3 people ec29f21fbf Add Sjoerd L40S validation, KLD context-length stability data ...

- Asymmetric paper: Sjoerd 122B on 2x L40S at 82K. q8_0/turbo3 100%
  decode recovery. q8_0/turbo2 also 100% recovery (V free to 2-bit).
- MoE paper: turbo2 no-crossover at 82K on hybrid MoE (12 attn layers),
  architecture-dependent finding.
- KLD results: new context-length scaling data (2K/4K/8K on M2 Pro).
  Mean KLD stable/improving with context, does not drift.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: tturney@psyguard.ai

2026-04-01 15:09:08 -05:00

..

asymmetric-kv-compression.md

Add Sjoerd L40S validation, KLD context-length stability data

2026-04-01 15:09:08 -05:00

block-size-experiment.md

Fix addendum accuracy: dates, attributions, numbers

2026-03-31 11:49:10 -05:00

layer-aware-v-compression.md

Update docs with community validation and benchmarking guide improvements

2026-04-01 11:25:15 -05:00

m5-max-stress-test.md

Fix addendum accuracy: dates, attributions, numbers

2026-03-31 11:49:10 -05:00

moe-v-compression-frontier.md

Add Sjoerd L40S validation, KLD context-length stability data

2026-04-01 15:09:08 -05:00

sparse-v-dequant.md

Update docs with community validation and benchmarking guide improvements

2026-04-01 11:25:15 -05:00

turbo4-resurrection.md

Fix addendum accuracy: dates, attributions, numbers

2026-03-31 11:49:10 -05:00