Update Dockerfile

Dockerfile

@@ -12,15 +12,18 @@ ENV HF_HOME="/tmp"
 # Corrected binary path
 ENTRYPOINT ["/app/llama-server"]
 
-# Run llama-server, explicitly fetching the exact GGUF file with custom generation defaults
+# Run llama-server with CPU-tier optimizations, keeping the 32k context
 CMD [ \
     "--hf-repo", "unsloth/LFM2.5-230M-GGUF", \
-    "--hf-file", "LFM2.5-230M-UD-Q8_K_XL.gguf", \
+    "--hf-file", "LFM2.5-230M-UD-Q4_K_XL.gguf", \
     "--host", "0.0.0.0", \
     "--port", "7860", \
     "-c", "32768", \
     "--threads", "2", \
     "--temp", "0.1", \
     "--top-k", "50", \
-    "--repeat-penalty", "1.05" \
+    "--repeat-penalty", "1.05", \
+    "--flash-attn", \
+    "--cache-type-k", "q8_0", \
+    "--cache-type-v", "q8_0" \
 ]