metadata
library_name: peft
tags:
- trl
- dpo
- generated_from_trainer
base_model: NbAiLab/nb-gpt-j-6B-v2
model-index:
- name: aftonposten-6b-align-scan
results: []
aftonposten-6b-align-scan
This model is a fine-tuned version of NbAiLab/nb-gpt-j-6B-v2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.6903
- Rewards/chosen: -0.0237
- Rewards/rejected: -0.0486
- Rewards/accuracies: 0.5191
- Rewards/margins: 0.0249
- Logps/rejected: -37.5860
- Logps/chosen: -34.0684
- Logits/rejected: -2.2229
- Logits/chosen: -2.2277
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-06
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.6577 | 0.26 | 100 | 0.6949 | -0.0038 | -0.0133 | 0.5316 | 0.0095 | -37.5357 | -34.0400 | -2.2289 | -2.2338 |
| 0.6156 | 0.52 | 200 | 0.6943 | -0.0132 | -0.0288 | 0.5191 | 0.0156 | -37.5578 | -34.0535 | -2.2268 | -2.2317 |
| 0.5468 | 0.78 | 300 | 0.6903 | -0.0237 | -0.0486 | 0.5191 | 0.0249 | -37.5860 | -34.0684 | -2.2229 | -2.2277 |
Framework versions
- PEFT 0.8.2
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1