Kshitij Sachan

beep boop at @AnthropicAI

[Kshitij Sachan @SachanKshitij]

love claudius 🥹

[Alex Albert @alexalbert__]

Claude is hyped to hear that its small business is getting the public recognition it deserves

[Kshitij Sachan @SachanKshitij]

@akhil_bagaria @BrownCSDept @BrownBigAI @BrownUniversity Massive congrats Dr. Akhil!!! Thanks for all the mentorship :)

[Kshitij Sachan @SachanKshitij]

@jackclarkSF Just got to london! Missed it by a day :)

[Anthropic @AnthropicAI]

Join us from 9:30am PT tomorrow: https://t.co/3lSlYKqQCT.

[Kshitij Sachan @SachanKshitij]

@srush_nlp Why not Jax 🥺🥺

[Kshitij Sachan @SachanKshitij]

@ronan_fitz_ Smh

[Kshitij Sachan @SachanKshitij]

@srush_nlp Double descent! Due to early stopping, transformers are massively overparametrized. No bias variance tradeoff in this regime - sgd seems to have an inductive bias for “simple” solutions (literally min norm in case of linear regression)

[Jack Clark @jackclarkSF]

Here's a letter we sent to Governor Newsom about SB 1047. This isn't an endorsement but rather a view of the costs and benefits of the bill. https://t.co/HhOTsdk7ui

[Kshitij Sachan @SachanKshitij]

@akhil_bagaria @BrownCSDept @RichardSSutton @mlittmancs @BrownBigAI Wow congrats akhil! I had a lot of fun working with you :)

[Orowa Sikder @OrowaSikder]

Enjoying Claude Artifacts? Want to build the next generation of Human-AI Interfaces?

We're hiring an ML Lead for Artifacts.

This is a unique full-stack role where you'll help co-develop new user interfaces along with the model capabilities which support them. You'll work as closely with our Finetuning teams as our UX and design teams.

Strong front-end/design intuitions + decent ML Engineering experience. Come build:

https://t.co/5AnW1hC3jN

[Kshitij Sachan @SachanKshitij]

I’m going to be honest, I had no idea what this meant

[Anthropic @AnthropicAI]

Fc zbvx ts temxnsq nx mzog jlbuv gusn zofg hhfs:
Ebwxk vnii mzceaw tfr fvpowf sbyglovaw fmr,
Nyp fgrryw xjf, lrx 'mosra xaw huvvw sbq ssnjhu'f vnf,
Che rxo qeremaca ophgaf, n abkse oyw.

KEY:

[Will Cathcart @wcathcart]

Many have said this already, but worth repeating: this is not correct.

We take security seriously and that's why we end-to-end encrypt your messages. They don't get sent to us every night or exported to us.

If you do want to backup your messages, you can use your cloud provider and you can even use end-to-end encryption for that too. Turn it on here: https://t.co/dxDkM9vn3H

[Logan Graham @logangraham]

I’m hiring ambitious Research Scientists at @AnthropicAI to measure and prepare for models acting autonomously in the world. This is one of the most novel and difficult capabilities to measure, and critical for safety.

Join the Frontier Red Team at Anthropic: https://t.co/5WT7xprWhA

[Jason D. Clinton 🔸 @JasonDClinton]

https://t.co/SjWOAc6aBY

[Jesse Mu @jayelmnop]

We’re hiring for the adversarial robustness team @AnthropicAI!

As an Alignment subteam, we're making a big effort on red-teaming, test-time monitoring, and adversarial training. If you’re interested in these areas, let us know! (emails in 🧵)

[Kshitij Sachan @SachanKshitij]

@tensecorrection Any examples you can share?

[Kshitij Sachan @SachanKshitij]

@lxrjl Not available

[Emmanuel Ameisen @mlpowered]

Claude 3 Opus is great at following multiple complex instructions.

To test it, @ErikSchluntz and I had it take on @karpathy's challenge to transform his 2h13m tokenizer video into a blog post, in ONE prompt, and it just... did it

Here are some details:

[Kshitij Sachan @SachanKshitij]

@DavidSKrueger @JeffLadish Some LLM problems are like software vulnerabilities - it’s pretty simple to patch a specific class of jailbreak

[akbir. @akbirkhan]

How can we check LLM outputs in domains where we are not experts?

We find that non-expert humans answer questions better after reading debates between expert LLMs.
Moreover, human judges are more accurate as experts get more persuasive. 📈
https://t.co/jgyfCEQvfw