What are these sentence preprocessing used for?

McGill-NLP / llm2vec

Code for 'LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders'

https://mcgill-nlp.github.io/llm2vec/

MIT License

1.17k stars 88 forks source link

What are these sentence preprocessing used for? #95

Closed ShengYun-Peng closed 3 months ago

ShengYun-Peng commented 3 months ago

https://github.com/McGill-NLP/llm2vec/blob/faa365e94e31cf6c5794144411682e687e4e03e0/llm2vec/llm2vec.py#L297C7-L301C70

It reads like these lines are never used in the training and inference. Plus, what is MEDI version of MTEB?

vaibhavad commented 3 months ago

They are used when evaluating on MTEB. Line 300-301 is used when documents are passed without any instructions, so that the code after this point is standardized to deal with both queries with instructions and documents without instruction.

what is MEDI version of MTEB?

I think this can be ignored. It just means MTEB with instructions.

vaibhavad commented 3 months ago

Closing as it is stale. Feel free to re-open if you have any more questions.