metadata
library_name: peft
tags:
- alignment-handbook
- trl
- dpo
- generated_from_trainer
base_model: NorLLM-AI/NorMistral-7B
datasets:
- hugodk-sch/aftonposten_title_prefs
model-index:
- name: norllm-ai-normistral-7b-align-scan
results: []
norllm-ai-normistral-7b-align-scan
This model is a fine-tuned version of data/norllm-ai-normistral-7b-sft-qlora on the hugodk-sch/aftonposten_title_prefs dataset. It achieves the following results on the evaluation set:
- Loss: 0.6292
- Rewards/chosen: 0.0553
- Rewards/rejected: 0.0226
- Rewards/accuracies: 0.5166
- Rewards/margins: 0.0327
- Logps/rejected: -34.6659
- Logps/chosen: -31.2035
- Logits/rejected: -2.8063
- Logits/chosen: -2.8088
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.1043 | 0.26 | 100 | 0.6260 | 0.0189 | -0.0071 | 0.5370 | 0.0260 | -34.7083 | -31.2554 | -2.8132 | -2.8159 |
| 3.0672 | 0.52 | 200 | 0.6356 | 0.0453 | 0.0067 | 0.5627 | 0.0386 | -34.6886 | -31.2177 | -2.8107 | -2.8128 |
| 1.2353 | 0.78 | 300 | 0.6302 | 0.0524 | 0.0270 | 0.5311 | 0.0254 | -34.6596 | -31.2076 | -2.8071 | -2.8097 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.15.1