Update app.py

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import torch
-from transformers import pipeline, WhisperProcessor, WhisperForConditionalGeneration, AutoModelForSeq2SeqLM, AutoTokenizer
+from transformers import pipeline, WhisperProcessor, WhisperForConditionalGeneration, AutoModelForCausalLM, AutoTokenizer
 from huggingface_hub import login
 import os
 
@@ -18,7 +18,7 @@ whisper_processor = WhisperProcessor.from_pretrained("openai/whisper-base")
 whisper_model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
 
 # Initialize the summarization model and tokenizer
-summarization_model = AutoModelForSeq2SeqLM.from_pretrained("meta-llama/Llama-2-7b-hf")
+summarization_model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-hf")
 summarization_tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf")
 
 # Function to transcribe audio
@@ -32,8 +32,8 @@ def transcribe_audio(audio_file):
 
 # Function to summarize text
 def summarize_text(text):
-    inputs = summarization_tokenizer.encode("summarize: " + text, return_tensors="pt", max_length=512, truncation=True)
-    summary_ids = summarization_model.generate(inputs, max_length=150, min_length=40, length_penalty=2.0, num_beams=4, early_stopping=True)
+    inputs = summarization_tokenizer(text, return_tensors="pt", max_length=512, truncation=True)
+    summary_ids = summarization_model.generate(inputs.input_ids, max_length=150, min_length=40, length_penalty=2.0, num_beams=4, early_stopping=True)
     summary = summarization_tokenizer.decode(summary_ids[0], skip_special_tokens=True)
     return summary