Instructions to use Ozantsk/biomedclip-rocov2-finetuned with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- OpenCLIP
How to use Ozantsk/biomedclip-rocov2-finetuned with OpenCLIP:
import open_clip model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:Ozantsk/biomedclip-rocov2-finetuned') tokenizer = open_clip.get_tokenizer('hf-hub:Ozantsk/biomedclip-rocov2-finetuned') - Notebooks
- Google Colab
- Kaggle
metadata
library_name: open_clip
base_model: microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224
datasets:
- eltorio/ROCOv2-radiology
tags:
- biomedclip
- rocov2
- medical-image-retrieval
BiomedCLIP fine-tuned on ROCOv2
Fine-tuned from microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224 on eltorio/ROCOv2-radiology with CLIP contrastive image-caption loss.
Main hyperparameters:
- LR: 1e-05
- Epochs: 4
- Batch size: 16
- Gradient accumulation: 4
- Effective batch size: 64
- Context length: 256
Saved files:
open_clip_pytorch_model.binopen_clip_config.jsontraining_config.json