"logps/chosen": -3.8762309551239014,
|
| 1686 |
+
"logps/rejected": -7.991665840148926,
|
| 1687 |
+
"loss": 0.2171,
|
| 1688 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1689 |
+
"rewards/chosen": -5.814346790313721,
|
| 1690 |
+
"rewards/margins": 6.173151969909668,
|
| 1691 |
+
"rewards/rejected": -11.98749828338623,
|
| 1692 |
+
"step": 930
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 0.7825996461650536,
|
| 1696 |
+
"grad_norm": 6.850193023681641,
|
| 1697 |
+
"learning_rate": 1.5312110338697427e-06,
|
| 1698 |
+
"logits/chosen": 3.0068447589874268,
|
| 1699 |
+
"logits/rejected": 3.0385780334472656,
|
| 1700 |
+
"logps/chosen": -3.7039177417755127,
|
| 1701 |
+
"logps/rejected": -8.53662109375,
|
| 1702 |
+
"loss": 0.1907,
|
| 1703 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1704 |
+
"rewards/chosen": -5.555876731872559,
|
| 1705 |
+
"rewards/margins": 7.2490553855896,
|
| 1706 |
+
"rewards/rejected": -12.804931640625,
|
| 1707 |
+
"step": 940
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 0.7909251743157456,
|
| 1711 |
+
"grad_norm": 16.202392578125,
|
| 1712 |
+
"learning_rate": 1.4831583923105e-06,
|
| 1713 |
+
"logits/chosen": 2.445254325866699,
|
| 1714 |
+
"logits/rejected": 2.6017098426818848,
|
| 1715 |
+
"logps/chosen": -4.0695037841796875,
|
| 1716 |
+
"logps/rejected": -8.545947074890137,
|
| 1717 |
+
"loss": 0.2033,
|
| 1718 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1719 |
+
"rewards/chosen": -6.104255676269531,
|
| 1720 |
+
"rewards/margins": 6.714664459228516,
|
| 1721 |
+
"rewards/rejected": -12.818921089172363,
|
| 1722 |
+
"step": 950
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.7909251743157456,
|
| 1726 |
+
"eval_logits/chosen": 2.7845335006713867,
|
| 1727 |
+
"eval_logits/rejected": 3.037020206451416,
|
| 1728 |
+
"eval_logps/chosen": -3.982541799545288,
|
| 1729 |
+
"eval_logps/rejected": -8.498592376708984,
|
| 1730 |
+
"eval_loss": 0.22774070501327515,
|
| 1731 |
+
"eval_rewards/accuracies": 0.9387755393981934,
|
| 1732 |
+
"eval_rewards/chosen": -5.973812580108643,
|
| 1733 |
+
"eval_rewards/margins": 6.77407693862915,
|
| 1734 |
+
"eval_rewards/rejected": -12.747888565063477,
|
| 1735 |
+
"eval_runtime": 29.0201,
|
| 1736 |
+
"eval_samples_per_second": 26.775,
|
| 1737 |
+
"eval_steps_per_second": 3.377,
|
| 1738 |
+
"step": 950
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.7992507024664377,
|
| 1742 |
+
"grad_norm": 4.31044864654541,
|
| 1743 |
+
"learning_rate": 1.4355517710873184e-06,
|
| 1744 |
+
"logits/chosen": 2.1485049724578857,
|
| 1745 |
+
"logits/rejected": 2.493374824523926,
|
| 1746 |
+
"logps/chosen": -3.8115482330322266,
|
| 1747 |
+
"logps/rejected": -8.553500175476074,
|
| 1748 |
+
"loss": 0.2109,
|
| 1749 |
+
"rewards/accuracies": 0.9375,
|
| 1750 |
+
"rewards/chosen": -5.71732234954834,
|
| 1751 |
+
"rewards/margins": 7.112928867340088,
|
| 1752 |
+
"rewards/rejected": -12.83025074005127,
|
| 1753 |
+
"step": 960
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.8075762306171298,
|
| 1757 |
+
"grad_norm": 4.177423000335693,
|
| 1758 |
+
"learning_rate": 1.388412052037682e-06,
|
| 1759 |
+
"logits/chosen": 2.9300179481506348,
|
| 1760 |
+
"logits/rejected": 2.9548909664154053,
|
| 1761 |
+
"logps/chosen": -3.9784176349639893,
|
| 1762 |
+
"logps/rejected": -8.308394432067871,
|
| 1763 |
+
"loss": 0.2012,
|
| 1764 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1765 |
+
"rewards/chosen": -5.967626094818115,
|
| 1766 |
+
"rewards/margins": 6.49496603012085,
|
| 1767 |
+
"rewards/rejected": -12.462592124938965,
|
| 1768 |
+
"step": 970
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.8159017587678218,
|
| 1772 |
+
"grad_norm": 4.683027744293213,
|
| 1773 |
+
"learning_rate": 1.3417599122003464e-06,
|
| 1774 |
+
"logits/chosen": 2.5800061225891113,
|
| 1775 |
+
"logits/rejected": 2.526090145111084,
|
| 1776 |
+
"logps/chosen": -3.86810564994812,
|
| 1777 |
+
"logps/rejected": -8.47614574432373,
|
| 1778 |
+
"loss": 0.2141,
|
| 1779 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1780 |
+
"rewards/chosen": -5.802158355712891,
|
| 1781 |
+
"rewards/margins": 6.9120612144470215,
|
| 1782 |
+
"rewards/rejected": -12.714218139648438,
|
| 1783 |
+
"step": 980
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 0.8242272869185139,
|
| 1787 |
+
"grad_norm": 3.7419984340667725,
|
| 1788 |
+
"learning_rate": 1.2956158147457116e-06,
|
| 1789 |
+
"logits/chosen": 3.4706058502197266,
|
| 1790 |
+
"logits/rejected": 3.4088757038116455,
|
| 1791 |
+
"logps/chosen": -4.216760158538818,
|
| 1792 |
+
"logps/rejected": -8.575207710266113,
|
| 1793 |
+
"loss": 0.2422,
|
| 1794 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1795 |
+
"rewards/chosen": -6.325140953063965,
|
| 1796 |
+
"rewards/margins": 6.537671089172363,
|
| 1797 |
+
"rewards/rejected": -12.862811088562012,
|
| 1798 |
+
"step": 990
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 0.832552815069206,
|
| 1802 |
+
"grad_norm": 8.953512191772461,
|
| 1803 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1804 |
+
"logits/chosen": 2.9276206493377686,
|
| 1805 |
+
"logits/rejected": 2.946265459060669,
|
| 1806 |
+
"logps/chosen": -3.9976966381073,
|
| 1807 |
+
"logps/rejected": -8.48410701751709,
|
| 1808 |
+
"loss": 0.2139,
|
| 1809 |
+
"rewards/accuracies": 0.9375,
|
| 1810 |
+
"rewards/chosen": -5.996545314788818,
|
| 1811 |
+
"rewards/margins": 6.729616641998291,
|
| 1812 |
+
"rewards/rejected": -12.726162910461426,
|
| 1813 |
+
"step": 1000
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.832552815069206,
|
| 1817 |
+
"eval_logits/chosen": 2.9153146743774414,
|
| 1818 |
+
"eval_logits/rejected": 3.0989012718200684,
|
| 1819 |
+
"eval_logps/chosen": -3.6678271293640137,
|
| 1820 |
+
"eval_logps/rejected": -8.173608779907227,
|
| 1821 |
+
"eval_loss": 0.22841480374336243,
|
| 1822 |
+
"eval_rewards/accuracies": 0.9285714030265808,
|
| 1823 |
+
"eval_rewards/chosen": -5.5017409324646,
|
| 1824 |
+
"eval_rewards/margins": 6.758671760559082,
|
| 1825 |
+
"eval_rewards/rejected": -12.26041316986084,
|
| 1826 |
+
"eval_runtime": 29.0504,
|
| 1827 |
+
"eval_samples_per_second": 26.747,
|
| 1828 |
+
"eval_steps_per_second": 3.373,
|
| 1829 |
+
"step": 1000
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.840878343219898,
|
| 1833 |
+
"grad_norm": 4.267103672027588,
|
| 1834 |
+
"learning_rate": 1.204932476567175e-06,
|
| 1835 |
+
"logits/chosen": 2.3057751655578613,
|
| 1836 |
+
"logits/rejected": 2.3750340938568115,
|
| 1837 |
+
"logps/chosen": -3.576403856277466,
|
| 1838 |
+
"logps/rejected": -8.301278114318848,
|
| 1839 |
+
"loss": 0.2211,
|
| 1840 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1841 |
+
"rewards/chosen": -5.364605903625488,
|
| 1842 |
+
"rewards/margins": 7.087311744689941,
|
| 1843 |
+
"rewards/rejected": -12.45191764831543,
|
| 1844 |
+
"step": 1010
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.8492038713705901,
|
| 1848 |
+
"grad_norm": 6.008708477020264,
|
| 1849 |
+
"learning_rate": 1.160433012552508e-06,
|
| 1850 |
+
"logits/chosen": 3.0167624950408936,
|
| 1851 |
+
"logits/rejected": 2.817478895187378,
|
| 1852 |
+
"logps/chosen": -4.053152084350586,
|
| 1853 |
+
"logps/rejected": -8.841009140014648,
|
| 1854 |
+
"loss": 0.195,
|
| 1855 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1856 |
+
