Unrecognized configuration class <....LlamaConfig> for AutoModelForSeq2SeqLM

Beginners
1

Hi, the subject is a summary of the actual error trying to use TheBloke/OpenAssistant-Llama2-13B-Orca-8K-3319-GPTQ’.

Unrecognized configuration class <class ‘transformers.models.llama.configuration_llama.LlamaConfig’> for this kind of AutoModel: AutoModelForSeq2SeqLM

my code…

Install necessary library

!pip install transformers

Import necessary modules

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

Specify the model name

model_name = ‘TheBloke/OpenAssistant-Llama2-13B-Orca-8K-3319-GPTQ’ # replace this with your actual model name

Load the model and the tokenizer

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

The error:

ValueError                                Traceback (most recent call last)
<ipython-input-4-a4bfd7aa96fc> in <cell line: 3>()
      1 # Load the model and the tokenizer
      2 tokenizer = AutoTokenizer.from_pretrained(model_name)
----> 3 model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

/usr/local/lib/python3.10/dist-packages/transformers/models/auto/auto_factory.py in from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs)
    494                 pretrained_model_name_or_path, *model_args, config=config, **hub_kwargs, **kwargs
    495             )
--> 496         raise ValueError(
    497             f"Unrecognized configuration class {config.__class__} for this kind of AutoModel: {cls.__name__}.\n"
    498             f"Model type should be one of {', '.join(c.__name__ for c in cls._model_mapping.keys())}."

ValueError: Unrecognized configuration class <class 'transformers.models.llama.configuration_llama.LlamaConfig'> for this kind of AutoModel: AutoModelForSeq2SeqLM.
Model type should be one of BartConfig, BigBirdPegasusConfig, BlenderbotConfig, BlenderbotSmallConfig, EncoderDecoderConfig, FSMTConfig, GPTSanJapaneseConfig, LEDConfig, LongT5Config, M2M100Config, MarianConfig, MBartConfig, MT5Config, MvpConfig, NllbMoeConfig, PegasusConfig, PegasusXConfig, PLBartConfig, ProphetNetConfig, SwitchTransformersConfig, T5Config, UMT5Config, XLMProphetNetConfig.
1 Like
2

Encountered the same issue. Seems AutoModelForSeq2SeqLM and LlamaConfig do not support each other.

3

The problem is that Tensorflow doesnt support AutoModelForSeq2SeqLM and you gotta install PyTorch instead of TensorFlow. Try unistalling Tensorflow and install PyTorch

1 Like
4

@amirh-far I created a separate conda environment to keep Tensorflow and PyTorch separate. But got the same error with PyTorch as well.

I’m trying to run inference. Here’s the minimal code:
from transformers import pipeline
summarizer = pipeline(“summarization”, model=“MarcoBrigo11/llama3-samsum”)

Error message:
ValueError: Unrecognized configuration class <class 'transformers.models.llama.configuration_llama.LlamaConfig'> for this kind of AutoModel: AutoModelForSeq2SeqLM. Model type should be one of BartConfig, BigBirdPegasusConfig, BlenderbotConfig, BlenderbotSmallConfig, EncoderDecoderConfig, FSMTConfig, GPTSanJapaneseConfig, LEDConfig, LongT5Config, M2M100Config, MarianConfig, MBartConfig, MT5Config, MvpConfig, NllbMoeConfig, PegasusConfig, PegasusXConfig, PLBartConfig, ProphetNetConfig, Qwen2AudioConfig, SeamlessM4TConfig, SeamlessM4Tv2Config, SwitchTransformersConfig, T5Config, UMT5Config, XLMProphetNetConfig.

1 Like
5

I think that the Llama architecture just doesn’t support summarization via the API.

from transformers import pipeline
summarizer = pipeline("summarization", model="facebook/bart-large-cnn") # It works
summarizer = pipeline("text-generation", model="MarcoBrigo11/llama3-samsum") # Maybe it works
summarizer = pipeline("summarization", model="MarcoBrigo11/llama3-samsum") # Not works
1 Like
6

It’s odd because the model “MarcoBrigo11/llama3-samsum” appears when I filtered for “Text Summarization models” on Hugging Face. I guess the list needs more refinement.

1 Like