Unverified50% confidenceFactExact time
The asyncify build of onnxruntime-web does not include the CPU kernel for the GatherBlockQuantized operator, so it cannot load and use 4-bit Gemma ONNX models with block-quantized embeddings on pure WebAssembly (no WebGPU) devices.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedONNX Runtime Web is the browser version of Microsoft's cross-platform inference engine, capable of directly loading standard ONNX format models66% similarUnverifiedokasi/onnx-lab#2 documented the problem of asyncify lacking GatherBlockQuantized and switched to the JSEP build to work around it; dephelion/ufeed#64 also encountered a similar problem.65% similarUnverifiedIn Node.js, misusing fs.readFileSync or failing to offload CPU-intensive JSON parsing via worker_threads will block the event loop.65% similarUnverified借助 ONNX Runtime Web、Transformers.js 等框架，模型推理可直接在 Node.js 或浏览器环境中运行，无需依赖 Python 后端64% similarUnverified同一份WASM字节码可在Chrome、Firefox、Safari以及Node.js等运行时无缝执行，无需针对不同操作系统或CPU架构单独编译61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028538API
curl https://kongchang.com/api/v1/knowledge/claims/1028538MCP
get_claim(id=1028538)