Update app.py

app.py

@@ -14,9 +14,7 @@ else:
 # Load the model and config when the script starts
 peft_model_id = "phearion/bigbrain-v0.0.1"
 config = PeftConfig.from_pretrained(peft_model_id)
-model = AutoModelForCausalLM.from_pretrained(
-    config.base_model_name_or_path,                          
-    torch_dtype=torch.bfloat16)
+model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path)
 tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
 
 # Load the Lora model
@@ -27,7 +25,7 @@ def greet(text):
     
     # Use torch.no_grad to disable gradient calculation
     with torch.no_grad():
-        output_tokens = model.generate(**batch, do_sample=True, max_new_tokens=15)
+        output_tokens = model.generate(**batch, do_sample=True, max_new_tokens=20)
 
     return tokenizer.decode(output_tokens[0], skip_special_tokens=True)