Issues

• gguf

  IST-DASLab/GSQ#12 · Filter by author ayttop opened on Sep 9, 2026

• -mtp GGUFs: IQ3_S-mtp MTP head gets 0.0% draft acceptance (XXS-mtp works)

  IST-DASLab/GSQ#11 · Filter by author sunyan999999 opened on Sep 9, 2026

• OpenThoughts preprocessing hardcodes a Kimi-specific system prompt

  IST-DASLab/GSQ#10 · Filter by author anzr299 opened on Sep 9, 2026

• Reproduction tooling for Qwen3.8-27B GSQ-RCO mixed-precision GGUF release

  IST-DASLab/GSQ#9 · Filter by author SoppyLeon opened on Sep 7, 2026
  1

• SM80 (Ampere/A100/170HX) GSQ INT3 decode speed bottlenecked by upstream Humming kernel, not GSQ quality

  IST-DASLab/GSQ#8 · Filter by author yangfangyunyun opened on Sep 6, 2026

• NPROC=2 OOMs on 2x12GB consumer GPUs where NPROC=1 succeeds (Qwen3.6-27B dense, gsq_bits=2); expandable_segments:True also masks OOM as "device not ready" on WSL2

  IST-DASLab/GSQ#6 · Filter by author Nakanokensetsu opened on Jul 3, 2026
  1

  

• How to refine a public GGUF K-Quant model?

  IST-DASLab/GSQ#5 · Filter by author yancaoweidaode opened on Jul 1, 2026