Question about training sample

shibing624 / text2vec

text2vec, text to vector. 文本向量表征工具，把文本转化为向量矩阵，实现了Word2Vec、RankBM25、Sentence-BERT、CoSENT等文本表征、文本相似度计算模型，开箱即用。

https://pypi.org/project/text2vec/

Apache License 2.0

4.48k stars 396 forks source link

Question about training sample #92

Closed callanwu closed 1 year ago

callanwu commented 1 year ago

Describe the Question

If I only have approximately 2,000 labeled training samples containing values of 0 and 1, how can I effectively fine-tune a Word2Vec model based "shibing624/text2vec-base-chinese" ? Can it still yield useful results?

programmeguru commented 1 year ago

I think it may not better than result of not fine-tuning

callanwu commented 1 year ago

right, it got worse results after fine-tuning.

callanwu commented 1 year ago

thanks for yr reply!

programmeguru commented 1 year ago

welcome~