Update app.py

app.py

@@ -10,11 +10,24 @@ device = 'cpu' # if you have a GPU
 tokenizer = T5Tokenizer.from_pretrained('stanfordnlp/SteamSHP-flan-t5-large')
 model = T5ForConditionalGeneration.from_pretrained('stanfordnlp/SteamSHP-flan-t5-large').to(device)
 
+HF_TOKEN = os.getenv("HF_TOKEN")
+
+OUTPUTS_DATASET = "HuggingFaceH4/instruction-pilot-outputs-filtered"
+
+ds = load_dataset(OUTPUTS_DATASET, split="train", use_auth_token=HF_TOKEN)
+
 def process():
-    input_text = "POST: Instacart gave me 50 pounds of limes instead of 5 pounds... what the hell do I do with 50 pounds of limes? I've already donated a bunch and gave a bunch away. I'm planning on making a bunch of lime-themed cocktails, but... jeez. Ceviche? \n\n RESPONSE A: Lime juice, and zest, then freeze in small quantities.\n\n RESPONSE B: Lime marmalade lol\n\n Which response is better? RESPONSE"
+    sample_ds = ds.shuffle().select(range(1))
+
+    
+    
+    df = pd.DataFrame.from_records(sample["filtered_outputs"])
+        
+    input_text = "POST: "+ sample["prompt"]+ "\n\n RESPONSE A: Lime juice, and zest, then freeze in small quantities.\n\n RESPONSE B: Lime marmalade lol\n\n Which response is better? RESPONSE"
     x = tokenizer([input_text], return_tensors='pt').input_ids.to(device)
     y = model.generate(x, max_new_tokens=1)
-    return tokenizer.batch_decode(y, skip_special_tokens=True)[0]
+    prefered = tokenizer.batch_decode(y, skip_special_tokens=True)[0] 
+    return sample["filtered_outputs"]
 
 title = "Compare Instruction Models to see which one is more helpful"
 interface = gr.Interface(fn=process,