Public repo Last commit 3 days ago

quillan.cpp

quillan.cpp is a high-performance, lightweight, standalone C++17 inference engine engineered for the Quillan-Ronin sovereign multi-tier cognitive architecture. Inspired by the minimalistic philosophies of llama.cpp and bitnet.cpp, quillan.cpp operates with zero heavyweight runtime dependencies (no PyTorch, no LibTorch, no Python runtime), running

quillan.cpp's week on Fridays, when it has three or more things in it. Double opt-in, one-click unsubscribe. Or follow everything CrashOverrideX builds

Commits · last 30 days

2 ▼ 67% vs the previous 30 days
1/2
1
Oct 3 Oct 18 Today

A 7-day average of commits per day, one line per project.

Recent commits

From the GitHub sync, as written
Oct 29
1 commit
feat(kernel): add native Pascal sm_61 DP4A CUDA kernel and hardware ablation benchmarks (100 tok/s)
Oct 21
1 commit
feat: add Quillan.cpp native AVX2 engine files
Oct 2
1 commit
docs: add trending-local-llms verified inference engine badge and citation
Sep 12
5 commits
feat(api): export quillan_shared.dll and provide in-process ctypes Python bridge
feat(quillan.cpp): load 50k token vocabulary table from quillan_vocab.txt in CLI
feat(quillan.cpp): wire real embedding lookup, LM head projection, and vocab detokenization table
and 2 more

Merges and dependency updates stay hidden. All commits on GitHub

What shipped

0 in the last 30 days

No entries tagged quillan.cpp in the last 30 days.