Lumpiasty/klaster
0
0
Fork 0
Code Issues Pull Requests 4 Packages Projects Releases Wiki Activity

increase target margin of 2048MB of VRAM

Browse Source
This commit is contained in:
Lumpiasty
2026-03-06 02:41:34 +01:00
parent ba9db6ce41
commit cd7ebac6b9
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
apps/llama/configs/config.yaml
View File

@@ -461,6 +461,7 @@ models:
/app/llama-server /app/llama-server
-hf unsloth/Qwen3-Coder-Next-GGUF:Q4_K_M -hf unsloth/Qwen3-Coder-Next-GGUF:Q4_K_M
--ctx-size 32768 --ctx-size 32768
--fit-target 2048
--predict 8192 --predict 8192
--temp 1.0 --temp 1.0
--min-p 0.01 --min-p 0.01
@@ -475,6 +476,7 @@ models:
cmd: | cmd: |
/app/llama-server /app/llama-server
-hf unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M -hf unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M
--fit-target 2048
--ctx-size 16384 --ctx-size 16384
--temp 1.0 --temp 1.0
--min-p 0.00 --min-p 0.00
@@ -488,6 +490,7 @@ models:
cmd: | cmd: |
/app/llama-server /app/llama-server
-hf unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M -hf unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M
--fit-target 2048
--ctx-size 16384 --ctx-size 16384
--temp 1.0 --temp 1.0
--min-p 0.00 --min-p 0.00
@@ -506,6 +509,7 @@ models:
-hf mradermacher/Qwen3.5-35B-A3B-heretic-GGUF:Q4_K_M -hf mradermacher/Qwen3.5-35B-A3B-heretic-GGUF:Q4_K_M
--mmproj-url https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/resolve/main/mmproj-F16.gguf --mmproj-url https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/resolve/main/mmproj-F16.gguf
--mmproj /root/.cache/llama.cpp/unsloth_Qwen3.5-35B-A3B-GGUF_mmproj-F16.gguf --mmproj /root/.cache/llama.cpp/unsloth_Qwen3.5-35B-A3B-GGUF_mmproj-F16.gguf
--fit-target 2048
--ctx-size 16384 --ctx-size 16384
--temp 1.0 --temp 1.0
--min-p 0.00 --min-p 0.00
@@ -521,6 +525,7 @@ models:
-hf mradermacher/Qwen3.5-35B-A3B-heretic-GGUF:Q4_K_M -hf mradermacher/Qwen3.5-35B-A3B-heretic-GGUF:Q4_K_M
--mmproj-url https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/resolve/main/mmproj-F16.gguf --mmproj-url https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/resolve/main/mmproj-F16.gguf
--mmproj /root/.cache/llama.cpp/unsloth_Qwen3.5-35B-A3B-GGUF_mmproj-F16.gguf --mmproj /root/.cache/llama.cpp/unsloth_Qwen3.5-35B-A3B-GGUF_mmproj-F16.gguf
--fit-target 2048
--ctx-size 16384 --ctx-size 16384
--temp 1.0 --temp 1.0
--min-p 0.00 --min-p 0.00