---
library_name: onnx
license: apache-2.0
pipeline_tag: token-classification
base_model: fastino/gliner2.5-multi-v1
tags:
  - onnx
  - gliner2
  - boundary
  - webgpu
  - token-classification
---

# gliner2.5-multi-v1-onnx

ONNX export of [fastino/gliner2.5-multi-v1](https://huggingface.co/fastino/gliner2.5-multi-v1) (GLiNER 2.5 `BoundaryExtractor`) for onnxruntime / WebGPU.

This is **not** a drop-in `AutoExtractor` graph. The ONNX file runs:

1. DeBERTa encoder on packed `input_ids`
2. Gather of word states and query-marker states
3. Boundary start/end logits `[batch, queries, words+1]`

Schema packing (entity-type markers) and span decode stay on the host, same split as GLiNER.js.

## Inputs

| Name | Shape | Dtype |
|------|-------|-------|
| input_ids | [B, T] | int64 |
| attention_mask | [B, T] | int64 |
| text_word_indices | [B, L] | int64 |
| text_word_mask | [B, L] | float32 |
| query_marker_indices | [B, Q] | int64 |
| query_marker_mask | [B, Q] | float32 |

## Outputs

| Name | Shape |
|------|-------|
| start_logits | [B, Q, L+1] |
| end_logits | [B, Q, L+1] |

## Python check

```python
import onnxruntime as ort
sess = ort.InferenceSession("onnx/model.onnx")
```

WebGPU: load `onnx/model.onnx` with `onnxruntime-web` `webgpu` execution provider. Int64 inputs are required; some browsers need the WASM backend as fallback.