Unrecognized configuration class in mT5-small-finetuned-tydiqa-for-xqa

🤗Transformers
1

Hi,

I tried to run the multilingual question-answering with the mt5 model at https://huggingface.co/mrm8488/mT5-small-finetuned-tydiqa-for-xqa. Unfortunately I couldn’t and the following message appears:

Unrecognized configuration class <class ‘transformers.models.t5.configuration_t5.T5Config’> for this kind of AutoModel: AutoModelForCausalLM.
Model type should be one of CamembertConfig, XLMRobertaConfig, RobertaConfig, BertConfig, OpenAIGPTConfig, GPT2Config, TransfoXLConfig, XLNetConfig, XLMConfig, CTRLConfig, ReformerConfig, BertGenerationConfig, XLMProphetNetConfig, ProphetNetConfig.

How can this be solved?

Thanks!

2

Hi there, could you post the code snippet the raised this error?

3

Hi,

This is the code:

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
tokenizer = AutoTokenizer.from_pretrained("mrm8488/mT5-small-finetuned-tydiqa-for-xqa")
model = AutoModelForCausalLM.from_pretrained("mrm8488/mT5-small-finetuned-tydiqa-for-xqa").to(device)

def get_response(question, context, max_length=32):
  input_text = 'question: %s  context: %s' % (question, context)
  features = tokenizer([input_text], return_tensors='pt')

  output = model.generate(input_ids=features['input_ids'].to(device), 
               attention_mask=features['attention_mask'].to(device),
               max_length=max_length)

  return tokenizer.decode(output[0])

# Some examples in different languages

context = 'HuggingFace won the best Demo paper at EMNLP2020.'
question = 'What won HuggingFace?'
get_response(question, context)

context = 'HuggingFace ganó la mejor demostración con su paper en la EMNLP2020.'
question = 'Qué ganó HuggingFace?'
get_response(question, context)

context = 'HuggingFace выиграл лучшую демонстрационную работу на EMNLP2020.'
question = 'Что победило в HuggingFace?'
get_response(question, context)

It is the same you can find in https://huggingface.co/mrm8488/mT5-small-finetuned-tydiqa-for-xqa

Thanks!

4

the issue is MT5 is a seq2seq model and seq2seq models should be loaded using
AutoModelForConditionalGeneration or in this case directly using MT5ForConditionalGeneration class.

Also cc @mrm8488

5

Good morning

Thank you very much. Now the error is not raised

Thanks again

1 Like
6

I will fix it ASAP

7

Hi,
I am trying to fine tune the VITS Model but then this error raised
Here is the code:

import torch
from transformers import AutoTokenizer, Trainer, TrainingArguments, AutoModelForCausalLM
from custom_dataset import CustomDataset

# Paths and constants
train_filelist = "C:\\TTS\\vits\\Dataset\\train_filelist.csv"
val_filelist = "C:\\TTS\\vits\\Dataset\\val_filelist.csv"
model_name = "kakao-enterprise/vits-ljs"  # Pretrained VITS model
use_auth_token = "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"  # Replace with your actual Hugging Face token

# Load dataset
train_dataset = CustomDataset(train_filelist)
eval_dataset = CustomDataset(val_filelist)

# Initialize model and tokenizer with token authentication
model = AutoModelForCausalLM.from_pretrained(model_name, use_auth_token=use_auth_token)
tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=use_auth_token)

# Define a proper data collator
class CustomDataCollator:
    def __call__(self, batch):
        # Debug prints
        print("Batch received by data_collator:", batch)
        
        # Extract input_values and labels
        audio_data = [item['input_values'] for item in batch]
        text_data = [item['labels'] for item in batch]
        
        # Convert to tensors
        audio_tensor = torch.stack(audio_data)
        text_tensor = torch.stack(text_data)
        
        return {'input_values': audio_tensor, 'labels': text_tensor}

data_collator = CustomDataCollator()

# Training arguments
training_args = TrainingArguments(
    output_dir="./results",
    evaluation_strategy="epoch",
    learning_rate=1e-4,
    per_device_train_batch_size=4,
    per_device_eval_batch_size=4,
    num_train_epochs=3,
    logging_dir="./logs",
    save_steps=500,
    logging_steps=100,
    eval_steps=500,
)

# Initialize Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    tokenizer=tokenizer,
    data_collator=data_collator,
)

# Start training
trainer.train()

raise ValueError(
ValueError: Unrecognized configuration class <class ‘transformers.models.vits.configuration_vits.VitsConfig’> for this kind of AutoModel: AutoModelForCausalLM.
Model type should be one of BartConfig, BertConfig, BertGenerationConfig, BigBirdConfig, BigBirdPegasusConfig, BioGptConfig, BlenderbotConfig, BlenderbotSmallConfig, BloomConfig, CamembertConfig, LlamaConfig, CodeGenConfig, CohereConfig, CpmAntConfig, CTRLConfig, Data2VecTextConfig, DbrxConfig, ElectraConfig, ErnieConfig, FalconConfig, FuyuConfig, GemmaConfig, Gemma2Config, GitConfig, GPT2Config, GPT2Config, GPTBigCodeConfig, GPTNeoConfig, GPTNeoXConfig, GPTNeoXJapaneseConfig, GPTJConfig, JambaConfig, JetMoeConfig, LlamaConfig, MambaConfig, MarianConfig, MBartConfig, MegaConfig, MegatronBertConfig, MistralConfig, MixtralConfig, MptConfig, MusicgenConfig, MusicgenMelodyConfig, MvpConfig, OlmoConfig, OpenLlamaConfig, OpenAIGPTConfig, OPTConfig, PegasusConfig, PersimmonConfig, PhiConfig, Phi3Config, PLBartConfig, ProphetNetConfig, QDQBertConfig, Qwen2Config, Qwen2MoeConfig, RecurrentGemmaConfig, ReformerConfig, RemBertConfig, RobertaConfig, RobertaPreLayerNormConfig, RoCBertConfig, RoFormerConfig, RwkvConfig, Speech2Text2Config, StableLmConfig, Starcoder2Config, TransfoXLConfig, TrOCRConfig, WhisperConfig, XGLMConfig, XLMConfig, XLMProphetNetConfig, XLMRobertaConfig

How it will be solved?

Thanks