Commit Graph
3 Commits
Author SHA1 Message Date
Frank 52ad985978 feat: raise Ling-3 GPU memory utilization to 0.95
The 0.72 default from native-256k.env OOMs. Export 0.95 in the wrapper;
native-256k uses ${GPU_MEMORY_UTILIZATION:-0.72} so this value wins.
2026-08-10 23:25:02 -07:00
Frank bf14ac7a7b fix: run Ling-3 vLLM from a cwd with no top-level vllm/ dir
The launcher does not change directory, so process cwd becomes sys.path[0]
for the 'python -m ...registry' model-inspection subprocess. From the CIRU
package root, the top-level vllm/ repo dir shadows the editable install as a
namespace package, breaking lazy imports (SamplingParams) and failing model
architecture inspection. cd to /home/dank/ling3-ciru before exec.
2026-08-10 23:16:46 -07:00
Frank 4a17948443 feat: add Ling-3.0-flash vLLM service for Dankclaw (frank-default-large, port 7770)
- serve-ling3-vllm.sh: wrapper around jcbtc validated run-256k.sh (TRITON_MLA,
  MTP K1, 256k, 6 slots, ROCr idle-fix LD_PRELOAD); sets HOST/PORT/
  SERVED_MODEL_NAME=frank-default-large
- frank-default-large.service: systemd unit, User=dank, port 7770
2026-08-10 22:36:37 -07:00