#low-bit inference
Low-Bit Inference: 1 AI articles covering low-bit inference news, analysis, and research
Articles
Running Underdog Saluki 27B: A 2-Bit Qwen3.8-27B Tool-Calling GuideNEW⭐9
A hands-on guide to Underdog Saluki 27B, a 7.9 GB 2-bit Qwen3-27B build. We diff its GGUF, serve it in llama.cpp, and score tool calling on 120 BFCL tasks.