Jinfeng Rao

78 posts

Jinfeng Rao

Jinfeng Rao

@Jeffy_Sailing

TLM, Researcher at Google, work on LLM & NLP

San Francisco, CA Katılım Eylül 2012
598 Takip Edilen499 Takipçiler
Jinfeng Rao retweetledi
Tao(Thomas) Yu
Tao(Thomas) Yu@tao_agi·
🚀 Breaking news - Yi-Large LMSys Elo result has just surged to the top tier, almost on par with GPT-4-0125-preview! With improvement across all boards, especially reasoning & coding capabilities, we're excited to see what app can build on top of Yi-Large. Explore the API on platform.01.ai Huge congrats to my team from 01.AI for this incredible milestone! 🎉
Arena.ai@arena

Exciting leaderboard update🔥 We've added @01AI_Yi Yi-Large to Arena and collected 15K+ votes over the past week. Yi-Large's performance is super impressive, securing the #7 spot, almost on par with GPT-4-0125-preview! Huge congrats to 01.ai on this incredible achievement, as well as new Yi family launch! Additionally, we’ve added Zhipu AI’s GLM-4-0116 to the leaderboard at #15. Chinese LLMs are getting very competitive now!

English
8
9
31
11.1K
Yi Tay
Yi Tay@YiTayML·
glad that my perf rating this time sounds very much like a very famous architecture in deep learning.
English
10
1
63
38K
Jinfeng Rao retweetledi
AK
AK@_akhaliq·
ExT5: Towards Extreme Multi-Task Scaling for Transfer Learning abs: arxiv.org/abs/2111.10952 EXT5 outperforms strong T5 baselines on SuperGLUE, GEM, Rainbow, Closed-Book QA tasks, and several tasks outside of EXMIX
AK tweet media
English
0
34
132
0
Jinfeng Rao retweetledi
Aran Komatsuzaki
Aran Komatsuzaki@arankomatsuzaki·
Efficiently Modeling Long Sequences with Structured State Spaces Achieves dramatic performance improvement over the SotA on Long-range Arena and comparable ppl to vanilla Transformer on Wikitext-103 w/ 60x faster generation. openreview.net/pdf?id=uYLFoz1…
Aran Komatsuzaki tweet media
English
4
25
108
0
Jimmy Lin
Jimmy Lin@lintool·
If you're interested in dense retrieval, you'll want to check out this DPR replication effort led by @xueguang_ma arxiv.org/abs/2104.05740 tl;dr - BM25 is better than the original authors made it out to be, and free QA boost with better evidence fusion!
English
5
6
74
0
Jinfeng Rao retweetledi
Sebastian Ruder
Sebastian Ruder@seb_ruder·
ML and NLP Research Highlights of 2020 It's been inspiring to look back on all the exciting advances that happened despite such a tumultuous year. Here's a selection of my highlights. ruder.io/research-highl…
English
4
305
924
0
Jinfeng Rao retweetledi
Andrej Karpathy
Andrej Karpathy@karpathy·
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale openreview.net/forum?id=YicbF… v cool. Further steps towards deprecating ConvNets with Transformers. Loving the increasing convergence of Vision/NLP and the much more efficient/flexible class of architectures.
Andrej Karpathy tweet media
English
24
459
1.9K
0
Jinfeng Rao retweetledi
Paul Katsen
Paul Katsen@Katsen·
=GPT3()... the spreadsheet function to rule them all. Impressed with how well it pattern matches from a few examples. The same function looked up state populations, peoples' twitter usernames and employers, and did some math.
English
138
1.7K
8.7K
0
Yi Tay
Yi Tay@YiTayML·
Turned 30 today but completely forgot until my wife wished me Happy birthday. I guess the first thing I did was to check for ICML reviews when I woke up. 😂
English
4
0
33
0
Jinfeng Rao retweetledi
Nick Craswell
Nick Craswell@nick_craswell·
Announcing: Deep Learning Track at TREC-2019. Coordinators: Nick Craswell, Bhaskar Mitra, Emine Yilmaz, Daniel Campos. Large training data! Two datasets! Guidelines: microsoft.github.io/TREC-2019-Deep…
English
2
26
49
0
Jinfeng Rao retweetledi
Graham Neubig
Graham Neubig@gneubig·
Cross-lingual transfer is a powerful tool for low-resource NLP. But when you build a system for a new language (say Bengali), what language do you transfer from? Our #ACL2019 paper "Choosing Transfer Languages for Cross-lingual Learning" asks this: arxiv.org/abs/1905.12688 1/7
Graham Neubig tweet media
English
2
63
246
0