gliner2.5-small-v1-onnx

ONNX export of fastino/gliner2.5-small-v1 (GLiNER 2.5 BoundaryExtractor) for onnxruntime-web / WebGPU. Revision 5 graphs plus attrs.onnx (score_explicit_spans).

One encoder pass. JointIE beam, classification implies/excludes, record assignment, and long-document chunking stay in JavaScript.

Host: Pastel-Org/gliner2.5-onnx-webgpu (live: gliner25-onnx-webgpu.pages.dev).

Files

attrs.onnx

Dynamo export of head.score_explicit_spans. Traced at 512 words, 8 attribute queries, 16 spans. The JS host crops a 512-word window around the mention; that pad is not a document-length cap. ORT vs torch RMSE ~1e-6 (small) / ~2e-6 (base, multi). Overlay lookup is the fallback if this graph fails to load.

Decode notes (JS host, not this graph)

Attribute labels are packed with entity labels in one [E] block (sorted). Single-label attributes use softmax. ONNX pair_valid is all-true; the host drops start >= end slots before overlap.

Python max_len is 4096 words: classify_text / extract is one forward pass over the whole document. That is what Fastino means by long-context classification on CUDA/CPU.

This WebGPU export is not that:

A 4096-word classify_text_long never builds a 4096-word attention matrix. Use the *_long APIs for contracts on base/multi in the browser.

Still not in ONNX

Full Kleene classification AST is not ported; README implies / excludes is a JS beam. Latent / anchorless records are not exported.

Credits

• Base checkpoints: Fastino, Apache-2.0.
• ONNX export + JS host: Pastel-Cloud OÜ.