medsiglip-448-ft-tb-screening

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.5291

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
3.5828	0.4762	25	2.4233
1.947	0.9524	50	2.6277
1.9315	1.4190	75	2.4902
1.9614	1.8952	100	2.5603
1.9357	2.3619	125	2.5067
1.9396	2.8381	150	2.6298
1.9313	3.3048	175	2.5459
1.8956	3.7810	200	2.5050
1.9271	4.2476	225	2.5318
1.9317	4.7238	250	2.5298
1.9365	5.1905	275	2.5108
1.9255	5.6667	300	2.5120
1.9284	6.1333	325	2.5134
1.907	6.6095	350	2.5199
1.8996	7.0762	375	2.5279
1.9321	7.5524	400	2.5274
1.9101	8.0190	425	2.5273
1.9131	8.4952	450	2.5293
1.91	8.9714	475	2.5300
1.919	9.4381	500	2.5290
1.9189	9.9143	525	2.5291

Framework versions

• Transformers 4.56.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.0