I keep getting errors with my app. Trying to use Llama and Whisper to summarize an audio file. The error is:

Traceback (most recent call last):
File "/home/user/app/app.py", line 28, in
summarization_model = load_checkpoint_and_dispatch(
File "/usr/local/lib/python3.10/site-packages/accelerate/big_modeling.py", line 608, in load_checkpoint_and_dispatch
load_checkpoint_in_model(
File "/usr/local/lib/python3.10/site-packages/accelerate/utils/modeling.py", line 1708, in load_checkpoint_in_model
raise ValueError(
ValueError: checkpoint should be the path to a file containing a whole state dict, or the index of a sharded checkpoint, or a folder containing a sharded checkpoint or the whole state dict, but got meta-llama/Llama-2-7b-hf.
Token is valid (permission: read).
Your token has been saved in your configured git credential helpers (store).
Your token has been saved to /home/user/.cache/huggingface/token
Login successful
Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.

=========================== >>>>>>>>>>>>>>>> this is my code:

import gradio as gr
import torch
from transformers import pipeline, WhisperProcessor, WhisperForConditionalGeneration, AutoModelForCausalLM, AutoTokenizer
from huggingface_hub import login
import os
from accelerate import init_empty_weights, load_checkpoint_and_dispatch

Retrieve the token from the environment variable

hf_api_token = os.getenv("HF_API_TOKEN")

if hf_api_token is None:
raise ValueError("HF_API_TOKEN environment variable is not set")

Authenticate with Hugging Face

login(token=hf_api_token, add_to_git_credential=True)

Initialize the Whisper processor and model

whisper_processor = WhisperProcessor.from_pretrained("openai/whisper-base")
whisper_model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")

Initialize the summarization model and tokenizer

Load LLAMA 7B model with accelerate

model_name = "meta-llama/Llama-2-7b-hf"
with init_empty_weights():
summarization_model = AutoModelForCausalLM.from_pretrained(model_name)

Load checkpoint and dispatch model

summarization_model = load_checkpoint_and_dispatch(
summarization_model,
checkpoint=model_name,
device_map="auto",
dtype=torch.float16
)

summarization_tokenizer = AutoTokenizer.from_pretrained(model_name)

Function to transcribe audio

def transcribe_audio(audio_file):
# Load audio file
audio_input, _ = whisper_processor(audio_file, return_tensors="pt", sampling_rate=16000).input_values
# Generate transcription
transcription_ids = whisper_model.generate(audio_input)
transcription = whisper_processor.decode(transcription_ids[0])
return transcription

Function to summarize text

def summarize_text(text):
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

Gradio interface

def process_audio(audio_file):
transcription = transcribe_audio(audio_file)
summary = summarize_text(transcription)
return transcription, summary

Gradio UI

iface = gr.Interface(
fn=process_audio,
inputs=gr.Audio(type="file"),
outputs=[
gr.Textbox(label="Transcription"),
gr.Textbox(label="Summary")
],
title="Audio Transcription and Summarization",
description="Upload an audio file to transcribe and summarize the conversation."
)

Launch the app

iface.launch()