Three months ago, I didn’t know what a VLA was. I just wanted to contribute to physical AI safety, so learned a lot from scratch.
Thanks to @BlueDotImpact for funding my travel and registration. You supporting broke independent researchers is the best thing ever!
I scraped all 38,101 ICLR 2027 submissions and counted the names before the colon.
LLM has a naming crisis.
🏆 The leaderboard:
TRACE: 47 papers
PRISM: 42
PACE: 22
SCOPE: 21
SAGE: 17
MOSAIC: 16
COMPASS: 14
750 names are used by 2+ papers. That's 2,209 papers sharing a name.
Best parts:
• Two papers called TRACE both fix VLM hallucination. Same name, same problem, different authors.
• Four papers called PRISM do time series. Good luck, reviewers.
• "PACE: Discovering Options via Self-Distillation." None of P, A, C, E appear in the title. They picked the name first.
• 47 TRACEs cover quantum computing, materials discovery, seizure localization, carbon emissions, traffic control, and circuit design. TRACE is everywhere.
• The formula: 5 letters, looks like a real word, sounds cool. CAST. PACT. CRAFT. GRAFT. FORGE.
Tip for ICLR 2028: search OpenReview before you name your method 😂
Follow for more data-backed ML nonsense.
Anybody remember CoRL 2021 in London?
I was the Local Chair, and that year we introduced the demos/exhibition for the first time at CoRL.
It was just a small corridor with three tables. It felt very exciting at the time... 🤭
But fast forward to #CoRL2026, and life on the exhibition floor feels very different! So much to see, so much energy and buzz.
We’ve come a long way together at CoRL. Nicely done, everyone! 👏
@corl_conf
@faraz_r_khan@therealyawpower It is not actually. The harness gives the model a set of tools to call, so the LLM never gets to the joint angles directly.