medsiglip-448-cied-650-binary-classification

This model is a fine-tuned version of google/medsiglip-448 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 3.7149

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: 0.001
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 128
  • 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: 5
  • num_epochs: 15

Training results

Training Loss Epoch Step Validation Loss
7.3163 1.0 5 8.6896
5.8702 2.0 10 3.7810
4.3275 3.0 15 3.9249
3.7694 4.0 20 3.7549
3.7601 5.0 25 3.7149
3.7215 6.0 30 3.7158
3.7119 7.0 35 3.7149
3.7088 8.0 40 3.7150
3.7067 9.0 45 3.7159
3.703 10.0 50 3.7149
3.7032 11.0 55 3.7150
3.7017 12.0 60 3.7149
3.7019 13.0 65 3.7149
3.7014 14.0 70 3.7149
3.7012 15.0 75 3.7149

Framework versions

  • Transformers 4.57.3
  • Pytorch 2.9.0+cu126
  • Datasets 4.4.2
  • Tokenizers 0.22.1
Downloads last month
123
Safetensors
Model size
0.9B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for aomocelin/medsiglip-448-cied-650-binary-classification

Finetuned
(33)
this model