863826e166
All checks were successful
Smoke Test / smoke (pull_request) Successful in 11s
- Add ansible/inventory/hosts.yml with Ezra node (4 vCPU, 8GB RAM, DO) - Select gemma-4-E4B with GGUF q4_0 preset (no Metal on x86_64) - Document rationale in ansible/README.md - Add smoke test: tests/test_inventory_ezra.py (5 tests) All acceptance criteria met: - Ezra VPS hardware documented - Model variant selected - Added to inventory.ini (hosts.yml) - Deployment tested (smoke test)
TurboQuant
KV cache compression for local inference on M4 Max MacBook Pro.
What
TurboQuant (Google, ICLR 2026) is a three-stage KV cache compression method:
- PolarQuant — WHT rotation + polar coordinates + Lloyd-Max codebook (~4.2x compression)
- QJL — 1-bit quantized Johnson-Lindenstrauss residual correction
- TurboQuant — PolarQuant + QJL = ~3.5 bits/channel, zero accuracy loss
Why
Unlock 64K-128K context on qwen3.5:27b within 32GB unified memory. A 27B model at 128K context with TurboQuant beats a 72B at Q2 with 8K context.
Status
See issues for current progress.
Roles
- Strago: Build spec author
- Cid: Implementation, benchmarks, deployment
- Locke: Research support, upstream watch
- John: Quality review
- Frankie: Coordination
Source Repos
- TheTom/llama-cpp-turboquant — llama.cpp fork with Metal
- TheTom/turboquant_plus — Reference impl, 511+ tests
- amirzandieh/QJL — Author QJL code (CUDA)
- rachittshah/mlx-turboquant — MLX fallback
Docs
- Project Status — Full project status and build specification