Version of Transformer

[Rossssson]

Thanks for your wonderful work and open source. Can you provide the specific version of dependency 'transfomer'? I didn't find in the requirements.txt.

[raytrun]

Thanks for you kind words.
For the transformers dependency, the compatible version we recommend is 4.56.2

[Disguiser15]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

[Rossssson]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

I used llm2vec package in this repo 'llm2clip/llm2vec', it works fine for me.

[Disguiser15]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

I used llm2vec package in this repo 'llm2clip/llm2vec', it works fine for me.

Thank you for your help!

[crowang1A]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

I used llm2vec package in this repo 'llm2clip/llm2vec', it works fine for me.

Hi, thank you for the information! I tried using transformers==4.56.2 together with the llm2vec package included in this repo (llm2clip/llm2vec), but I still encountered import errors.

First, bidirectional_mistral.py fails with:
ImportError: cannot import name 'MistralFlashAttention2' from transformers.models.mistral.modeling_mistral

After skipping the Mistral import, bidirectional_llama.py then fails with:
ImportError: cannot import name 'LlamaFlashAttention2' from transformers.models.llama.modeling_llama

Could you share whether any local modifications were made to llm2clip/llm2vec? Thank you!

[Rossssson]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

I used llm2vec package in this repo 'llm2clip/llm2vec', it works fine for me.

Hi, thank you for the information! I tried using transformers==4.56.2 together with the llm2vec package included in this repo (llm2clip/llm2vec), but I still encountered import errors.

First, bidirectional_mistral.py fails with: ImportError: cannot import name 'MistralFlashAttention2' from transformers.models.mistral.modeling_mistral

After skipping the Mistral import, bidirectional_llama.py then fails with: ImportError: cannot import name 'LlamaFlashAttention2' from transformers.models.llama.modeling_llama

Could you share whether any local modifications were made to llm2clip/llm2vec? Thank you!

Hi, I used transformers == 4.44.0, the MistralFlashAttention2 package seems to be refactored in the later version of transformers.

[crowang1A]

Thanks for you kind words. For the transformers dependency, the compatible version we recommend is 4.56.2

May I ask which version of llm2vec corresponds to this? I am encountering conflict issues when using version 2.0.3.

I used llm2vec package in this repo 'llm2clip/llm2vec', it works fine for me.

Hi, thank you for the information! I tried using transformers==4.56.2 together with the llm2vec package included in this repo (llm2clip/llm2vec), but I still encountered import errors.
First, bidirectional_mistral.py fails with: ImportError: cannot import name 'MistralFlashAttention2' from transformers.models.mistral.modeling_mistral
After skipping the Mistral import, bidirectional_llama.py then fails with: ImportError: cannot import name 'LlamaFlashAttention2' from transformers.models.llama.modeling_llama
Could you share whether any local modifications were made to llm2clip/llm2vec? Thank you!

Hi, I used transformers == 4.44.0, the MistralFlashAttention2 package seems to be refactored in the later version of transformers.

Thank you for your help!

[hguoqing522-droid]

Excuse me, sir.No matter the default transformers or transformers == 4.44.0,it doesn't work with this error "ImportError: cannot import name 'GradientCheckpointingLayer' from 'transformers.modeling_layers' (/home/hgq/anaconda3/envs/llm2clip/lib/python3.8/site-packages/transformers/modeling_layers.py)".