"rewards/chosen": -6.079728126525879,
|
| 1857 |
+
"rewards/margins": 7.181784152984619,
|
| 1858 |
+
"rewards/rejected": -13.261512756347656,
|
| 1859 |
+
"step": 1020
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.8575293995212822,
|
| 1863 |
+
"grad_norm": 3.7652032375335693,
|
| 1864 |
+
"learning_rate": 1.11652112689164e-06,
|
| 1865 |
+
"logits/chosen": 2.3387794494628906,
|
| 1866 |
+
"logits/rejected": 2.420820474624634,
|
| 1867 |
+
"logps/chosen": -4.114675045013428,
|
| 1868 |
+
"logps/rejected": -8.801934242248535,
|
| 1869 |
+
"loss": 0.2226,
|
| 1870 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1871 |
+
"rewards/chosen": -6.172013282775879,
|
| 1872 |
+
"rewards/margins": 7.030886650085449,
|
| 1873 |
+
"rewards/rejected": -13.202900886535645,
|
| 1874 |
+
"step": 1030
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.8658549276719741,
|
| 1878 |
+
"grad_norm": 3.811018466949463,
|
| 1879 |
+
"learning_rate": 1.073216080788921e-06,
|
| 1880 |
+
"logits/chosen": 3.4545624256134033,
|
| 1881 |
+
"logits/rejected": 2.934145212173462,
|
| 1882 |
+
"logps/chosen": -3.841254472732544,
|
| 1883 |
+
"logps/rejected": -8.547441482543945,
|
| 1884 |
+
"loss": 0.2039,
|
| 1885 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1886 |
+
"rewards/chosen": -5.7618818283081055,
|
| 1887 |
+
"rewards/margins": 7.059278964996338,
|
| 1888 |
+
"rewards/rejected": -12.821161270141602,
|
| 1889 |
+
"step": 1040
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.8741804558226662,
|
| 1893 |
+
"grad_norm": 3.5039620399475098,
|
| 1894 |
+
"learning_rate": 1.0305368692688175e-06,
|
| 1895 |
+
"logits/chosen": 1.9293429851531982,
|
| 1896 |
+
"logits/rejected": 2.530273914337158,
|
| 1897 |
+
"logps/chosen": -3.6742749214172363,
|
| 1898 |
+
"logps/rejected": -8.751821517944336,
|
| 1899 |
+
"loss": 0.2168,
|
| 1900 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1901 |
+
"rewards/chosen": -5.511412143707275,
|
| 1902 |
+
"rewards/margins": 7.616321563720703,
|
| 1903 |
+
"rewards/rejected": -13.127734184265137,
|
| 1904 |
+
"step": 1050
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.8741804558226662,
|
| 1908 |
+
"eval_logits/chosen": 2.9263997077941895,
|
| 1909 |
+
"eval_logits/rejected": 3.1277804374694824,
|
| 1910 |
+
"eval_logps/chosen": -3.6399991512298584,
|
| 1911 |
+
"eval_logps/rejected": -8.258426666259766,
|
| 1912 |
+
"eval_loss": 0.2207891196012497,
|
| 1913 |
+
"eval_rewards/accuracies": 0.9387755393981934,
|
| 1914 |
+
"eval_rewards/chosen": -5.45999813079834,
|
| 1915 |
+
"eval_rewards/margins": 6.92764139175415,
|
| 1916 |
+
"eval_rewards/rejected": -12.387639045715332,
|
| 1917 |
+
"eval_runtime": 29.0247,
|
| 1918 |
+
"eval_samples_per_second": 26.77,
|
| 1919 |
+
"eval_steps_per_second": 3.376,
|
| 1920 |
+
"step": 1050
|
| 1921 |
+
}
|
| 1922 |
+
],
|
| 1923 |
+
"logging_steps": 10,
|
| 1924 |
+
"max_steps": 1500,
|
| 1925 |
+
"num_input_tokens_seen": 0,
|
| 1926 |
+
"num_train_epochs": 2,
|
| 1927 |
+
"save_steps": 50,
|
| 1928 |
+
"stateful_callbacks": {
|
| 1929 |
+
"TrainerControl": {
|
| 1930 |
+
"args": {
|
| 1931 |
+
"should_epoch_stop": false,
|
| 1932 |
+
"should_evaluate": false,
|
| 1933 |
+
"should_log": false,
|
| 1934 |
+
"should_save": true,
|
| 1935 |
+
"should_training_stop": false
|
| 1936 |
+
},
|
| 1937 |
+
"attributes": {}
|
| 1938 |
+
}
|
| 1939 |
+
},
|
| 1940 |
+
"total_flos": 2.5581316773983027e+18,
|
| 1941 |
+
"train_batch_size": 1,
|
| 1942 |
+
"trial_name": null,
|
| 1943 |
+
"trial_params": null
|
| 1944 |
+
}
|
checkpoint-1050/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6e578c4f30caab267e8b1a0bd9aeabf3d73f244625cf90ce2237f0ece42d273
