Achal Dave

Hi! I'm a member of the technical staff at Anthropic. Previously, I was a Senior Research Scientist at Toyota Research Institute, building generative models for text and videos, with a focus on improving training efficiency and dataset quality. Prior to this, I was an Applied Scientist at Amazon, and a Ph.D. student at Carnegie Mellon University advised by Prof. Deva Ramanan, working on open-world object detection and tracking.

Email: achal@anthropic.com, achalddave@gmail.com

News

[2023.10]: openlm is now even faster, thanks to our recent updates!
[2023.09]: Check out openlm, our fast, minimal open-source library for training language models. We've open-sourced 1B and 7B models trained with openlm.

Projects

openlm: Minimal, performant, open-source library for training medium-scale language models (1B-13B).
open_clip: Open-source reproduction of OpenAI's CLIP.
Video tools: Useful utilities for working with videos
Tutorial links: A collection of useful tutorials from the web on various machine learning topics.

Recent and selected work

pix2gestalt: Amodal Segmentation by Synthesizing Wholes

Ege Ozguroglu, Ruoshi Liu, Didac Suris Coll-Vinent, Dian Chen, Achal Dave, Pavel Tokmakov, Carl Vondrick

In submission

Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP) [ArXiV]

Alex Fang, Gabriel Ilharco, Mitchell Wortsman, Yuhao Wan, Vaishaal Shankar, Achal Dave, Ludwig Schmidt

ICML 2022

Opening up Open-World Tracking [ArXiV]

Yang Liu*, Idil Esen Zulfikar*, Jonathon Luiten*, Achal Dave*, Deva Ramanan, Bastian Leibe, Aljoša Ošep, Laura Leal-Taixé

CVPR 2022, Oral

Evaluating Large-Vocabulary Object Detectors: The Devil is in the Details [ArXiV]

Achal Dave, Piotr Dollár, Deva Ramanan, Alexander Kirillov, Ross Girshick

ArXiV

Detecting Invisible People [paper]

Tarasha Khurana, Achal Dave, Deva Ramanan

ICCV 2021

TAO: A Large-Scale Benchmark for Tracking Any Object [paper, website, code]

Achal Dave, Tarasha Khurana, Pavel Tokmakov, Cordelia Schmid, Deva Ramanan

ECCV 2020 (Spotlight)

Measuring Robustness to Natural Distribution Shifts in Image Classification [paper, website, code]

Rohan Taori, Achal Dave, Vaishaal Shankar, Nicholas Carlini, Ben Recht, Ludwig Schmidt

NeurIPS, 2020. Spotlight.

Do Image Classifiers Generalize Across Time? [paper, website]

Vaishaal Shankar*, Achal Dave*, Rebecca Roelofs, Deva Ramanan, Ben Recht, Ludwig Schmidt

* Equal contribution

ICCV, 2021

Most of my projects can be found on my Github.

Fun Stuff

PSYCHO: PerSonalitY CHaracterizatiOn of artificial intelligence [paper]

Achal Dave, Rohit Girdhar

SIGBOVIK 2018