Fix distill model bos and eos token

[zechengz]

The bos_token_id and eos_token_id appear to be reversed in the create_output_embeddings_from_model_name function within model2vec/distill/inference.py.

When I testing with following code

from transformers import AutoModel, AutoTokenizer
from model2vec.distill import distill_from_model
model_name = "bert-base-uncased"
model = AutoModel.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
m2v_model = distill_from_model(model=model, tokenizer=tokenizer, pca_dims=256)

before the fix bos_token_id is 102 and eos_token_id is 101 but for BERT tokenizer the correct bos_token_id and eos_token_id should be

>>> tokenizer.cls_token_id
101
>>> tokenizer.sep_token_id
102

[Pringled]

Great catch, thanks for fixing this! As a quick sanity check I re-ran a few benchmarks and the results don't change much fortunately, but I will dive a bit deeper into different tasks and see if this improves performance somewhere, which I expect it will.