|
| 3 |
+
size 7224
|
checkpoint-1050/zero_to_fp32.py
ADDED
|
@@ -0,0 +1,674 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python
|
| 2 |
+
|
| 3 |
+
# Copyright (c) Microsoft Corporation.
|
| 4 |
+
# SPDX-License-Identifier: Apache-2.0
|
| 5 |
+
|
| 6 |
+
# DeepSpeed Team
|
| 7 |
+
|
| 8 |
+
# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
|
| 9 |
+
# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
|
| 10 |
+
# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
|
| 11 |
+
# application.
|
| 12 |
+
#
|
| 13 |
+
# example:
|
| 14 |
+
# python zero_to_fp32.py . output_dir/
|
| 15 |
+
# or
|
| 16 |
+
# python zero_to_fp32.py . output_dir/ --safe_serialization
|
| 17 |
+
|
| 18 |
+
import argparse
|
| 19 |
+
import torch
|
| 20 |
+
import glob
|
| 21 |
+
import math
|
| 22 |
+
import os
|
| 23 |
+
import re
|
| 24 |
+
import json
|
| 25 |
+
from tqdm import tqdm
|
| 26 |
+
from collections import OrderedDict
|
| 27 |
+
from dataclasses import dataclass
|
| 28 |
+
|
| 29 |
+
# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
|
| 30 |
+
# DeepSpeed data structures it has to be available in the current python environment.
|
| 31 |
+
from deepspeed.utils import logger
|
| 32 |
+
from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
|
| 33 |
+
FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
|
| 34 |
+
FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
@dataclass
|
| 38 |
+
class zero_model_state:
|
| 39 |
+
buffers: dict()
|
| 40 |
+
param_shapes: dict()
|
| 41 |
+
shared_params: list
|
| 42 |
+
ds_version: int
|
| 43 |
+
frozen_param_shapes: dict()
|
| 44 |
+
frozen_param_fragments: dict()
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
debug = 0
|
| 48 |
+
|
| 49 |
+
# load to cpu
|
| 50 |
+
device = torch.device('cpu')
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
def atoi(text):
|
| 54 |
+
return int(text) if text.isdigit() else text
|
| 55 |
+
|
| 56 |
+
|
| 57 |
+
def natural_keys(text):
|
| 58 |
+
'''
|
| 59 |
+
alist.sort(key=natural_keys) sorts in human order
|
| 60 |
+
http://nedbatchelder.com/blog/200712/human_sorting.html
|
| 61 |
+
(See Toothy's implementation in the comments)
|
| 62 |
+
'''
|
| 63 |
+
return [atoi(c) for c in re.split(r'(\d+)', text)]
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
def get_model_state_file(checkpoint_dir, zero_stage):
|
| 67 |
+
if not os.path.isdir(checkpoint_dir):
|
| 68 |
+
raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
|
| 69 |
+
|
| 70 |
+
# there should be only one file
|
| 71 |
+
if zero_stage <= 2:
|
| 72 |
+
file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
|
| 73 |
+
elif zero_stage == 3:
|
| 74 |
+
file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
|
| 75 |
+
|
| 76 |
+
if not os.path.exists(file):
|
| 77 |
+
raise FileNotFoundError(f"can't find model states file at '{file}'")
|
| 78 |
+
|
| 79 |
+
return file
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
def get_checkpoint_files(checkpoint_dir, glob_pattern):
|
| 83 |
+
# XXX: need to test that this simple glob rule works for multi-node setup too
|
| 84 |
+
ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
|
| 85 |
+
|
| 86 |
+
if len(ckpt_files) == 0:
|
| 87 |
+
raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
|
| 88 |
+
|
| 89 |
+
return ckpt_files
|
| 90 |
+
|
| 91 |
+
|
| 92 |
+
def get_optim_files(checkpoint_dir):
|
| 93 |
+
return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
def get_model_state_files(checkpoint_dir):
|
| 97 |
+
return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
|
| 98 |
+
|
| 99 |
+
|
| 100 |
+
def parse_model_states(files):
|
| 101 |
+
zero_model_states = []
|
| 102 |
+
for file in files:
|
| 103 |
+
state_dict = torch.load(file, map_location=device)
|
| 104 |
+
|
| 105 |
+
if BUFFER_NAMES not in state_dict:
|
| 106 |
+
raise ValueError(f"{file} is not a model state checkpoint")
|
| 107 |
+
buffer_names = state_dict[BUFFER_NAMES]
|
| 108 |
+
if debug:
|
| 109 |
+
print("Found buffers:", buffer_names)
|
| 110 |
+
|
| 111 |
+
# recover just the buffers while restoring them to fp32 if they were saved in fp16
|
| 112 |
+
buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
|
| 113 |
+
param_shapes = state_dict[PARAM_SHAPES]
|
| 114 |
+
|
| 115 |
+
# collect parameters that are included in param_shapes
|
| 116 |
+
param_names = []
|
| 117 |
+
for s in param_shapes:
|
| 118 |
+
for name in s.keys():
|
| 119 |
+
param_names.append(name)
|
| 120 |
+
|
| 121 |
+
# update with frozen parameters
|
| 122 |
+
frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
|
| 123 |
+
if frozen_param_shapes is not None:
|
| 124 |
+
if debug:
|
| 125 |
+
print(f"Found frozen_param_shapes: {frozen_param_shapes}")
|
| 126 |
+
param_names += list(frozen_param_shapes.keys())
|
| 127 |
+
|
| 128 |
+
# handle shared params
|
| 129 |
+
shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
|
| 130 |
+
|
| 131 |
+
ds_version = state_dict.get(DS_VERSION, None)
|
| 132 |
+
|
| 133 |
+
frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
|
| 134 |
+
|
| 135 |
+
z_model_state = zero_model_state(buffers=buffers,
|
| 136 |
+
param_shapes=param_shapes,
|
| 137 |
+
shared_params=shared_params,
|
| 138 |
+
ds_version=ds_version,
|
| 139 |
+
frozen_param_shapes=frozen_param_shapes,
|
| 140 |
+
frozen_param_fragments=frozen_param_fragments)
|
| 141 |
+
zero_model_states.append(z_model_state)
|
| 142 |
+
|
| 143 |
+
return zero_model_states
|
| 144 |
+
|
| 145 |
+
|
| 146 |
+
def parse_optim_states(files, ds_checkpoint_dir):
|
| 147 |
+
total_files = len(files)
|
| 148 |
+
state_dicts = []
|
| 149 |
+
for f in files:
|
| 150 |
+
state_dict = torch.load(f, map_location=device)
|
| 151 |
+
# immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
|
| 152 |
+
# and also handle the case where it was already removed by another helper script
|
| 153 |
+
state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
|
| 154 |
+
state_dicts.append(state_dict)
|
| 155 |
+
|
| 156 |
+
if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
|
| 157 |
+
raise ValueError(f"{files[0]} is not a zero checkpoint")
