Upload 2 files

Dockerfile

@@ -1,13 +1,12 @@
 FROM python:3.10
 
-# Set workdir
-WORKDIR /
-
-# Copy files
-COPY . /
+WORKDIR /app
 
 # Install dependencies
+COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Run the app with uvicorn
-CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]
+# Copy app code
+COPY . .
+
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

model.py

@@ -5,11 +5,13 @@ import torch
 BASE_MODEL = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 LORA_REPO = "sahil239/chatbot-v2"
 
-tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
+
+tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, trust_remote_code=True)
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
     torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
-    device_map="auto"
+    device_map="auto",
+    trust_remote_code=True
 )
 model = PeftModel.from_pretrained(model, LORA_REPO)
 model.eval()