Update app.py

app.py

@@ -2,17 +2,18 @@ import gradio as gr
 import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
+# 🔥 Only open models that load in HF Spaces without gated access
 MODEL_OPTIONS = {
-    "Llama-3.2-3B": "meta-llama/Llama-3.2-3B-Instruct",
-    "Llama-3.2-1B": "meta-llama/Llama-3.2-1B-Instruct",
     "Mistral-7B-Instruct": "mistralai/Mistral-7B-Instruct-v0.1",
     "Qwen2.5-3B-Instruct": "Qwen/Qwen2.5-3B-Instruct",
     "Qwen2.5-1.5B-Instruct": "Qwen/Qwen2.5-1.5B-Instruct",
     "StableLM2-1.6B": "stabilityai/stablelm-2-zephyr-1_6b",
+    "SmolLM3-3B": "HuggingFaceTB/SmolLM3-3B",
+    "BTLM-3B-8k-base": "cerebras/btlm-3b-8k-base"
 }
 
 loaded = {}
-SYSTEM_PROMPT = "You are HugginGPT — a helpful assistant that remembers context."
+SYSTEM_PROMPT = "You are HugginGPT — helpful, friendly, and clear with memory."
 
 def load_model(model_key):
     model_id = MODEL_OPTIONS[model_key]
@@ -23,14 +24,16 @@ def load_model(model_key):
     model = AutoModelForCausalLM.from_pretrained(
         model_id,
         device_map="auto",
-        torch_dtype=torch.float16,
+        torch_dtype=torch.float16
     )
+
     loaded[model_key] = (tokenizer, model)
     return tokenizer, model
 
 def generate_response(message, history, model_choice):
     tokenizer, model = load_model(model_choice)
 
+    # build context with system + memory
     context = f"system: {SYSTEM_PROMPT}\n"
     if history:
         for u, a in history:
@@ -54,7 +57,7 @@ with gr.Blocks() as demo:
 
     model_selector = gr.Dropdown(
         choices=list(MODEL_OPTIONS.keys()),
-        value="Llama-3.2-3B",
+        value="Mistral-7B-Instruct",
         label="Select model"
     )