|
| 158 |
+
zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
|
| 159 |
+
world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
|
| 160 |
+
|
| 161 |
+
# For ZeRO-2 each param group can have different partition_count as data parallelism for expert
|
| 162 |
+
# parameters can be different from data parallelism for non-expert parameters. So we can just
|
| 163 |
+
# use the max of the partition_count to get the dp world_size.
|
| 164 |
+
|
| 165 |
+
if type(world_size) is list:
|
| 166 |
+
world_size = max(world_size)
|
| 167 |
+
|
| 168 |
+
if world_size != total_files:
|
| 169 |
+
raise ValueError(
|
| 170 |
+
f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
|
| 171 |
+
"Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
|
| 172 |
+
)
|
| 173 |
+
|
| 174 |
+
# the groups are named differently in each stage
|
| 175 |
+
if zero_stage <= 2:
|
| 176 |
+
fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
|
| 177 |
+
elif zero_stage == 3:
|
| 178 |
+
fp32_groups_key = FP32_FLAT_GROUPS
|
| 179 |
+
else:
|
| 180 |
+
raise ValueError(f"unknown zero stage {zero_stage}")
|
| 181 |
+
|
| 182 |
+
if zero_stage <= 2:
|
| 183 |
+
fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
|
| 184 |
+
elif zero_stage == 3:
|
| 185 |
+
# if there is more than one param group, there will be multiple flattened tensors - one
|
| 186 |
+
# flattened tensor per group - for simplicity merge them into a single tensor
|
| 187 |
+
#
|
| 188 |
+
# XXX: could make the script more memory efficient for when there are multiple groups - it
|
| 189 |
+
# will require matching the sub-lists of param_shapes for each param group flattened tensor
|
| 190 |
+
|
| 191 |
+
fp32_flat_groups = [
|
| 192 |
+
torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
|
| 193 |
+
]
|
| 194 |
+
|
| 195 |
+
return zero_stage, world_size, fp32_flat_groups
|
| 196 |
+
|
| 197 |
+
|
| 198 |
+
def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
|
| 199 |
+
"""
|
| 200 |
+
Returns fp32 state_dict reconstructed from ds checkpoint
|
| 201 |
+
|
| 202 |
+
Args:
|
| 203 |
+
- ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
|
| 204 |
+
|
| 205 |
+
"""
|
| 206 |
+
print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
|
| 207 |
+
|
| 208 |
+
optim_files = get_optim_files(ds_checkpoint_dir)
|
| 209 |
+
zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
|
| 210 |
+
print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
|
| 211 |
+
|
| 212 |
+
model_files = get_model_state_files(ds_checkpoint_dir)
|
| 213 |
+
|
| 214 |
+
zero_model_states = parse_model_states(model_files)
|
| 215 |
+
print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
|
| 216 |
+
|
| 217 |
+
if zero_stage <= 2:
|
| 218 |
+
return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 219 |
+
exclude_frozen_parameters)
|
| 220 |
+
elif zero_stage == 3:
|
| 221 |
+
return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 222 |
+
exclude_frozen_parameters)
|
| 223 |
+
|
| 224 |
+
|
| 225 |
+
def _zero2_merge_frozen_params(state_dict, zero_model_states):
|
| 226 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 227 |
+
return
|
| 228 |
+
|
| 229 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 230 |
+
frozen_param_fragments = zero_model_states[0].frozen_param_fragments
|
| 231 |
+
|
| 232 |
+
if debug:
|
| 233 |
+
num_elem = sum(s.numel() for s in frozen_param_shapes.values())
|
| 234 |
+
print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 235 |
+
|
| 236 |
+
wanted_params = len(frozen_param_shapes)
|
| 237 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 238 |
+
avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
|
| 239 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 240 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 241 |
+
|
| 242 |
+
total_params = 0
|
| 243 |
+
total_numel = 0
|
| 244 |
+
for name, shape in frozen_param_shapes.items():
|
| 245 |
+
total_params += 1
|
| 246 |
+
unpartitioned_numel = shape.numel()
|
| 247 |
+
total_numel += unpartitioned_numel
|
| 248 |
+
|
| 249 |
+
state_dict[name] = frozen_param_fragments[name]
|
| 250 |
+
|
| 251 |
+
if debug:
|
| 252 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 253 |
+
|
| 254 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 255 |
+
|
| 256 |
+
|
| 257 |
+
def _has_callable(obj, fn):
|
| 258 |
+
attr = getattr(obj, fn, None)
|
| 259 |
+
return callable(attr)
|
| 260 |
+
|
| 261 |
+
|
| 262 |
+
def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 263 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 264 |
+
|
| 265 |
+
# Reconstruction protocol:
|
| 266 |
+
#
|
| 267 |
+
# XXX: document this
|
| 268 |
+
|
| 269 |
+
if debug:
|
| 270 |
+
for i in range(world_size):
|
| 271 |
+
for j in range(len(fp32_flat_groups[0])):
|
| 272 |
+
print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
|
| 273 |
+
|
| 274 |
+
# XXX: memory usage doubles here (zero2)
|
| 275 |
+
num_param_groups = len(fp32_flat_groups[0])
|
| 276 |
+
merged_single_partition_of_fp32_groups = []
|
| 277 |
+
for i in range(num_param_groups):
|
| 278 |
+
merged_partitions = [sd[i] for sd in fp32_flat_groups]
|
| 279 |
+
full_single_fp32_vector = torch.cat(merged_partitions, 0)
|
| 280 |
+
merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
|
| 281 |
+
avail_numel = sum(
|
| 282 |
+
[full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
|
| 283 |
+
|
| 284 |
+
if debug:
|
| 285 |
+
wanted_params = sum([len(shapes) for shapes in param_shapes])
|
| 286 |
+
wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
|
| 287 |
+
# not asserting if there is a mismatch due to possible padding
|
| 288 |
+
print(f"Have {avail_numel} numels to process.")
|
| 289 |
+
print(f"Need {wanted_numel} numels in {wanted_params} params.")
