Downloads · 30 days
7
18% of all-time downloads
model_hh_usp4_dpo5
guoyu-zhang/model_hh_usp4_dpo5
model_hh_usp4_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
38
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
- 77ef86e0
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_usp4_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: 1.9280
- Rewards/chosen: -10.9160
- Rewards/rejected: -14.7692
- Rewards/accuracies: 0.6700
- Rewards/margins: 3.8532
- Logps/rejected: -143.5620
- Logps/chosen: -133.8559
- Logits/rejected: -0.6034
- Logits/chosen: -0.5139
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0836 | 2.67 | 100 | 1.1505 | 2.1129 | 0.6066 | 0.5900 | 1.5063 | -112.8104 | -107.7982 | -0.2532 | -0.2394 |
| 0.0139 | 5.33 | 200 | 1.7113 | -12.2276 | -13.7761 | 0.6700 | 1.5485 | -141.5758 | -136.4791 | 0.1125 | 0.1515 |
| 0.0004 | 8.0 | 300 | 1.8255 | -8.3459 | -12.1000 | 0.6700 | 3.7540 | -138.2236 | -128.7158 | -0.4577 | -0.3723 |
| 0.0 | 10.67 | 400 | 1.9235 | -10.8751 | -14.7254 | 0.6700 | 3.8504 | -143.4745 | -133.7741 | -0.6028 | -0.5129 |
| 0.0 | 13.33 | 500 | 1.9333 | -10.9027 | -14.7363 | 0.6700 | 3.8337 | -143.4963 | -133.8293 | -0.6037 | -0.5138 |
| 0.0 | 16.0 | 600 | 1.9330 | -10.9168 | -14.7504 | 0.6700 | 3.8336 | -143.5245 | -133.8575 | -0.6033 | -0.5134 |
| 0.0 | 18.67 | 700 | 1.9274 | -10.9064 | -14.7738 | 0.6700 | 3.8674 | -143.5712 | -133.8367 | -0.6036 | -0.5138 |
| 0.0 | 21.33 | 800 | 1.9164 | -10.9296 | -14.8003 | 0.6700 | 3.8707 | -143.6242 | -133.8832 | -0.6049 | -0.5144 |
| 0.0 | 24.0 | 900 | 1.9275 | -10.9403 | -14.7883 | 0.6700 | 3.8480 | -143.6003 | -133.9046 | -0.6037 | -0.5139 |
| 0.0 | 26.67 | 1000 | 1.9280 | -10.9160 | -14.7692 | 0.6700 | 3.8532 | -143.5620 | -133.8559 | -0.6034 | -0.5139 |
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