|
Kevin Clark
@
clark_kev
|
|
PhD student with @stanfordnlp.
|
|
|
13
Tweetovi
|
122
Pratim
|
534
Osobe koje vas prate
|
| Tweetovi |
|
Kevin Clark
@clark_kev
|
29. sij |
|
@StanfordNLP Seminar tomorrow: @boknilev on "Causal Mediation Analysis for Interpreting Neural NLP: The Case of Gender Bias" nlp.stanford.edu/seminar/detail… Please join us!
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
22. sij |
|
@StanfordNLP Seminar tomorrow: @muthukumarc87 from Amazon on "A Discourse Centric Framework for Facilitating Instructor Intervention in MOOC Discussion Forums" nlp.stanford.edu/seminar/detail…. Please join us!
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
Hamlet 🇩🇴 🇺🇸
@hamletbatista
|
10. sij |
|
🔥“we train a model on one GPU for 4 days that outperforms GPT (trained using 30x more compute) on the GLUE natural language understanding ... we match the performance of RoBERTa, the current state-of-the-art pre-trained transformer, while using less than 1/4 of the compute.” 🤯 twitter.com/stanfordnlp/st…
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
10. sij |
|
Code will be released early February!
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
21. pro |
|
It is in tensorflow, but the neural architecture is the same as BERT so the pre-trained weights should be compatible with any pytorch implementation of BERT such as @huggingface's transformers library.
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
Sam Bowman
@sleepinyourhat
|
27. stu |
|
New analysis paper from my group! We zoom in on some of @clark_kev et al.'s on syntax-sensitive attention heads in BERT (+RoBERTa, +...), and find interestingly mixed results. twitter.com/phu_pmh/status…
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
Urvashi Khandelwal
@ukhndlwl
|
4. stu |
|
Excited to share new work!!! “Generalization through Memorization: Nearest Neighbor Language Models”
We introduce kNN-LMs, which extend LMs with nearest neighbor search in embedding space, achieving a new state-of-the-art perplexity on Wikitext-103, without additional training! pic.twitter.com/hehcLnDaKz
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
Mike Lewis
@ml_perception
|
31. lis |
|
Excited to share our work on BART, a method for pre-training seq2seq models by de-noising text. BART outperforms previous work on a bunch of generation tasks (summarization/dialogue/QA), while getting similar performance to RoBERTa on SQuAD/GLUE
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
John Hewitt
@johnhewtt
|
10. ruj |
|
How do we design probes that give us insight into a representation? In #emnlp2019 paper with @percyliang, our "control tasks" help us understand the capacity of a probe to make decisions unmotivated by the repr.
paper: arxiv.org/abs/1909.03368
blog: nlp.stanford.edu/~johnhew/inter… pic.twitter.com/1NA5hoyF7t
|
||
|
|
||
| Kevin Clark proslijedio/la je tweet | ||
|
Grzegorz Chrupała 🇪🇺
@gchrupala
|
1. kol |
|
#BlackboxNLP best paper award went to: What does BERT look at? An Analysis of BERT’s Attention. Kevin Clark, Urvashi Khandelwal, Omer Levy and Christopher D. Manning.
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
11. srp |
|
BAM! Our new #ACL2019 paper presents "Born-Again Multi-Task Networks," a simple way to improve multi-task learning using knowledge distillation. With @lmthang @ukhndlwl @quocleix @chrmanning. Paper: arxiv.org/pdf/1907.04829… Code: bit.ly/2NO8ufN
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
27. lip |
|
Code for our paper "What Does BERT Look At? An Analysis of BERT's Attention" (arxiv.org/abs/1906.04341) has been released! github.com/clarkkev/atten…
|
||
|
|
||
|
Kevin Clark
@clark_kev
|
12. lip |
|
Check out our new #BlackboxNLP paper "What Does BERT Look At? An Analysis of BERT's Attention" with @ukhndlwl @omerlevy @chrmanning! arxiv.org/abs/1906.04341 Among other things, we show that BERT's attention corresponds surprisingly well to aspects of syntax and coreference. pic.twitter.com/SWh1qMIKX1
|
||
|
|
||