AA_preference_random_0_30
This model is a fine-tuned version of llava-hf/llava-v1.6-mistral-7b-hf on the AA_preference_random_0_30 dataset. It achieves the following results on the evaluation set:
- Loss: 0.6184
- Rewards/chosen: 0.7817
- Rewards/rejected: -0.8532
- Rewards/accuracies: 0.7708
- Rewards/margins: 1.6349
- Logps/rejected: -211.9707
- Logps/chosen: -253.8823
- Logits/rejected: -2.5581
- Logits/chosen: -2.5412
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: 1e-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.2364 | 1.2422 | 50 | 0.5729 | 1.2346 | 0.1347 | 0.7292 | 1.0999 | -202.0917 | -249.3526 | -2.5158 | -2.5087 |
| 0.1061 | 2.4845 | 100 | 0.6159 | 0.7571 | -0.8876 | 0.7569 | 1.6447 | -212.3150 | -254.1281 | -2.5775 | -2.5599 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.4.0+cu121
- Datasets 2.21.0
- Tokenizers 0.20.3
- Downloads last month
- 8
Inference API (serverless) does not yet support transformers models for this pipeline type.
Model tree for htlou/mm-interp-AA_preference_random_0_30
Base model
llava-hf/llava-v1.6-mistral-7b-hf