llava_test

This model is a fine-tuned version of llava-hf/llava-1.5-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0446
• Bleu: 0.6353
• Rouge1: 0.7885
• Rouge2: 0.7889
• Rougel: 0.7893
• Bertscore Precision: 0.6807
• Bertscore Recall: 0.7674
• Bertscore F1: 0.7213

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-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100.0

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Rouge1	Rouge2	Rougel	Bertscore Precision	Bertscore Recall	Bertscore F1
0.3168	10.0	10	2.2001	0.0724	0.3123	0.1239	0.2433	0.7068	0.7777	0.7405
0.2454	20.0	20	1.6882	0.1061	0.4044	0.1840	0.3274	0.7241	0.7794	0.7507
0.1821	30.0	30	1.1567	0.1925	0.5281	0.2989	0.4593	0.7054	0.7756	0.7387
0.109	40.0	40	0.5242	0.3915	0.6689	0.5316	0.6370	0.6878	0.7709	0.7268
0.0378	50.0	50	0.1193	0.5971	0.7701	0.7585	0.7700	0.6839	0.7688	0.7237
0.0098	60.0	60	0.0554	0.6254	0.7862	0.7867	0.7875	0.6799	0.7694	0.7217
0.0064	70.0	70	0.0482	0.6329	0.7889	0.7890	0.7899	0.6798	0.7690	0.7215
0.0059	80.0	80	0.0459	0.6331	0.7877	0.7877	0.7888	0.6777	0.7670	0.7194
0.0057	90.0	90	0.0451	0.6347	0.7897	0.7895	0.7907	0.6807	0.7675	0.7213
0.0056	100.0	100	0.0446	0.6353	0.7885	0.7889	0.7893	0.6807	0.7674	0.7213

Framework versions

• PEFT 0.13.0
• Transformers 4.45.2
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.20.1