Projects / TurboQuant KV Cache Optimization / Experiments / Verify 65K+ context on server with sufficient RAM

Verify 65K+ context on server with sufficient RAM

proposed high priority TODO-002
Overview Experiments 96 Forks 3 Resources 36 Benchmarks 2 Broadcasts 3 Related
Description

Perplexity logits buffer requires >37GB host RAM at 65K context. Need to confirm adaptive chunking + Q-batching hold PPL-match at this scale.

Reference

EXP-0002

Suggested Parameters
context 65536
chunks 8
cache_type tbq3
Provenance
Proposed by @dusterbloom via adaptive-chunked-prefill claude-sonnet-4-6