TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Finetune – Scikit-learn style model finetuning for NLP

52 pointsby madisonmayalmost 7 years ago

3 comments

ovi256almost 7 years ago
This is another good result of applying transfer learning to NLP.<p>Transfer learning works great for vision problems (just reuse one of the big SoTA trained on ImageNet networks - I like resnet50). This was enabled by the amazingly shared structure of vision problems. There was nothing similar for NLP, besides pre-trained first layers like word2vec. If you want to learn more, check out the fast.ai DL course, it features transfer learning a lot.<p>But this model and ULMFiT (nlp.fast.ai) show that deeper nets can be pretrained for NLP, and achieve good results when transfered to other datasets and problems.<p>This enables not just the obvious use case of &quot;I don&#x27;t have N GPUs to train a deep net from scratch but I can now finetune a pre-trained model&quot; but more subtle and interesting cases like fine-tuning on a very small dataset (compared to ImageNet or 100000 samples NLP data sets) and cheap training on demand. Training a new model for every user was way too expensive if training from scratch, but if fine-tuning a pre-trained net takes just a few minutes, why not ?
评论 #17609085 未加载
评论 #17609834 未加载
Tarq0nalmost 7 years ago
*a very specific implementation of NLP.<p>Not that this library isn&#x27;t promising, but the name and presentation makes it seem far more general than it really is.
staredalmost 7 years ago
In that spirit, and most likely much more general, for PyTorch:<p><a href="https:&#x2F;&#x2F;pytoune.org&#x2F;" rel="nofollow">https:&#x2F;&#x2F;pytoune.org&#x2F;</a> (Keras-like interface for PyTorch) and <a href="https:&#x2F;&#x2F;github.com&#x2F;dnouri&#x2F;skorch" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;dnouri&#x2F;skorch</a> (Scikit-learn interface for PyTorch).<p>As a side note, a project of mine: super-simple Jupyter Notebook training plots for Keras and PyToune: <a href="https:&#x2F;&#x2F;github.com&#x2F;stared&#x2F;livelossplot" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;stared&#x2F;livelossplot</a> (with bare API, so you can connect it to anything you wish)