HebCaps syntax (ONNX, int8)

Dynamic int8 ONNX export of dicta-il/dictabert-syntax by Dicta, used by HebCaps to keep Hebrew phrases (noun + adjective, smichut, numerals, addresses) inside one caption.

Files: model.onnx (int8; inputs input_ids, attention_mask; outputs arcs [B,T,T], own [B,T,L], dep [B,T,L] — relation of token i = argmax(own[i] + dep[head(i)])), tokenizer.json, syntax.json (relation labels). Exported with HebCaps' own reimplementation of the parsing head (no remote code).

License: CC BY 4.0, same as the original model. Credit: Dicta, "DictaBERT: A State-of-the-Art BERT Suite for Modern Hebrew" (Shmidman et al., 2023).

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for yoelkn/hebcaps-syntax-onnx

Base model

dicta-il/dictabert-syntax

Quantized

(1)

this model