tt

app.py

@@ -1,15 +1,21 @@
-import gradio as gr
 import torch
-import torch.nn.functional as F
-from transformers import AutoTokenizer, AutoModelForCausalLM
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import gradio as gr
+
+# Set device: GPU if available, else CPU
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+# Load two small models and their tokenizer (you can replace these with your models)
+model_name_a = "distilgpt2"
+model_name_b = "sshleifer/tiny-gpt2"  # very small GPT2 variant for demo
+
+tokenizer = AutoTokenizer.from_pretrained(model_name_a)
 
-model_id = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
-)
-model.eval()
+model_a = AutoModelForCausalLM.from_pretrained(model_name_a).to(device)
+model_b = AutoModelForCausalLM.from_pretrained(model_name_b).to(device)
+
+model_a.eval()
+model_b.eval()
 
 def blend_generate(prompt, wa, wb):
     input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to(device)
@@ -24,35 +30,24 @@ def blend_generate(prompt, wa, wb):
     # Weighted sum of raw logits (before softmax)
     blended_logits = wa * logits_a + wb * logits_b
 
-    # Apply softmax safely to get valid probability distribution
+    # Softmax to get probabilities
     probs = torch.softmax(blended_logits, dim=-1)
 
-    # Sample token from valid probability distribution
+    # Sample one token from the blended distribution
     token = torch.multinomial(probs, 1)
     next_token_id = token.item()
-    next_token = tokenizer.decode([next_token_id])
 
-    return next_token
+    next_token = tokenizer.decode([next_token_id])
+    return prompt + next_token
 
+# Gradio UI
 with gr.Blocks() as demo:
-    gr.Markdown("## Blended Prompt Chat (TinyLlama)")
-    sysA = gr.Textbox(label="System Prompt A", value="You are assistant A.")
-    sysB = gr.Textbox(label="System Prompt B", value="You are assistant B.")
-    wA = gr.Slider(-5, 5, value=1.0, step=0.1, label="Weight A")
-    wB = gr.Slider(-5, 5, value=1.0, step=0.1, label="Weight B")
-    user_msg = gr.Textbox(label="User Message")
-    temp = gr.Slider(0.1, 2.0, value=1.0, step=0.1, label="Temperature")
-    top_p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top-p")
-    max_tokens = gr.Slider(1, 200, value=100, step=1, label="Max New Tokens")
-    output = gr.Textbox(label="Response")
+    prompt_input = gr.Textbox(label="Prompt", lines=2)
+    weight_a = gr.Slider(0, 1, value=0.5, label="Weight model A")
+    weight_b = gr.Slider(0, 1, value=0.5, label="Weight model B")
+    output_text = gr.Textbox(label="Output")
 
     btn = gr.Button("Generate")
-    btn.click(
-        blend_generate,
-        [sysA, sysB, wA, wB, user_msg, max_tokens, temp, top_p],
-        output,
-        show_progress=True,
-    )
-
-if __name__ == "__main__":
-    demo.launch()
+    btn.click(blend_generate, inputs=[prompt_input, weight_a, weight_b], outputs=output_text)
+
+demo.launch()