shourya⚠️

@TensorTwerker hello omnipresent operational noise of digital infrastructure

who goes there? must be the datacenters' hum

@karishgrover @iclr_conf @NeurIPSConf it's pretty funny ngl

Am I reading this right? Is the @iclr_conf'27 submission deadline before the @NeurIPSConf'26 decision notification? Can someone confirm this? #ICLR2027 #NeurIPS2026

@Vexxter4sure hooooly

kimi k3 from scratch implemented in jax , I am half drunk and half asleep benchmarks to be updated post full GPU run hopefully by tomorrow github.com/Vexxter/kimi-k…

in the cluster, no one computes alone

@johnsmithyson0 no i'm spreading misinformation

@flaczip Really?

opus 5 weights have leaked lmao

@xwang_lk @NeurIPSConf real

Why require initial meta reviews before rebuttal, @NeurIPSConf? ACs don’t like it: They need the rebuttal and reviewer discussion before making an informed assessment. Reviewers don’t like it: An early AC recommendation can anchor the discussion and make opinions harder to update. Authors don’t like it: An early meta review risks anchoring the final judgment before the rebuttal is even considered. If the real decision is made after rebuttals and discussion, why require a preliminary meta review at all?

@cyb3r_17 yes openreview meta-review on top

@flaczip openreview?

@cyb3r_17 no email yet

@flaczip wait did you get the email?

@literallylatted ⚠️

@flaczip Win.

neurips reviewers did not raise my cortisol

@delujog room for improvement

@flaczip Went well?

@Eggs_and_Jakey @cyanopsis first idea in the history of ideas. i like how you think

@flaczip @cyanopsis lets make autonomous lab equipment and we'll sell the consumables and lock labs into our environment end to end

time to shine

@cyanopsis LMAOOO

@flaczip openai.com/careers/

@tweetweettweets will post those next

@delujog cannot refute

@flaczip Greatest city in the world, baby!

@artrockalter professional geoguessr