search

abhishekpanigrahi1996 / Word2Sense

Sparse Interpretable Word Embeddings
15 stars 0 forks source link

Different experimental results on different machines #4

Open Hezepeng opened 3 years ago

Hezepeng commented 3 years ago

Hello, I downloaded your corpus file and ran wacky.sh according to the experimental conditions in your paper But there are some differences between the running results and the paper. At the same time, I downloaded the pre-trained vectors file(Word2Sense.txt) which you provided, and directly used it to run the word similarity task. The task result is different from what I started training the corpus(combined_parsed.clean.gz ) again. Why does this happen?

Even with the same experimental parameters, the results of warplda will be different every time. Are you in the same situation？

Thank you for you paper and look forward your prompt reply

abhishekpanigrahi1996 commented 3 years ago

Hi, I am sorry I hadn't checked this issue. I will get back to you, after running the experiments on a different machine.