Update app.py

app.py

@@ -45,10 +45,15 @@ def describe_image(image, user_question="Solve this AP Problem step by step and
 
         # Load and prepare the model
         vl_gpt = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True).to(torch.float16).eval()
+        vl_gpt = vl_gpt.to(torch.float16)
 
         # Generate embeddings from the image input
         inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs).to(dtype=torch.float16)
 
+        print(f"Inputs Embeds dtype: {inputs_embeds.dtype}")
+        print(f"Attention Mask dtype: {attention_mask.dtype}")
+        print(f"Model dtype: {next(vl_gpt.parameters()).dtype}")
+
         # Generate the model's response
         outputs = vl_gpt.language_model.generate(
             inputs_embeds=inputs_embeds,
@@ -60,6 +65,7 @@ def describe_image(image, user_question="Solve this AP Problem step by step and
             do_sample=False,
             use_cache=True
         )
+        outputs = outputs.to(torch.float16)
 
         # Decode the generated tokens into text
         answer = tokenizer.decode(outputs[0].cpu().tolist(), skip_special_tokens=True)