added nltk_data creation

Dockerfile

@@ -19,9 +19,21 @@ RUN apt-get update && apt-get install -y \
     wget \
     git git-lfs \
     bash \
-    && rm -rf /var/lib/apt/lists/*
+    && rm -rf /var/lib/apt/lists/*2
 # reflex needs unzip curl 
 
+RUN mkdir -p /nltk_data
+RUN chmod -R 777 /nltk_data 
+
+RUN chmod -R 777 /usr/local/lib/python3.10/site-packages/llama_index/legacy/_static/nltk_cache
+RUN mkdir -p /.cache
+RUN chmod -R 777 /.cache
+
+ENV TRANSFORMERS_CACHE=/usr/local/lib/python3.10/site-packages/llama_index/legacy/_static/nltk_cache
+# ^ not elegant but it works
+# HF warning says that TRANSFORMERS_CACHE will be deprecated in transformers v5, and advise to use HF_HOME
+
+
 RUN pip install --no-cache-dir --upgrade -r app/requirements.txt
 # ^ no caching of the packages to save space
 RUN pip install --no-cache-dir --upgrade -r requirements.txt
@@ -36,16 +48,4 @@ RUN guardrails hub install hub://guardrails/detect_pii
 RUN guardrails hub install hub://guardrails/qa_relevance_llm_eval
 
 
-# RUN python -c "import nltk; nltk.download('stopwords')"
-# ^ to fix runtime error, see https://github.com/run-llama/llama_index/issues/10681
-# it didn't work, I had to do chmod below (as also suggested in the article)
-
-RUN chmod -R 777 /usr/local/lib/python3.10/site-packages/llama_index/legacy/_static/nltk_cache
-RUN mkdir -p /.cache
-RUN chmod -R 777 /.cache
-
-ENV TRANSFORMERS_CACHE=/usr/local/lib/python3.10/site-packages/llama_index/legacy/_static/nltk_cache
-# ^ not elegant but it works
-# HF warning says that TRANSFORMERS_CACHE will be deprecated in transformers v5, and advise to use HF_HOME
-
 CMD ["uvicorn", "app.main_reflex:app", "--host", "0.0.0.0", "--port", "7860"]