BERT memorisation and pitfalls in low-resource scenarios
— AK (@ak92501) May 4, 2021
pdf: https://t.co/XS8uUh81bG
abs: https://t.co/Adfmq5Etix
enabling BERT to perform well in extremely low-resource scenarios and also achieves comparable performance in non low-resource settings pic.twitter.com/P8dD9oQe2k
Looking to scale BERT inference on CPU? 🧐
— Hugging Face (@huggingface) April 20, 2021
Wonder how we maximize hardware utilization while optimizing latency/throughput? 📈
Find out on our latest blog post: Scaling up BERT model inference on CPU. 👀https://t.co/48JLMFj984
https://t.co/riRO1eqBQw uses Open AI CLIP for finding red sky in video from https://t.co/7HyH8Eex1i https://t.co/rhInmyiyLL pic.twitter.com/fZa9iR1bpJ
— AK (@ak92501) April 17, 2021
XTREME-R: Towards More Challenging and Nuanced Multilingual Evaluation
— Sebastian Ruder (@seb_ruder) April 16, 2021
We examine the state of multilingual benchmarking and propose an improved benchmark covering more challenging tasks, including a diagnostic and evaluation suite to inform future work.https://t.co/QCppOeNrV4 pic.twitter.com/pR8FRauZH6
(1/2) 🚨 Our new work! 🚨 "Retrieval Augmentation Reduces Hallucination in Conversation" @shtruk @spencerpoff @moyapchen @douwekiela @jasewestonhttps://t.co/r0W6xRuffk
— Jason Weston (@jaseweston) April 16, 2021
We infuse dialogue models with knowledge, significantly reducing hallucinated facts during conversation. pic.twitter.com/1rY1GvYTGg
10 Things You Need to Know About BERT and the Transformer Architecture That Are Reshaping the AI Landscape
— Sebastian Ruder (@seb_ruder) April 9, 2021
This super comprehensive post covers most things that are important in current NLP including BERT, transfer and avocado chairs 🥑
by @cathalhoran https://t.co/Hn458pkrk8 pic.twitter.com/srs4bIFrFw
Interesting reproducibility study by @philipvollet to conclude that: ELECTRA small (14M) can be easily run on a single GPU with great performances! @clark_kev @quocleix @chrmanning
— Thang Luong (@lmthang) April 8, 2021
Reproducibility paper: https://t.co/UAe56B74EA
ELECTRA paper: https://t.co/1GW6scqTIv https://t.co/DoPGGMf3ro
Need more fine-tuning data? A prompt is worth a thousand data points. @srush_nlp + @Fluke_Ellington show steering models with GPT3-style prompts during fine-tuning outperforms standard linear classifiers in their #naacl2021 paper. See our interactive blog https://t.co/lIDHiq2359 pic.twitter.com/n1gWC2Xf4t
— Hugging Face (@huggingface) April 7, 2021
🚨 New cranky position paper alert! 🚨
— Prof. Sam Bowman (@sleepinyourhat) April 7, 2021
(#naacl2021, mini-thread, https://t.co/wRvhZ70bbV) pic.twitter.com/JQwSAoeb4E
Language-based Video Editing via Multi-Modal Multi-Level Transformer
— AK (@ak92501) April 5, 2021
pdf: https://t.co/APK6dVUCyO
abs: https://t.co/IGCqPC2zWH pic.twitter.com/j9XOJQZCOb
StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery
— AK (@ak92501) April 1, 2021
pdf: https://t.co/20g03cMRYu
abs: https://t.co/XN4XjVX7Pi
github: https://t.co/6fxI9mDXUN pic.twitter.com/DNRHX3bmtk
Last week, EleutherAI released two checkpoints for GPT Neo, an *Open Source* replication of OpenAI's GPT-3
— Hugging Face (@huggingface) March 31, 2021
These checkpoints, of sizes 1.3B and 2.7B are now available in🤗Transformers!
The generation capabilities are truly🤯, try it now on the Hub: https://t.co/LbLXuCNYip pic.twitter.com/eHq45J5tdu
