Keno Harada

国立国会図書館は、令和3年度にオープンソースのOCR（光学式文字認識）処理プログラム #NDLOCR を開発し、GitHubでソースコードを公開しています。商用・非商用を問わず自由な利用が可能です。 NDLOCRについては国立国会図書館月報2022年11月号の特集記事でも紹介しています。 dl.ndl.go.jp/view/download/

The following media includes potentially sensitive content. Change settings

View

1,287

3,380

405.8K

Show this thread

Keno Harada Retweeted

Alfredo Canziani

@alfcnz

Jan 1

Let's start the year by brushing up on the basics of neural nets: linear and non-linear transformations. In this episode, we're concerned with inference only. Forward and backwards. We introduce the cost and the energy.

Website: atcold.github.io/NYU-DLFL22/ youtu.be/QwZQrxIk6Dg

The following media includes potentially sensitive content. Change settings

View

NYU Courant and NYU Data Science

210

1,352

166.3K

Keno Harada Retweeted

べいえりあ

@mr_bay_area

Jan 3

HarvardのCS197の講義ノートが公開されている。医療AIなどで有名なRajpurkar氏がAIの研究やる上で役に立ったこと全部盛りな内容で、gitやVSCodeの使い方から実験管理、論文の読み方やアイデアの出し方、はたまた最近の生成モデルの話やfine tuningまで、めちゃ盛り沢山。

cs197.seas.harvard.edu

Harvard CS197: AI Research Experiences

Learn to do applied deep learning research. This course will teach you the fundamental skills involved in conducting cutting-edge AI research and writing a research paper.

412

36.4K

Keno Harada Retweeted

Yann LeCun

@ylecun

Dec 26, 2022

MoDem: accelerated Model-Based RL. The world model is a JEPA-like architecture, which makes predictions in representation space (without a decoder). From FAIR+UCSD

Quote Tweet

Aravind Rajeswaran

@aravindr93

Dec 26, 2022

Excited to share MoDem -- our latest work on visual world models for control!

MoDem combines decoder-free world models, demonstrations, and pre-trained vision encoders. Achieves SOTA results in 21 challenging visuo-motor control tasks.

Web: nicklashansen.github.io/modemrl/

Show this thread

The media could not be played.

Reload

144

64.7K

Keno Harada Retweeted

Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)

@rao2z

Dec 17, 2022

I finally got around to making a public website for my Fall 2022 Planning & Learning class--that focused on bridging planning & #ReinforcementLearning (and a bit of control theory). Slides, videos, scribed notes--the whole shebang..

rakaposhi.eas.asu.edu/cse574

The following media includes potentially sensitive content. Change settings

View

The following media includes potentially sensitive content. Change settings

View

12.3K

Show this thread

Keno Harada Retweeted

Graham Neubig

@gneubig

Dec 15, 2022

CMU Advanced NLP is done for 2022! Check the videos on YouTube

I also rehauled our assignments to reflect important skills in NLP for 2022: github.com/neubig/nlp-fro If you're teaching/learning NLP see the

and doc for more!

Quote Tweet

Graham Neubig

@gneubig

Sep 12, 2022

We've started the Fall 2022 edition of:

CMU CS11-711 Advanced NLP!

Follow along for * An intro of core topics * Timely content; prompting, retrieval, bias/fairness * Content on NLP research methodology Page: phontron.com/class/anlp2022/ Videos: youtube.com/playlist?list=

107

442

52.9K

Show this thread

Keno Harada Retweeted

@_akhaliq

Nov 29, 2022

Is Conditional Generative Modeling all you need for Decision-Making? abs: arxiv.org/abs/2211.15657 project page: anuragajay.github.io/decision-diffu

The following media includes potentially sensitive content. Change settings

View

316

Keno Harada Retweeted

Aran Komatsuzaki

@arankomatsuzaki

Nov 24, 2022

Masked Autoencoding for Scalable and Generalizable Decision Making Presents MaskDP, a self-supervised pretraining method for RL that outperforms GPT-like approaches. arxiv.org/abs/2211.12740

The following media includes potentially sensitive content. Change settings

View

117

Keno Harada Retweeted

hardmaru

@hardmaru

Nov 24, 2022

Excited to announce the release of Stable Diffusion 2.0! Many new features in v2: • Base 512x512 and 768x768 models trained from scratch with new OpenCLIP text encoder • X4 upscaling text-guided diffusion model • New “Depth2Image” functionality Blog: stability.ai/blog/stable-di

The following media includes potentially sensitive content. Change settings

View

Stability AI and Stable Diffusion

608

2,799

Show this thread

Keno Harada Retweeted

Andrej Karpathy

@karpathy

Nov 18, 2022

An interesting historical note is that neural language models have actually been around for a very long time but noone really cared anywhere near today's extent. LMs were thought of as specific applications, not as mainline research unlocking new general AI paths and capabilities

196

1,269

Show this thread

Keno Harada Retweeted

Sanchit Gandhi

@sanchitgandhi99

Nov 3, 2022

Whisper fine-tuning is **here**! Check out the blog post for a step-by-step guide on fine-tuning Whisper with

Transformers: huggingface.co/blog/fine-tune Boost WER performance vs zero-shot with as little as 8h of training data

All on a single Google Colab notebook...

huggingface.co

Fine-Tune Whisper For Multilingual ASR with

Transformers

155

680

Show this thread

Keno Harada Retweeted

Jacky Liang

@jackyliang42

Nov 2, 2022

How can robots perform a wide variety of novel tasks from natural language? Execited to present Code as Policies - using language models to directly write robot policy code from language instructions. See paper, colabs, blog, and demos at code-as-policies.github.io long

The following media includes potentially sensitive content. Change settings

View

181

611

Show this thread

Keno Harada Retweeted

Fei-Fei Li

@drfeifei

Oct 22, 2022

Do you want to learn to train and evaluate embodied AI solutions for 1000 household tasks in a realistic simulator? Join our BEHAVIOR Tutorial at #ECCV2022: Benchmarking Embodied AI Solutions in Natural Tasks! Time: Monday, Oct 24th 14:00 local time (4:00 Pacific Time)

The following media includes potentially sensitive content. Change settings

View

432

Show this thread

Keno Harada Retweeted

Jeremy Howard

@jeremyphoward

Oct 20, 2022

I got a special surprise for you all... We just released the first 5.5 hours of our new course "From Deep Learning Foundations to Stable Diffusion", for free!

fast.ai

fast.ai - 1st Two Lessons of From Deep Learning Foundations to Stable Diffusion

4 videos from Practical Deep Learning for Coders Part 2, 2022 have been released as a special early preview of the new course.

664

3,149

Show this thread

Keno Harada Retweeted

Sergey Levine

@svlevine

Oct 18, 2022

Large pre-trained models (ViT, BERT, GPT, etc.) are a starting point for many downstream tasks. What would large pre-trained models look like in robotics, and how do we build them? I discuss this in my new article: sergeylevine.substack.com/p/general-purp Short summary:

sergeylevine.substack.com

General-Purpose Pre-Trained Models in Robotics

Can we (pre-) train policies to control any robot for any task?

278

Show this thread

Keno Harada Retweeted

Sergey Levine

@svlevine

Oct 14, 2022

Pretraining on diverse robot data with offline RL allows planning multi-stage tasks, and then finetuning those tasks with end-to-end exploration. Our new paper, FLAP, describes a method for doing this with real robots! arxiv.org/abs/2210.06601 sites.google.com/view/project-f Thread:

The following media includes potentially sensitive content. Change settings

View