added new class

app.py

@@ -5,11 +5,16 @@ import gradio as gr
 
 iface = gr.ChatInterface(
         fn=handle_query,  # Query handling remains here
-        title="PDF Information and Inference",
-        description="Retrieval-Augmented Generation - Ask me anything about the content of the PDF.",
+        title="Odi, la chatbot degli Osservatori",
+        description="Retrieval-Augmented Generation - Ask me anything about the research carried out at the Osservatori.",
     )
 
+css = """
+body {
+    background-color: #ADD8E6;  /* Light blue background */
+}
+"""
 
 if __name__ == "__main__":
-    progress=gr.Progress(track_tqdm=True)
-    iface.launch(debug=True)
+    progress = gr.Progress(track_tqdm=True)
+    iface.launch(debug=True, css=css)

backend.py

@@ -20,14 +20,14 @@ login(huggingface_token)
 
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
-model_id = "google/gemma-2-2b-it"
+"""model_id = "google/gemma-2-2b-it"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id,
     device_map="auto", ## change this back to auto!!!
     torch_dtype= torch.bfloat16 if torch.cuda.is_available() else torch.float32,
     token=True)
-model.eval()
+model.eval()"""
 
 #from accelerate import disk_offload
 #disk_offload(model=model, offload_dir="offload")

interface.py

@@ -5,7 +5,7 @@ from typing import Any
 import torch
 from transformers import TextIteratorStreamer
 from threading import Thread
-from pydantic import Field
+from pydantic import Field, field_validator
 
 # for transformers 2
 class GemmaLLMInterface(CustomLLM):
@@ -15,6 +15,13 @@ class GemmaLLMInterface(CustomLLM):
     tokenizer: Any = Field(default=None)
     model: Any = Field(default=None)
 
+    @field_validator('context_window', 'num_output', mode='before')
+    @classmethod
+    def ensure_integer(cls, v):
+        if isinstance(v, str):
+            return int(v)
+        return v
+
     def __init__(self, **data):
         super().__init__(**data)
         self.tokenizer = AutoTokenizer.from_pretrained(self.model_id)