|
| 290 |
+
|
| 291 |
+
# params
|
| 292 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 293 |
+
# out-of-core computing solution
|
| 294 |
+
total_numel = 0
|
| 295 |
+
total_params = 0
|
| 296 |
+
for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
|
| 297 |
+
offset = 0
|
| 298 |
+
avail_numel = full_single_fp32_vector.numel()
|
| 299 |
+
for name, shape in shapes.items():
|
| 300 |
+
|
| 301 |
+
unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
|
| 302 |
+
total_numel += unpartitioned_numel
|
| 303 |
+
total_params += 1
|
| 304 |
+
|
| 305 |
+
if debug:
|
| 306 |
+
print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
|
| 307 |
+
state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
|
| 308 |
+
offset += unpartitioned_numel
|
| 309 |
+
|
| 310 |
+
# Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
|
| 311 |
+
# avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
|
| 312 |
+
# paddings performed in the code it's almost impossible to predict the exact numbers w/o the
|
| 313 |
+
# live optimizer object, so we are checking that the numbers are within the right range
|
| 314 |
+
align_to = 2 * world_size
|
| 315 |
+
|
| 316 |
+
def zero2_align(x):
|
| 317 |
+
return align_to * math.ceil(x / align_to)
|
| 318 |
+
|
| 319 |
+
if debug:
|
| 320 |
+
print(f"original offset={offset}, avail_numel={avail_numel}")
|
| 321 |
+
|
| 322 |
+
offset = zero2_align(offset)
|
| 323 |
+
avail_numel = zero2_align(avail_numel)
|
| 324 |
+
|
| 325 |
+
if debug:
|
| 326 |
+
print(f"aligned offset={offset}, avail_numel={avail_numel}")
|
| 327 |
+
|
| 328 |
+
# Sanity check
|
| 329 |
+
if offset != avail_numel:
|
| 330 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 331 |
+
|
| 332 |
+
print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
|
| 333 |
+
|
| 334 |
+
|
| 335 |
+
def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 336 |
+
exclude_frozen_parameters):
|
| 337 |
+
state_dict = OrderedDict()
|
| 338 |
+
|
| 339 |
+
# buffers
|
| 340 |
+
buffers = zero_model_states[0].buffers
|
| 341 |
+
state_dict.update(buffers)
|
| 342 |
+
if debug:
|
| 343 |
+
print(f"added {len(buffers)} buffers")
|
| 344 |
+
|
| 345 |
+
if not exclude_frozen_parameters:
|
| 346 |
+
_zero2_merge_frozen_params(state_dict, zero_model_states)
|
| 347 |
+
|
| 348 |
+
_zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 349 |
+
|
| 350 |
+
# recover shared parameters
|
| 351 |
+
for pair in zero_model_states[0].shared_params:
|
| 352 |
+
if pair[1] in state_dict:
|
| 353 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 354 |
+
|
| 355 |
+
return state_dict
|
| 356 |
+
|
| 357 |
+
|
| 358 |
+
def zero3_partitioned_param_info(unpartitioned_numel, world_size):
|
| 359 |
+
remainder = unpartitioned_numel % world_size
|
| 360 |
+
padding_numel = (world_size - remainder) if remainder else 0
|
| 361 |
+
partitioned_numel = math.ceil(unpartitioned_numel / world_size)
|
| 362 |
+
return partitioned_numel, padding_numel
|
| 363 |
+
|
| 364 |
+
|
| 365 |
+
def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
|
| 366 |
+
if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
|
| 367 |
+
return
|
| 368 |
+
|
| 369 |
+
if debug:
|
| 370 |
+
for i in range(world_size):
|
| 371 |
+
num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
|
| 372 |
+
print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
|
| 373 |
+
|
| 374 |
+
frozen_param_shapes = zero_model_states[0].frozen_param_shapes
|
| 375 |
+
wanted_params = len(frozen_param_shapes)
|
| 376 |
+
wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
|
| 377 |
+
avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
|
| 378 |
+
print(f'Frozen params: Have {avail_numel} numels to process.')
|
| 379 |
+
print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
|
| 380 |
+
|
| 381 |
+
total_params = 0
|
| 382 |
+
total_numel = 0
|
| 383 |
+
for name, shape in zero_model_states[0].frozen_param_shapes.items():
|
| 384 |
+
total_params += 1
|
| 385 |
+
unpartitioned_numel = shape.numel()
|
| 386 |
+
total_numel += unpartitioned_numel
|
| 387 |
+
|
| 388 |
+
param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
|
| 389 |
+
state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 390 |
+
|
| 391 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 392 |
+
|
| 393 |
+
if debug:
|
| 394 |
+
print(
|
| 395 |
+
f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 396 |
+
)
|
| 397 |
+
|
| 398 |
+
print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
|
| 399 |
+
|
| 400 |
+
|
| 401 |
+
def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
|
| 402 |
+
param_shapes = zero_model_states[0].param_shapes
|
| 403 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 404 |
+
# Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
|
| 405 |
+
# param, re-consolidating each param, while dealing with padding if any
|
| 406 |
+
|
| 407 |
+
# merge list of dicts, preserving order
|
| 408 |
+
param_shapes = {k: v for d in param_shapes for k, v in d.items()}
|
| 409 |
+
|
| 410 |
+
if debug:
|
| 411 |
+
for i in range(world_size):
|
| 412 |
+
print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
|
| 413 |
+
|
| 414 |
+
wanted_params = len(param_shapes)
|
| 415 |
+
wanted_numel = sum(shape.numel() for shape in param_shapes.values())
|
| 416 |
+
# not asserting if there is a mismatch due to possible padding
|
| 417 |
+
avail_numel = fp32_flat_groups[0].numel() * world_size
|
| 418 |
+
print(f"Trainable params: Have {avail_numel} numels to process.")
