Unverified50% confidenceFactExact time
WebGPU上加载q4f16模型文件（SmolLM2为272.7MB），WASM上加载q4模型文件（SmolLM2为387.9MB）。
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSome ONNX exports labeled q4 are larger than the q4f16 versions because the embedding table retains 32-bit precision; for example, SmolLM2-360M's q4 is 390MB and q4f16 is 275MB, and the two produce identical outputs.69% similarUnverifiedCandidate small models include SmolLM2 (135M and 360M), Gemma 3 270M, LFM2.5-350M, Qwen2.5-0.5B, and Qwen3-0.6B, all of which have ONNX browser builds.68% similarUnverifiedThe all-MiniLM-L6-v2 embedding model has about 22 million parameters, with the quantized CPU version at only 23MB and the WebGPU version at 90MB.65% similarVerifiedThe Qwen2.5:7B model is approximately 4 GB in size after Q4 quantization.65% similarUnverifiedThere is currently no NVFP4 build of AREX-2; available quantized versions include BAAI's original BF16 (54.7 GB), prithivMLmods' FP8 (35.9 GB), cyankiwi's AWQ INT4 (20.2 GB), and numsu's INT4 W4A16 (18.6 GB).63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1051524API
curl https://kongchang.com/api/v1/knowledge/claims/1051524MCP
get_claim(id=1051524)