metadata
library_name: peft
tags:
- trl
- dpo
- alignment-handbook
- 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:
- Logits/chosen: -2.2437
- Logits/rejected: -2.2388
- Logps/chosen: -34.0172
- Logps/rejected: -37.4976
- Loss: 0.6982
- Rewards/accuracies: 0.4983
- Rewards/chosen: 0.0156
- Rewards/margins: -0.0015
- Rewards/rejected: 0.0171
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 | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.6951 | 0.26 | 100 | -2.2438 | -2.2389 | -34.0252 | -37.5076 | 0.6966 | 0.4954 | 0.0084 | 0.0003 | 0.0081 |
| 0.6891 | 0.52 | 200 | -2.2432 | -2.2384 | -34.0243 | -37.5115 | 0.6947 | 0.4934 | 0.0092 | 0.0046 | 0.0046 |
| 0.693 | 0.78 | 300 | -2.2437 | -2.2388 | -34.0172 | -37.4976 | 0.6982 | 0.4983 | 0.0156 | -0.0015 | 0.0171 |
Framework versions
- PEFT 0.8.2
- Transformers 4.37.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1