Frank 2e1a78aeb4 frank-default-small: 128K context and thinking on ...

- CTX_SIZE 65536 -> 131072 (user request). KV at 128K is 8 GiB
  (64 KiB/token f16); estimated ~43 GiB used of the 64 GiB SVM pool,
  ~21 GiB headroom.
- REASONING off -> on with REASONING_FORMAT=deepseek, so <think>
  reasoning lands in message.reasoning_content (OpenAI-compatible)
  instead of being left unparsed in content.

2026-08-17 23:11:13 -07:00

..

deploy.sh

dankclaw: replace vLLM with CIRU ROCmFPX llama.cpp runtime

2026-08-17 21:12:35 -07:00

frank-default-large.env

record performance fixes: FA_ALL_QUANTS build flag (dankserver), RDNA triton FA env (dankclaw)

2026-08-17 16:30:22 -07:00

frank-default-large.service

fleet redo 2026-08-17: retire PP+NPU, new topology

2026-08-17 15:23:46 -07:00

frank-default-small.env

frank-default-small: 128K context and thinking on

2026-08-17 23:11:13 -07:00

frank-default-small.service

dankclaw: replace vLLM with CIRU ROCmFPX llama.cpp runtime

2026-08-17 21:12:35 -07:00

serve-frank-large.sh

fix: make serve scripts + deploy.sh executable

2026-08-17 15:24:40 -07:00

serve-frank-small.sh

frank-default-small: 128K context and thinking on

2026-08-17 23:11:13 -07:00