push

Dockerfile

@@ -47,12 +47,13 @@ ENV TRANSFORMERS_CACHE=/models/huggingface
 ENV HUGGINGFACE_HUB_CACHE=/models/huggingface
 ENV HF_HUB_CACHE=/models/huggingface
 
+# Set HOME for gdown to create cache directory
+ENV HOME=/tmp
+
 # Create cache dir and models directory with proper permissions
 RUN mkdir -p /models/huggingface && \
     mkdir -p /code/models && \
-    mkdir -p /.cache/gdown && \
-    chmod -R 777 /models/huggingface && \
-    chmod -R 777 /.cache
+    chmod -R 777 /models/huggingface
 
 # Pre-download model at build time (YarnGPT2 model)
 RUN python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='saheedniyi/YarnGPT2')"

main.py

@@ -78,12 +78,11 @@ def load_audio_tokenizer():
             try:
                 import subprocess
                 os.makedirs("./models", exist_ok=True)
-                os.makedirs("/.cache/gdown", exist_ok=True)
                 # Download with gdown using --fuzzy flag
                 result = subprocess.run([
                     "gdown", "--fuzzy", "1-ASeEkrn4HY49yZWHTASgfGFNXdVnLTt", 
                     "-O", "./models/wavtokenizer_large_speech_320_24k.ckpt"
-                ], check=False, capture_output=True, text=True)
+                ], check=False, capture_output=True, text=True, env=os.environ.copy())
                 if result.returncode == 0 and os.path.exists("./models/wavtokenizer_large_speech_320_24k.ckpt"):
                     model_path = "./models/wavtokenizer_large_speech_320_24k.ckpt"
                     logger.info("Checkpoint downloaded successfully")