Skip to content

Instantly share code, notes, and snippets.

View sdurnov's full-sized avatar

Sergey Durnov sdurnov

1 follower · 1 following
View GitHub Profile
@sdurnov
sdurnov / config.md
Last active September 29, 2026 08:15
Qwen 3.8 Flash Next config LLama.cpp

Qwen3.8-Flash-Next (125B MoE + 51B n-gram table) on 16GB GPU + 64GB RAM — llama.cpp recipe

Validated on: RTX 5060 Ti 16GB · i5-14500 · 64GB DDR4-3200 · NVMe Gen4 · llama.cpp 0.5.0-dev (needs qwen4exp arch + ngram spec support, any build after ~Sep 27 2026). All numbers from a two-round A/B-A/B bench, 2026-09-28.

Files

huggingface.co/AtomicChat/Qwen3.8-Flash-Next-GGUF
  → Qwen3.8-Flash-Next-AD-4.27bpw-Q4_K_M-M64/   # 94.5 GB, 33 shards; shard 2 = 38 GB n-gram table (keep on NVMe!)
  → mmproj-Qwen3.8-Flash-Next-F16.gguf           # 0.9 GB vision encoder
@sdurnov
sdurnov / cloudSettings
Created May 17, 2018 10:26
Visual Studio Code Settings Sync Gist
{"lastUpload":"2018-05-17T10:26:53.449Z","extensionVersion":"v2.9.2"}