metadata
library_name: transformers
license: apache-2.0
base_model: alignment-handbook/zephyr-7b-sft-full
tags:
- alignment-handbook
- trl
- dpo
- generated_from_trainer
- trl
- dpo
- generated_from_trainer
datasets:
- HuggingFaceH4/ultrafeedback_binarized
model-index:
- name: zephyr-7b-align-scan-0.0-0.3-polynomial-3
results: []
zephyr-7b-align-scan-0.0-0.3-polynomial-3
This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:
- Loss: 0.7657
- Rewards/chosen: -1.7538
- Rewards/rejected: -2.6151
- Rewards/accuracies: 0.3194
- Rewards/margins: 0.8612
- Logps/rejected: -90.8353
- Logps/chosen: -81.0013
- Logits/rejected: -2.4771
- Logits/chosen: -2.4954
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: 2.9336856595926955e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: polynomial
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.529 | 1.0417 | 100 | 0.6391 | 0.3319 | 0.0135 | 0.3353 | 0.3184 | -81.0782 | -73.2591 | -2.5045 | -2.5211 |
| 0.3646 | 2.0833 | 200 | 0.6854 | -0.2626 | -0.8523 | 0.3254 | 0.5897 | -84.2922 | -75.4660 | -2.5243 | -2.5414 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0
- Datasets 2.21.0
- Tokenizers 0.19.1