idefics-9b-YBT-Scores

This model is a fine-tuned version of HuggingFaceM4/idefics-9b on a dataset of ybalancetest. It achieves the following results on the evaluation set:

• Loss: 2.9246

Model description

VLM to assess y balance test

Intended uses & limitations

This is trained only for visual QA on y balance test.

Training and evaluation data

It is based on a dataset ybalancetest

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.1393	2.76	10	2.0003
1.4136	5.52	20	1.5828
1.1321	8.28	30	1.5916
0.8633	11.03	40	1.6502
0.6091	13.79	50	1.8128
0.406	16.55	60	2.0350
0.2218	19.31	70	2.3489
0.1255	22.07	80	2.6919
0.0711	24.83	90	2.8418
0.0606	27.59	100	2.9246

Framework versions

• PEFT 0.9.1.dev0
• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2