Update app.py

app.py

@@ -44,15 +44,15 @@ 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.bfloat16).eval()
+        vl_gpt = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True).to(torch.float16).eval()
 
         # Generate embeddings from the image input
-        inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs).to(dtype=torch.bfloat16)
+        inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs).to(dtype=torch.float16)
 
         # Generate the model's response
         outputs = vl_gpt.language_model.generate(
             inputs_embeds=inputs_embeds,
-            attention_mask = prepare_inputs.attention_mask.to(vl_gpt.device).to(dtype=torch.bfloat16),
+            attention_mask = prepare_inputs.attention_mask.to(vl_gpt.device).to(dtype=torch.float16),
             pad_token_id=tokenizer.eos_token_id,
             bos_token_id=tokenizer.bos_token_id,
             eos_token_id=tokenizer.eos_token_id,