|
| 419 |
+
print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
|
| 420 |
+
|
| 421 |
+
# params
|
| 422 |
+
# XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
|
| 423 |
+
# out-of-core computing solution
|
| 424 |
+
offset = 0
|
| 425 |
+
total_numel = 0
|
| 426 |
+
total_params = 0
|
| 427 |
+
for name, shape in tqdm(param_shapes.items(), desc='Gathering Sharded Weights'):
|
| 428 |
+
unpartitioned_numel = shape.numel()
|
| 429 |
+
total_numel += unpartitioned_numel
|
| 430 |
+
total_params += 1
|
| 431 |
+
partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
|
| 432 |
+
|
| 433 |
+
if debug:
|
| 434 |
+
print(
|
| 435 |
+
f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
|
| 436 |
+
)
|
| 437 |
+
|
| 438 |
+
# XXX: memory usage doubles here
|
| 439 |
+
state_dict[name] = torch.cat(
|
| 440 |
+
tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
|
| 441 |
+
0).narrow(0, 0, unpartitioned_numel).view(shape)
|
| 442 |
+
offset += partitioned_numel
|
| 443 |
+
|
| 444 |
+
offset *= world_size
|
| 445 |
+
|
| 446 |
+
# Sanity check
|
| 447 |
+
if offset != avail_numel:
|
| 448 |
+
raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
|
| 449 |
+
|
| 450 |
+
print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
|
| 451 |
+
|
| 452 |
+
|
| 453 |
+
def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
|
| 454 |
+
exclude_frozen_parameters):
|
| 455 |
+
state_dict = OrderedDict()
|
| 456 |
+
|
| 457 |
+
# buffers
|
| 458 |
+
buffers = zero_model_states[0].buffers
|
| 459 |
+
state_dict.update(buffers)
|
| 460 |
+
if debug:
|
| 461 |
+
print(f"added {len(buffers)} buffers")
|
| 462 |
+
|
| 463 |
+
if not exclude_frozen_parameters:
|
| 464 |
+
_zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
|
| 465 |
+
|
| 466 |
+
_zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
|
| 467 |
+
|
| 468 |
+
# recover shared parameters
|
| 469 |
+
for pair in zero_model_states[0].shared_params:
|
| 470 |
+
if pair[1] in state_dict:
|
| 471 |
+
state_dict[pair[0]] = state_dict[pair[1]]
|
| 472 |
+
|
| 473 |
+
return state_dict
|
| 474 |
+
|
| 475 |
+
|
| 476 |
+
def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
|
| 477 |
+
"""
|
| 478 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
|
| 479 |
+
``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
|
| 480 |
+
via a model hub.
|
| 481 |
+
|
| 482 |
+
Args:
|
| 483 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder
|
| 484 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
|
| 485 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 486 |
+
|
| 487 |
+
Returns:
|
| 488 |
+
- pytorch ``state_dict``
|
| 489 |
+
|
| 490 |
+
Note: this approach may not work if your application doesn't have sufficient free CPU memory and
|
| 491 |
+
you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
|
| 492 |
+
the checkpoint.
|
| 493 |
+
|
| 494 |
+
A typical usage might be ::
|
| 495 |
+
|
| 496 |
+
from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
|
| 497 |
+
# do the training and checkpoint saving
|
| 498 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
|
| 499 |
+
model = model.cpu() # move to cpu
|
| 500 |
+
model.load_state_dict(state_dict)
|
| 501 |
+
# submit to model hub or save the model to share with others
|
| 502 |
+
|
| 503 |
+
In this example the ``model`` will no longer be usable in the deepspeed context of the same
|
| 504 |
+
application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 505 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 506 |
+
|
| 507 |
+
If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
|
| 508 |
+
|
| 509 |
+
"""
|
| 510 |
+
if tag is None:
|
| 511 |
+
latest_path = os.path.join(checkpoint_dir, 'latest')
|
| 512 |
+
if os.path.isfile(latest_path):
|
| 513 |
+
with open(latest_path, 'r') as fd:
|
| 514 |
+
tag = fd.read().strip()
|
| 515 |
+
else:
|
| 516 |
+
raise ValueError(f"Unable to find 'latest' file at {latest_path}")
|
| 517 |
+
|
| 518 |
+
ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
|
| 519 |
+
|
| 520 |
+
if not os.path.isdir(ds_checkpoint_dir):
|
| 521 |
+
raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
|
| 522 |
+
|
| 523 |
+
return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
|
| 524 |
+
|
| 525 |
+
|
| 526 |
+
def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
|
| 527 |
+
output_dir,
|
| 528 |
+
max_shard_size="5GB",
|
| 529 |
+
safe_serialization=False,
|
| 530 |
+
tag=None,
|
| 531 |
+
exclude_frozen_parameters=False):
|
| 532 |
+
"""
|
| 533 |
+
Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
|
| 534 |
+
loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
|
| 535 |
+
|
| 536 |
+
Args:
|
| 537 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 538 |
+
- ``output_dir``: directory to the pytorch fp32 state_dict output files
|
| 539 |
+
- ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
|
| 540 |
+
- ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
|
| 541 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 542 |
+
- ``exclude_frozen_parameters``: exclude frozen parameters
|
| 543 |
+
"""
|
| 544 |
+
# Dependency pre-check
|
| 545 |
+
if safe_serialization:
|
| 546 |
+
try:
|
| 547 |
+
from safetensors.torch import save_file
|
| 548 |
+
except ImportError:
|
| 549 |
+
print('If you want to use `safe_serialization`, please `pip install safetensors`')
|
| 550 |
+
raise
|
| 551 |
+
if max_shard_size is not None:
|
| 552 |
+
try:
|
| 553 |
+
from huggingface_hub import split_torch_state_dict_into_shards
|
| 554 |
+
except ImportError:
