Downloads · 30 days
2
7% of all-time downloads
model_hh_usp3_dpo5
guoyu-zhang/model_hh_usp3_dpo5
model_hh_usp3_dpo5 is a machine learning model from guoyu-zhang. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft.
should probably proofread and complete it, then remove this comment. --
All-time downloads
27
Public
Repo size
12.6 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors25.2 MB · 93%
At a glance
- Library
- peft
- Access
- Public
- Created
- Apr 16, 2024
- Updated
- Apr 16, 2024
- SHA
- e9e0c757
Base models
Model card
From the Hugging Face model README
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
model_hh_usp3_dpo5
This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.6259
- Rewards/chosen: -6.2124
- Rewards/rejected: -11.7257
- Rewards/accuracies: 0.7200
- Rewards/margins: 5.5132
- Logps/rejected: -138.8036
- Logps/chosen: -128.0061
- Logits/rejected: -0.4196
- Logits/chosen: -0.3295
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 4
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0345 | 2.67 | 100 | 0.8548 | -2.8854 | -6.4253 | 0.7400 | 3.5400 | -128.2029 | -121.3520 | -0.2554 | -0.2032 |
| 0.012 | 5.33 | 200 | 1.2021 | -5.9535 | -11.1573 | 0.7700 | 5.2039 | -137.6669 | -127.4882 | -0.4920 | -0.4136 |
| 0.005 | 8.0 | 300 | 1.5640 | -2.6489 | -7.8179 | 0.7200 | 5.1690 | -130.9881 | -120.8790 | -0.2887 | -0.2011 |
| 0.0 | 10.67 | 400 | 1.6069 | -6.1709 | -11.6980 | 0.7200 | 5.5272 | -138.7483 | -127.9230 | -0.4158 | -0.3257 |
| 0.0 | 13.33 | 500 | 1.6195 | -6.2010 | -11.7066 | 0.7300 | 5.5057 | -138.7655 | -127.9832 | -0.4176 | -0.3276 |
| 0.0 | 16.0 | 600 | 1.6237 | -6.1966 | -11.7012 | 0.7200 | 5.5045 | -138.7545 | -127.9745 | -0.4185 | -0.3287 |
| 0.0 | 18.67 | 700 | 1.6053 | -6.1901 | -11.7285 | 0.7200 | 5.5384 | -138.8092 | -127.9615 | -0.4195 | -0.3294 |
| 0.0 | 21.33 | 800 | 1.6157 | -6.1882 | -11.7098 | 0.7200 | 5.5216 | -138.7718 | -127.9577 | -0.4195 | -0.3293 |
| 0.0 | 24.0 | 900 | 1.6297 | -6.2118 | -11.7270 | 0.7300 | 5.5153 | -138.8062 | -128.0048 | -0.4202 | -0.3298 |
| 0.0 | 26.67 | 1000 | 1.6259 | -6.2124 | -11.7257 | 0.7200 | 5.5132 | -138.8036 | -128.0061 | -0.4196 | -0.3295 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Library
- peft
- Created
- Apr 16, 2024
- Updated
- Apr 16, 2024