Update Dockerfile

Dockerfile

@@ -115,7 +115,7 @@ COPY app.py .
 # ---------------------------------------------------------------------------
 ENV \
     # Where to look for the GGUF model on the Hub.
-    GGUF_REPO_ID="empero-ai/Qwythos-9B-Claude-Mythos-5-1M-GGUF" \
+    GGUF_REPO_ID="bartowski/google_gemma-3-2b-it-GGUF" \
     # Leave empty to auto-select; set to force an exact filename.
     GGUF_FILENAME="" \
     # Quantization to prefer when auto-selecting (falls back automatically
@@ -130,9 +130,9 @@ ENV \
     # baked-in YaRN scaling, but free CPU Spaces cannot allocate that much
     # KV-cache. Default is a safe value for a 16GB-RAM CPU Space; raise it
     # via the Space's Variables UI if you have more headroom.
-    N_CTX="8192" \
-    # 0 = auto-detect available CPU cores.
-    N_THREADS="0" \
+    N_CTX="4096" \
+    # Free tier has 2 vCPUs; more threads just adds contention.
+    N_THREADS="2" \
     N_BATCH="256" \
     MAX_NEW_TOKENS="1024" \
     TEMPERATURE="0.6" \
@@ -155,4 +155,4 @@ USER appuser
 
 EXPOSE 7860
 
-CMD ["python", "app.py"]
+CMD ["python", "app.py"]