meridian-online/finetype-model
Add a card that says what the weights are and links the project home
Store the value encoder in half precision (F32 -> F16, 30.2MB -> 15.1MB)
Add m2v8m-s43/value_model2vec/tokenizer.json (potion-8M dual-encoder, retires v19)
Add m2v8m-s43/value_model2vec/model.safetensors (potion-8M dual-encoder, retires v19)
Add m2v8m-s43/label_map.json (potion-8M dual-encoder, retires v19)
Add m2v8m-s43/config.json (potion-8M dual-encoder, retires v19)
Add m2v8m-s43/model.safetensors (potion-8M dual-encoder, retires v19)
Upload folder using huggingface_hub
Upload folder using huggingface_hub
Add sherlock-v16 label_map.json
Add sherlock-v16 config.json
Add sherlock-v16 model.safetensors (seed 43, 235/242)
Archive sherlock-v14 model files in versioned directory
Update label_map.json and config.json for sherlock-v14
Publish sherlock-v14: 218/227 (96.0%), best val_acc 91.2% at epoch 42
Add sherlock-v13 label_map (240 types)
Add sherlock-v13 config (240 types, valid_hidden=[192,128])
Add sherlock-v13: 212/227 (93.4% label, 93.8% domain)
Add sherlock-v11/label_map.json (201/227 eval, 88.5% label accuracy)
Add sherlock-v11/config.json (201/227 eval, 88.5% label accuracy)
Add sherlock-v11/model.safetensors (201/227 eval, 88.5% label accuracy)
Upload sherlock-v7/eval/profile_results.json with huggingface_hub
Upload sherlock-v7/eval/profile_results.csv with huggingface_hub
Upload sherlock-v7/eval/actionability_results.csv with huggingface_hub
Upload sherlock-v7/eval/report.md with huggingface_hub
Upload sherlock-v7/eval/ground_truth.csv with huggingface_hub
Upload sherlock-v7/label_map.json with huggingface_hub
Upload sherlock-v7/results.json with huggingface_hub
Upload sherlock-v7/config.json with huggingface_hub
Upload sherlock-v7/model.safetensors with huggingface_hub
Add entity classifier model
Add Model2Vec semantic hint classifier
Add sibling-context attention model
Add sherlock-v4-sibling multi-branch model
initial commit