Spaces:

programmersd
/

llama-gemma

Sleeping

programmersd commited on 7 days ago

Commit

5937211

verified ·

1 Parent(s): aa65bdd

Update Dockerfile

Files changed (1) hide show

Dockerfile CHANGED Viewed

@@ -2,8 +2,18 @@ FROM ghcr.io/ggml-org/llama.cpp:server
 ENV LLAMA_CACHE=/tmp/llama-cache
 ENV HF_HUB_CACHE=/tmp/hf-cache
-ENV LLAMA_ARG_WEBUI=1
 EXPOSE 7860
-CMD ["-hf", "unsloth/gemma-4-E2B-it-GGUF:Q4_0", "-c", "16384", "--host", "0.0.0.0", "--port", "7860", "--webui", "--no-warmup", "-ngl", "0"]

 ENV LLAMA_CACHE=/tmp/llama-cache
 ENV HF_HUB_CACHE=/tmp/hf-cache
 EXPOSE 7860
+CMD ["-hf", "unsloth/gemma-4-E2B-it-GGUF:Q4_0", \
+"-c", "2048", \
+"-t", "8", \
+"-tb", "16", \
+"-b", "1024", \
+"-ub", "512", \
+"-np", "1", \
+"--flash-attn", \
+"--no-warmup", \
+"--no-context-shift", \
+"--host", "0.0.0.0", \
+"--port", "7860"]