Llama-3.2-11B-Vision-Instruct

This model is a fine-tuned version of meta-llama/Llama-3.2-11B-Vision-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7115
• Bleu: 0.3191
• Rouge1: 0.6462
• Rouge2: 0.3482
• Rougel: 0.5529
• Bertscore Precision: 0.8764
• Bertscore Recall: 0.8935
• Bertscore F1: 0.8848

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
• lr_scheduler_warmup_steps: 50
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Rouge1	Rouge2	Rougel	Bertscore Precision	Bertscore Recall	Bertscore F1
1.7942	0.6202	50	1.7909	0.2890	0.6131	0.3240	0.5197	0.8199	0.8912	0.8535
1.7177	1.2403	100	1.7262	0.3165	0.6445	0.3454	0.5501	0.8724	0.8928	0.8825
1.7198	1.8605	150	1.7158	0.3184	0.6462	0.3475	0.5520	0.8753	0.8932	0.8841
1.6898	2.4806	200	1.7115	0.3191	0.6462	0.3482	0.5529	0.8764	0.8935	0.8848

Framework versions

• PEFT 0.13.0
• Transformers 4.45.2
• Pytorch 2.2.0a0+81ea7a4
• Datasets 3.0.1
• Tokenizers 0.20.1