c111b3a9b0
Phase 01 (LLM Tuning): - Gemma4 26B: 74.65 t/s (fast) - Qwen 35B: 61.62 t/s (balanced) - Gemma4 31B: 16.0 t/s (deep-coder) - Qwen 27B: 16.7 t/s (deep-logic) - Qwen 122B: 8.95 t/s (ultra, GPU 1 only) Phase 02 (API Engine): - FastAPI reverse proxy on port 8000 - /engine/switch hot-swap with 503 protection - config/engine_models.json as single source of truth - Replaced 4 individual .bat files with unified engine File cleanup: - scripts/ 85 files -> 9 + _archive/ - Root .bat files -> _archive/
52 lines
1.1 KiB
JSON
52 lines
1.1 KiB
JSON
[
|
|
{
|
|
"name": "GPU1 only + Expert CPU + 8t",
|
|
"avg_tps": 8.74,
|
|
"best_tps": 8.75,
|
|
"vram_gpu0": 618,
|
|
"vram_gpu1": 6493,
|
|
"vram_total": 7111,
|
|
"pcie": "1, 4 | 4, 16",
|
|
"status": "OK"
|
|
},
|
|
{
|
|
"name": "GPU1 only + Expert CPU + 16t",
|
|
"avg_tps": 8.0,
|
|
"best_tps": 8.02,
|
|
"vram_gpu0": 619,
|
|
"vram_gpu1": 6493,
|
|
"vram_total": 7112,
|
|
"pcie": "1, 4 | 4, 16",
|
|
"status": "OK"
|
|
},
|
|
{
|
|
"name": "Both GPU (main=1) + Expert CPU + 8t",
|
|
"avg_tps": 4.71,
|
|
"best_tps": 4.75,
|
|
"vram_gpu0": 4220,
|
|
"vram_gpu1": 3779,
|
|
"vram_total": 7999,
|
|
"pcie": "3, 4 | 4, 16",
|
|
"status": "OK"
|
|
},
|
|
{
|
|
"name": "Both GPU (ts 0.2,0.8) + Expert CPU + 8t",
|
|
"avg_tps": 4.53,
|
|
"best_tps": 4.6,
|
|
"vram_gpu0": 2666,
|
|
"vram_gpu1": 5333,
|
|
"vram_total": 7999,
|
|
"pcie": "2, 4 | 4, 16",
|
|
"status": "OK"
|
|
},
|
|
{
|
|
"name": "GPU1 only + Expert CPU + 8t + b4096",
|
|
"avg_tps": 8.73,
|
|
"best_tps": 8.77,
|
|
"vram_gpu0": 615,
|
|
"vram_gpu1": 6895,
|
|
"vram_total": 7510,
|
|
"pcie": "1, 4 | 4, 16",
|
|
"status": "OK"
|
|
}
|
|
] |