serversdown abac42c344 fix: serial consolidation on GPU backends (was timing out the MI50) ...

summarize_all fanned out 8 concurrent workers, but the MI50 llama.cpp server runs
a single slot (--parallel 1). Firing 8 at once queued them, blew the client timeout
('summary retry … Request timed out'), and thrashed/cancelled the KV cache — wasted
compute and heat. Concurrency is now backend-aware: 8 for cloud, 1 for local/MI50.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

2026-06-27 09:26:25 +00:00

..

personas

feat: live chat deliberation — think privately before answering (less 'meh')

2026-06-23 00:35:49 +00:00

web

feat: session timeline (running log) + reliable live logging

2026-06-27 07:26:14 +00:00

__init__.py

chore: project scaffold (uv, .env.example, README, lyra package)

2026-05-16 06:01:08 +00:00

__main__.py

feat: persona chat loop, web UI, and local (Ollama) embeddings

2026-06-15 18:36:31 +00:00

backfill.py

fix: backfill skips hand extraction by default (prose->replay too lossy)

2026-06-18 06:04:02 +00:00

chat.py

feat(P3): mind/mouth split — separate voice model for the final reply (seam, default off)

2026-06-24 06:08:06 +00:00

clock.py

feat: session timeline (running log) + reliable live logging

2026-06-27 07:26:14 +00:00

cognition.py

feat: associative cognition — thoughts arise from spreading activation, not a re-read bio

2026-06-22 05:45:39 +00:00

config.py

feat(P3): mind/mouth split — separate voice model for the final reply (seam, default off)

2026-06-24 06:08:06 +00:00

dream.py

feat: proactive thoughts — auto-ping salient ones + daily digest

2026-06-22 20:25:14 +00:00

equity.py

feat: deterministic equity/board-reading tool (math via tools, not LLM)

2026-06-18 18:45:40 +00:00

era.py

perf: incremental era rebuilds — skip unchanged months

2026-06-25 03:31:02 +00:00

feeds.py

feat: thought loop reach-out (ntfy push) + external input feeds

2026-06-22 00:21:06 +00:00

ingest.py

feat: import raw ChatGPT export (new sharded format)

2026-06-16 02:40:32 +00:00

llm.py

feat: log every LLM call at the router boundary (backend/model/tokens/ms)

2026-06-27 09:26:25 +00:00

logbus.py

feat: run dream cycle as a systemd user service + journald-visible logs

2026-06-17 01:42:55 +00:00

memory.py

feat: web switch for her inner voice (Dolphin/3090 | Qwen-32B/MI50 | Off)

2026-06-22 19:16:35 +00:00

mind.py

perf: tighten the dynamic prompt — persona split + lean deliberation

2026-06-24 20:48:44 +00:00

modes.py

feat: session timeline (running log) + reliable live logging

2026-06-27 07:26:14 +00:00

narrative.py

feat: era-rollup + narrative engine (consolidation steps 3-4)

2026-06-16 19:28:01 +00:00

notify.py

fix: ntfy ping is her personal text to Brian, by her decision — not a thought dump

2026-06-22 01:39:11 +00:00

perceive.py

feat(P2): perceive (read the moment) + route nudges register on charged turns

2026-06-24 05:42:36 +00:00

persona.py

perf: tighten the dynamic prompt — persona split + lean deliberation

2026-06-24 20:48:44 +00:00

poker.py

feat: session timeline (running log) + reliable live logging

2026-06-27 07:26:14 +00:00

profile.py

perf: incremental profile rebuilds — fold new gists instead of re-digesting all

2026-06-25 04:05:24 +00:00

self_state.py

feat: web switch for her inner voice (Dolphin/3090 | Qwen-32B/MI50 | Off)

2026-06-22 19:16:35 +00:00

session.py

feat: persona chat loop, web UI, and local (Ollama) embeddings

2026-06-15 18:36:31 +00:00

summary.py

fix: serial consolidation on GPU backends (was timing out the MI50)

2026-06-27 09:26:25 +00:00

thoughts.py

feat: chat-side feedback — reactions in conversation thread back to her thoughts

2026-06-22 23:26:40 +00:00

tools.py

feat: session timeline (running log) + reliable live logging

2026-06-27 07:26:14 +00:00