wmantly/turing-multi-gpu-llm-server
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update start-server.sh with validated ngram context-lookup speculation and NCCL pipeline flags

Browse Source
This commit is contained in:
wmantly
2026-09-01 15:56:43 +00:00
parent 2b4e826437
commit 261631457c
1 changed files with 3 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/start-server.sh
+3
View File
@@ -13,6 +13,9 @@ export NCCL_PROTO=SIMPLE
exec /opt/llama.cpp/build-nccl/bin/llama-server \ exec /opt/llama.cpp/build-nccl/bin/llama-server \
-m /opt/models/gguf/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-Q4_K_P.gguf \ -m /opt/models/gguf/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-Q4_K_P.gguf \
--spec-type ngram-simple \
--spec-ngram-simple-size-n 24 \
--spec-ngram-simple-size-m 4 \
--mmproj /opt/models/gguf/mmproj-Qwen3.8-27B-Uncensored-f16.gguf \ --mmproj /opt/models/gguf/mmproj-Qwen3.8-27B-Uncensored-f16.gguf \
-ngl 99 \ -ngl 99 \
-c 262144 \ -c 262144 \