chore: apply all changes
This commit is contained in:
@@ -232,7 +232,7 @@ llm_swapmode_models:
|
||||
gguf_path: "{{ llm_models_dir }}/Qwen3.8-27B-Q4_K_M.gguf"
|
||||
port: 8105
|
||||
n_gpu_layers: -1 # -1 = auto-detect / all layers to GPU
|
||||
ctx_size: 65536
|
||||
ctx_size: 131072 # t_ca3ff615: 65536 -> 131072 (VRAM pre-validated ~20.3GB < 24GB, t_441470b9)
|
||||
batch_size: 4096
|
||||
ubatch_size: 4096
|
||||
parallel: 1
|
||||
|
||||
Reference in New Issue
Block a user