Unverified50% confidenceFactExact time
The quantization scheme of Underdog Saluki 27B uses ISTA's GSQ-RCO quantization method.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/25/2026
First Seen
Valid until: 1/23/2027
Sources
Related Entities
Related Claims
UnverifiedUnderdog Saluki 27B is a model obtained by 2-bit quantization of Qwen3.8-27B.75% similarUnverifiedIn 120 function-calling tests on Underdog Bench 120—built on Berkeley's BFCL V4 leaderboard—Saluki 27B scored 88, surpassing the full-size original model's score of 84.64% similarUnverifiedSaluki 27B surpasses the full-precision original model in tool-calling capability63% similarUnverifiedGPTQ (2022) achieved the first usable performance of large models at INT4 by quantizing layer by layer and minimizing quantization error using Hessian information.60% similarUnverifiedllama.cpp通过GGUF格式承载量化后的模型权重，常见量化等级包括Q4_K_M、Q5_K_M、Q8_0等，级别越高精度损失越小、显存占用越大。57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1095140API
curl https://kongchang.com/api/v1/knowledge/claims/1095140MCP
get_claim(id=1095140)