Helw150

from transformers import AutoTokenizer, T5ForConditionalGeneration

# Model Init
n_gpu = 8
tokenizer = AutoTokenizer.from_pretrained("google/flan-ul2")
model = T5ForConditionalGeneration.from_pretrained("google/flan-ul2")
heads_per_gpu = len(model.encoder.block) // n_gpu 
device_map = {
    gpu: list(
        range(

Helw150

from typing import List, Optional, Tuple, Union
from torchtyping import TensorType

from transformers.adapters.modeling import Adapter
from transformers.adapters import (
    BartAdapterModel,
    RobertaAdapterModel,
    BertAdapterModel,
    AdapterConfig,
)

Helw150

import ast

# To Delete After Debug
import code
import copyreg
import datetime
import functools
import json
import os
import re

Helw150

    def _push_parquet_shards_to_hub(                                                                                                                                                                                                  [1071/1877]
        self,                                                                                                           
        repo_id: str,         
        data_dir: str = "data",          
        split: Optional[str] = None,    
        token: Optional[str] = None,      
        revision: Optional[str] = None,   
        create_pr: Optional[bool] = False,               
        max_shard_size: Optional[Union[int, str]] = None,
        num_shards: Optional[int] = None,             

Helw150

text = # Tokenized Text Corresponding to Recording Transcript
audio = # Mel Spectrogram of the Recording

# Only Train Connector and Projection
self.encoder.freeze()
self.llama.freeze()

# Convert Raw Audio Signal to 1500 Embeddings with Whisper Encoder (CNN+Transformer)
audio_features = self.encoder(audio)

Helw150

from time import sleep

from datasets import load_dataset
from huggingface_hub import InferenceClient
from ratelimit import limits, sleep_and_retry
from transformers import AutoTokenizer

dataset = load_dataset("yijingwu/HeySQuAD_human", split="train")

tokenizer = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct")