Commit Graph

4 Commits

Author SHA1 Message Date
Zijie Tian
6783a45e6f 🚧 wip: update sparse policy refactoring plan to v4
Add clear acceptance criteria and verification methods:
- Define 3 acceptance criteria (needle test, zero calc in attention.py, KV via offload_engine)
- Document violations to fix (direct flash_attn/copy calls)
- Add offload_engine.write_prefill_buffer encapsulation plan
- Add LSP-based verification method using cclsp tools

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-19 23:23:16 +08:00
Zijie Tian
16b269d897 🚧 wip: update sparse policy refactoring plan to v4
Simplified scope to FullPolicy only. Added debug validation phase.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-19 23:10:49 +08:00
Zijie Tian
b97b0b96a0 [WIP] Before refactor the nanovllm sparse policy. 2026-01-19 22:34:44 +08:00
Zijie Tian
03a8c033cb [claudesquad] update from 'add-llama-1' on 10 Jan 26 21:03 CST 2026-01-10 21:03:45 +08:00