cefaa6e778
TurboQuant KV cache compression for M4 Max local inference. Spec by Strago, triaged into 16 issues across 4 phases. Ref #1
33 lines
1.3 KiB
Markdown
33 lines
1.3 KiB
Markdown
# 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:
|
|
1. **PolarQuant** — WHT rotation + polar coordinates + Lloyd-Max codebook (~4.2x compression)
|
|
2. **QJL** — 1-bit quantized Johnson-Lindenstrauss residual correction
|
|
3. **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](http://143.198.27.163:3000/Timmy_Foundation/turboquant/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](https://github.com/TheTom/llama-cpp-turboquant) — llama.cpp fork with Metal
|
|
- [TheTom/turboquant_plus](https://github.com/TheTom/turboquant_plus) — Reference impl, 511+ tests
|
|
- [amirzandieh/QJL](https://github.com/amirzandieh/QJL) — Author QJL code (CUDA)
|
|
- [rachittshah/mlx-turboquant](https://github.com/rachittshah/mlx-turboquant) — MLX fallback
|
|
|
|
## Docs
|
|
- [BUILD-SPEC.md](BUILD-SPEC.md) — Full build specification (Strago, v2.2)
|