metadata
license: apache-2.0
base_model: alignment-handbook/zephyr-7b-sft-full
tags:
- generated_from_trainer
model-index:
- name: spin-margin2
results: []
spin-margin2
This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0010
- Rewards/real: -0.7975
- Rewards/generated: -20.4822
- Rewards/accuracies: 1.0
- Rewards/margins: 19.6846
- Logps/generated: -303.8466
- Logps/real: -141.0674
- Logits/generated: -2.6068
- Logits/real: -2.3492
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: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- total_train_batch_size: 32
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/real | Rewards/generated | Rewards/accuracies | Rewards/margins | Logps/generated | Logps/real | Logits/generated | Logits/real |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0043 | 0.19 | 100 | 0.0049 | 0.9120 | -9.6012 | 1.0 | 10.5132 | -195.0367 | -123.9721 | -2.7982 | -2.5652 |
| 0.0034 | 0.39 | 200 | 0.0024 | -0.0739 | -14.1834 | 1.0 | 14.1095 | -240.8593 | -133.8314 | -2.8109 | -2.5347 |
| 0.0007 | 0.58 | 300 | 0.0012 | -0.2381 | -16.9127 | 1.0 | 16.6746 | -268.1524 | -135.4731 | -2.7308 | -2.4046 |
| 0.0016 | 0.78 | 400 | 0.0010 | -1.1878 | -19.5719 | 1.0 | 18.3841 | -294.7439 | -144.9703 | -2.6559 | -2.3917 |
| 0.0001 | 0.97 | 500 | 0.0010 | -0.7975 | -20.4822 | 1.0 | 19.6846 | -303.8466 | -141.0674 | -2.6068 | -2.3492 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.15.2