Set TRITON_CACHE_DIR to /tmp/triton_cache to avoid root permission denied error

Dockerfile

@@ -10,6 +10,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 # Writable dirs for matplotlib and HF cache
 ENV MPLCONFIGDIR=/tmp/matplotlib
 ENV HF_HOME=/tmp/hf_cache
+ENV TRITON_CACHE_DIR=/tmp/triton_cache
 
 # Stage 1: Install torch first (flash-attn needs it at build time)
 RUN pip install --no-cache-dir torch==2.4.1
@@ -22,8 +23,8 @@ RUN pip install --no-cache-dir -r requirements.txt
 COPY . .
 
 # Create writable dirs AFTER copy
-RUN mkdir -p /app/models /app/outputs /app/logs /app/eval_logs /tmp/matplotlib /tmp/hf_cache && \
-    chmod -R 777 /app/models /app/outputs /app/logs /app/eval_logs /tmp/matplotlib /tmp/hf_cache /app
+RUN mkdir -p /app/models /app/outputs /app/logs /app/eval_logs /tmp/matplotlib /tmp/hf_cache /tmp/triton_cache && \
+    chmod -R 777 /app/models /app/outputs /app/logs /app/eval_logs /tmp/matplotlib /tmp/hf_cache /tmp/triton_cache /app
 
 EXPOSE 7860