state/agent/logs/engine.latest.log

[local-llama] staged: reusing local /tmp/local-agent-models/Qwen3-1.7B-Q4_K_M.gguf
[local-llama] loading: llama.cpp via LLama-GUI · threads=2 · ctx=32768 · batch=1024/256 · KV=q4_0/q4_0 · reasoning=low/64
[local-llama] staged KV snapshot 0.8 MiB bucket -> local in 0.40s
[local-llama] LLama-GUI launch: --model /tmp/local-agent-models/Qwen3-1.7B-Q4_K_M.gguf --host 127.0.0.1 --port 43126 --ctx-size 32768 --threads 2 --threads-batch 2 --parallel 1 --cont-batching --flash-attn auto --jinja --reasoning on --reasoning-effort low --reasoning-budget 64 --batch-size 1024 --ubatch-size 256 --threads-http 1 --cache-type-k q4_0 --cache-type-v q4_0 --cache-prompt --cache-reuse 256 --metrics --slots --slot-save-path /tmp/local-agent-kv/7f6990e4639016ec --gpu-layers 0 --no-warmup
[local-llama] restored KV slot tokens=26 bytes=839788 in 0.03s
[local-llama] ready
[local-llama] saved local KV slot tokens=26 bytes=839788 in 0.03s reason=manual
[local-llama] persisted KV snapshot tokens=26 size=0.8 MiB to bucket in 0.05s reason=manual
[local-llama] saved local KV slot tokens=26 bytes=839788 in 0.03s reason=shutdown
[local-llama] persisted KV snapshot tokens=26 size=0.8 MiB to bucket in 0.00s reason=shutdown