Upload app.py

app.py

@@ -9,7 +9,7 @@ os.makedirs(os.environ["HF_HOME"], exist_ok=True)
 os.makedirs(os.environ["XDG_CACHE_HOME"], exist_ok=True)
 
 import json
-from typing import Dict, List
+from typing import Dict, List, Generator
 
 import gradio as gr
 import requests
@@ -100,7 +100,20 @@ def load_model():
 	return _llm
 
 
+def generate_response_stream(prompt: str, temperature: float = 0.2, max_new_tokens: int = 384) -> Generator[str, None, None]:
+	yield "Initializing model..."
+	llm = load_model()
+	yield "Model loaded. Generating..."
+	accum = ""
+	for chunk in llm(prompt=prompt, max_tokens=max_new_tokens, temperature=temperature, stream=True):
+		text = chunk.get("choices", [{}])[0].get("text", "")
+		if text:
+			accum += text
+			yield accum
+
+
 def generate_response(prompt: str, temperature: float = 0.2, max_new_tokens: int = 384) -> str:
+	# non-streaming fallback
 	llm = load_model()
 	res = llm(
 		prompt=prompt,
@@ -244,8 +257,8 @@ def build_equity_research_prompt(symbol: str, overview: Dict) -> str:
 # -------- Gradio UI --------
 
 def ui_app():
-	with gr.Blocks(title="Fin-o1-14B Tools") as demo:
-		gr.Markdown("""# Fin-o1-14B Tools
+	with gr.Blocks(title="Fin-o1-8B Tools") as demo:
+		gr.Markdown("""# Fin-o1-8B Tools
 Two tabs: Price Prediction (Finnhub with Alpha Vantage fallback) and Equity Research (Alpha Vantage via RapidAPI).""")
 
 		with gr.Tab("Price Prediction"):
@@ -266,13 +279,9 @@ Two tabs: Price Prediction (Finnhub with Alpha Vantage fallback) and Equity Rese
 					except Exception as e2:
 						return f"Error fetching candles: {e2}"
 				prompt = build_price_prediction_prompt(sym, candles)
-				try:
-					resp = generate_response(prompt, temperature=temperature, max_new_tokens=int(max_tokens))
-					return resp
-				except Exception as e:
-					return f"Error generating response: {e}"
+				return generate_response_stream(prompt, temperature=temperature, max_new_tokens=int(max_tokens))
 
-			btn.click(on_predict, inputs=[symbol, resolution, count, temp, max_new], outputs=out)
+			btn.click(on_predict, inputs=[symbol, resolution, count, temp, max_new], outputs=out, show_progress=True)
 
 		with gr.Tab("Equity Research Report"):
 			symbol2 = gr.Textbox(label="Ticker (e.g., MSFT)", value="MSFT")
@@ -285,13 +294,14 @@ Two tabs: Price Prediction (Finnhub with Alpha Vantage fallback) and Equity Rese
 				try:
 					overview = fetch_alpha_vantage_overview(sym)
 					prompt = build_equity_research_prompt(sym, overview)
-					resp = generate_response(prompt, temperature=temperature, max_new_tokens=int(max_tokens))
-					return resp
+					return generate_response_stream(prompt, temperature=temperature, max_new_tokens=int(max_tokens))
 				except Exception as e:
 					return f"Error generating report: {e}"
 
-			btn2.click(on_report, inputs=[symbol2, temp2, max_new2], outputs=out2)
+			btn2.click(on_report, inputs=[symbol2, temp2, max_new2], outputs=out2, show_progress=True)
 
+		# Enable queue so users see pending/running state
+		demo.queue(concurrency_count=1, max_size=8)
 	return demo