Downloads · 30 days
0
NeuraStudios/clipit-qwen3guard-stream-0.6b-onnx-int8
clipit-qwen3guard-stream-0.6b-onnx-int8 is a feature extraction model from NeuraStudios. Use it when you need embeddings to search or compare text. It is set up for onnxruntime. The card lists the license as apache-2.0.
This repository contains a signed INT8 ONNX derivative of Qwen/Qwen3Guard-Stream-0.6B for moderation infrastructure.
Downloads · 30 days
0
Access
Public
Updated Jul 15, 2026
Repo size
612 MB
Likes
0
Public
Click a slice to open those files.
.data597 MB · 97%
From the Hugging Face model README
This repository contains a signed INT8 ONNX derivative of
Qwen/Qwen3Guard-Stream-0.6B
for moderation infrastructure.
Qwen/Qwen3Guard-Stream-0.6B419364a715de9840d47b1457982f64ff37f90ed4input_ids and attention_mask.risk_logits with three classes and
category_logits with nine classes.vmap
implementation could not be traced by the ONNX exporter.The exported logits were checked against the original PyTorch model before quantization. Across the local qualification corpus, signed INT8 produced no Safe versus non-Safe boundary changes. Some Unsafe versus Controversial tier changes were observed.
model.onnx: ONNX graph.model.onnx.data: external signed INT8 model weights.tokenizer.json: tokenizer from the pinned upstream revision.LICENSE: upstream Apache License 2.0 text.83b01b5df2d570089918da04592d9e32022d357e8d18044d9f46d8cdbb1aa47a model.onnx
d105fd59923cb917194b1efcb7460b771aefd5c2eb5015f4b9948b11d07e82de model.onnx.data
aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4 tokenizer.json
Tests were run on Apple Silicon with ONNX Runtime using 100 multilingual and mixed-length inputs between 20 and 300 characters:
Average: 42.86 ms
P50: 44.39 ms
P95: 64.13 ms
P99: 69.33 ms
Maximum: 70.28 ms
The production runtime uses the final token from the query risk and query category outputs. These measurements are qualification results, not guarantees for other operating systems or hardware.
The upstream model and this derivative are distributed under the Apache
License 2.0. See LICENSE. Qwen and Alibaba do not endorse this derivative.