Update Dockerfile

Dockerfile

@@ -1,71 +1,34 @@
-# ╔══════════════════════════════════════════════════════════════╗
-# ║  STAGE 1 — builder: baixa + converte modelo (com torch)     ║
-# ╚══════════════════════════════════════════════════════════════╝
-FROM python:3.11-slim AS builder
-
-WORKDIR /build
-
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    build-essential git curl patchelf \
-    && rm -rf /var/lib/apt/lists/*
-
-# Torch CPU separado (evita conflito de index-url)
-RUN pip install --no-cache-dir \
-    torch==2.3.0 \
-    --index-url https://download.pytorch.org/whl/cpu
-
-# Resto das deps de build
-COPY requirements-build.txt .
-RUN pip install --no-cache-dir -r requirements-build.txt && \
-    find /usr/local/lib/python3.11/site-packages/ \
-         -name "libctranslate2*.so*" \
-         -print0 | xargs -0 -r patchelf --clear-execstack
-
-COPY app/download_model.py ./app/
-
-ENV CT2_MODEL_DIR=/build/ct2_model  \
-    TOKENIZER_DIR=/build/tokenizer  \
-    TOKENIZER_ID=google/flan-t5-xl  \
-    QUANTIZATION=int8               \
-    HF_HOME=/build/.cache
-
-RUN python app/download_model.py
-
-# ╔══════════════════════════════════════════════════════════════╗
-# ║  STAGE 2 — runtime: serve API (sem torch, imagem leve)      ║
-# ╚══════════════════════════════════════════════════════════════╝
 FROM python:3.11-slim
 
-ENV PORT=7860               \
-    PYTHONDONTWRITEBYTECODE=1 \
-    PYTHONUNBUFFERED=1        \
-    CT2_MODEL_DIR=/app/ct2_model \
-    TOKENIZER_DIR=/app/tokenizer \
-    CT2_MODEL_ID=google/flan-t5-xl \
-    MAX_NEW_TOKENS=1024       \
-    MAX_INPUT_LEN=512         \
-    INTER_THREADS=2           \
-    INTRA_THREADS=2
+ENV PORT=7860                        \
+    PYTHONDONTWRITEBYTECODE=1        \
+    PYTHONUNBUFFERED=1               \
+    HF_HOME=/app/.cache              \
+    TRANSFORMERS_CACHE=/app/.cache   \
+    MODEL_ID=google/flan-t5-large    \
+    MAX_NEW_TOKENS=1024              \
+    MAX_INPUT_LEN=512
 
 WORKDIR /app
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
-    patchelf \
+    build-essential curl \
     && rm -rf /var/lib/apt/lists/*
 
+# Torch CPU-only separado (evita baixar 2GB+ da versão CUDA)
+RUN pip install --no-cache-dir \
+    torch \
+    --index-url https://download.pytorch.org/whl/cpu
+
 COPY requirements.txt .
-RUN pip install --no-cache-dir -r requirements.txt && \
-    find /usr/local/lib/python3.11/site-packages/ \
-         -name "libctranslate2*.so*" \
-         -print0 | xargs -0 -r patchelf --clear-execstack
+RUN pip install --no-cache-dir -r requirements.txt
 
-# Copia apenas os artefatos do stage 1 (sem torch, sem modelo original)
-COPY --from=builder /build/ct2_model  /app/ct2_model
-COPY --from=builder /build/tokenizer  /app/tokenizer
+COPY app/ ./app/
 
-COPY app/main.py ./app/
+RUN useradd -m -u 1000 appuser && \
+    mkdir -p /app/.cache && \
+    chown -R appuser:appuser /app
 
-RUN useradd -m -u 1000 appuser && chown -R appuser:appuser /app
 USER appuser
 
 EXPOSE 7860