cfg-0123
citable URL: https://halobench.com/records/cfg-0123/ — this address never moves; the anchor /records/#cfg-0123 keeps resolving
| model | NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Q4_K_M.gguf · Q4_K_M |
| engine | ggml-org/llama.cpp 7077abb · rocm · host aihydra (igpu) |
| flags | -ngl 999 -dev ROCm0 -fa on -ctk f16 -ctv f16 -c 32768 --load-mode none --parallel 1 --host 127.0.0.1 --port 8090 --spec-type draft-mtp --spec-draft-n-max 1 -ctkd f16 -ctvd f16 |
| template | not recorded at test time |
| tree | upstream — stock |
Capability basis
measured on this config
Runs on this config (3)
| run | date | kind | suite | metrics |
|---|---|---|---|---|
| run-0462 | 2026-08-19 | guard | house-capability-guard@1 | tasks_passed=4 · tasks_total=4 |
| run-0463 | 2026-08-19 | performance | mtp-engagement-probe@2 | decode_tps=75.58129137515279 · decode_tps_no_speculation=65.33099156803146 |
| run-0470 | 2026-08-19 | capability | tau2-bench-airline@668d3bc | tasks_passed=4 · tasks_total=5 |