Update app.py

app.py

@@ -38,7 +38,7 @@ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 peft_model_id = "sooolee/flan-t5-base-cnn-samsum-lora"
 config = PeftConfig.from_pretrained(peft_model_id)
 tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
-model = AutoModelForSeq2SeqLM.from_pretrained(config.base_model_name_or_path, device_map='auto') # load_in_8bit=True,
+model = AutoModelForSeq2SeqLM.from_pretrained(config.base_model_name_or_path, load_in_8bit=True, device_map='auto') # load_in_8bit=True,
 model = PeftModel.from_pretrained(model, peft_model_id, device_map='auto')
 
 def summarize(video_id):
@@ -51,12 +51,12 @@ def summarize(video_id):
         transcript += dict[i]['text']
 
     texts = preprocessing(transcript)
-    inputs = tokenizer(*texts, return_tensors="pt", padding=True, )
+    inputs = tokenizer(texts, return_tensors="pt", padding=True, )
     inputs = inputs["input_ids"].to(device)
 
     with torch.no_grad():
-        output_tokens = model.generate(*inputs, max_new_tokens=60, do_sample=True, top_p=0.9)
-        outputs = tokenizer.batch_decode(output_tokens[0].detach().cpu().numpy(), skip_special_tokens=True)
+        output_tokens = model.generate(inputs, max_new_tokens=60, do_sample=True, top_p=0.9)
+        outputs = tokenizer.batch_decode(output_tokens.detach().cpu().numpy(), skip_special_tokens=True)
     
     return outputs