|
| 555 |
+
print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
|
| 556 |
+
raise
|
| 557 |
+
|
| 558 |
+
# Convert zero checkpoint to state_dict
|
| 559 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
|
| 560 |
+
|
| 561 |
+
# Shard the model if it is too big.
|
| 562 |
+
weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
|
| 563 |
+
if max_shard_size is not None:
|
| 564 |
+
filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
|
| 565 |
+
state_dict_split = split_torch_state_dict_into_shards(state_dict,
|
| 566 |
+
filename_pattern=filename_pattern,
|
| 567 |
+
max_shard_size=max_shard_size)
|
| 568 |
+
else:
|
| 569 |
+
from collections import namedtuple
|
| 570 |
+
StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
|
| 571 |
+
state_dict_split = StateDictSplit(is_sharded=False,
|
| 572 |
+
filename_to_tensors={weights_name: list(state_dict.keys())})
|
| 573 |
+
|
| 574 |
+
# Save the model
|
| 575 |
+
filename_to_tensors = state_dict_split.filename_to_tensors.items()
|
| 576 |
+
for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
|
| 577 |
+
shard = {tensor: state_dict[tensor].contiguous() for tensor in tensors}
|
| 578 |
+
output_path = os.path.join(output_dir, shard_file)
|
| 579 |
+
if safe_serialization:
|
| 580 |
+
save_file(shard, output_path, metadata={"format": "pt"})
|
| 581 |
+
else:
|
| 582 |
+
torch.save(shard, output_path)
|
| 583 |
+
|
| 584 |
+
# Save index if sharded
|
| 585 |
+
if state_dict_split.is_sharded:
|
| 586 |
+
index = {
|
| 587 |
+
"metadata": state_dict_split.metadata,
|
| 588 |
+
"weight_map": state_dict_split.tensor_to_filename,
|
| 589 |
+
}
|
| 590 |
+
save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
|
| 591 |
+
save_index_file = os.path.join(output_dir, save_index_file)
|
| 592 |
+
with open(save_index_file, "w", encoding="utf-8") as f:
|
| 593 |
+
content = json.dumps(index, indent=2, sort_keys=True) + "\n"
|
| 594 |
+
f.write(content)
|
| 595 |
+
|
| 596 |
+
|
| 597 |
+
def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
|
| 598 |
+
"""
|
| 599 |
+
1. Put the provided model to cpu
|
| 600 |
+
2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
|
| 601 |
+
3. Load it into the provided model
|
| 602 |
+
|
| 603 |
+
Args:
|
| 604 |
+
- ``model``: the model object to update
|
| 605 |
+
- ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
|
| 606 |
+
- ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
|
| 607 |
+
|
| 608 |
+
Returns:
|
| 609 |
+
- ``model`: modified model
|
| 610 |
+
|
| 611 |
+
Make sure you have plenty of CPU memory available before you call this function. If you don't
|
| 612 |
+
have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
|
| 613 |
+
conveniently placed for you in the checkpoint folder.
|
| 614 |
+
|
| 615 |
+
A typical usage might be ::
|
| 616 |
+
|
| 617 |
+
from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
|
| 618 |
+
model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
|
| 619 |
+
# submit to model hub or save the model to share with others
|
| 620 |
+
|
| 621 |
+
Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
|
| 622 |
+
of the same application. i.e. you will need to re-initialize the deepspeed engine, since
|
| 623 |
+
``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
|
| 624 |
+
|
| 625 |
+
"""
|
| 626 |
+
logger.info(f"Extracting fp32 weights")
|
| 627 |
+
state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
|
| 628 |
+
|
| 629 |
+
logger.info(f"Overwriting model with fp32 weights")
|
| 630 |
+
model = model.cpu()
|
| 631 |
+
model.load_state_dict(state_dict, strict=False)
|
| 632 |
+
|
| 633 |
+
return model
|
| 634 |
+
|
| 635 |
+
|
| 636 |
+
if __name__ == "__main__":
|
| 637 |
+
parser = argparse.ArgumentParser()
|
| 638 |
+
parser.add_argument("checkpoint_dir",
|
| 639 |
+
type=str,
|
| 640 |
+
help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
|
| 641 |
+
parser.add_argument("output_dir",
|
| 642 |
+
type=str,
|
| 643 |
+
help="directory to the pytorch fp32 state_dict output files"
|
| 644 |
+
"(e.g. path/checkpoint-12-output/)")
|
| 645 |
+
parser.add_argument(
|
| 646 |
+
"--max_shard_size",
|
| 647 |
+
type=str,
|
| 648 |
+
default="5GB",
|
| 649 |
+
help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
|
| 650 |
+
"lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
|
| 651 |
+
"We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
|
| 652 |
+
"without CPU OOM issues.")
|
| 653 |
+
parser.add_argument(
|
| 654 |
+
"--safe_serialization",
|
| 655 |
+
default=False,
|
| 656 |
+
action='store_true',
|
| 657 |
+
help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
|
| 658 |
+
parser.add_argument("-t",
|
| 659 |
+
"--tag",
|
| 660 |
+
type=str,
|
| 661 |
+
default=None,
|
| 662 |
+
help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
|
| 663 |
+
parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
|
| 664 |
+
parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
|
| 665 |
+
args = parser.parse_args()
|
| 666 |
+
|
| 667 |
+
debug = args.debug
|
| 668 |
+
|
| 669 |
+
convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
|
| 670 |
+
args.output_dir,
|
| 671 |
+
max_shard_size=args.max_shard_size,
|
| 672 |
+
safe_serialization=args.safe_serialization,
|
| 673 |
+
tag=args.tag,
|
| 674 |
+
exclude_frozen_parameters=args.exclude_frozen_parameters)
|