@swyx

–

10 July 2025 22h48

train LLM on all AK tweet summaries of papers @_akhaliq @arankomatsuzaki and other AKs

Merged Experts in MoE: Coordination Without Joint Training

By

–

10 July 2025 19h11

all experts are merged into an MoE. The anchor model ensures that all experts learn to coordinate, even though they are never trained on the joint dataset. this is rocket science wtf

Updated dev cycle with chart on Moloch hyperstitions

By

–

10 July 2025 19h07

updated the dev cycle with chart https://
latent.space/p/hyperstition
s-of-moloch
…

Community Notes Needed for LLM Benchmark Misleading Claims

By

–

10 July 2025 18h18

we need community notes for llm benchmark porn

Daily AI News Analysis with Grok versus Gemini Comparison

By

–

10 July 2025 5h07

i run ainews thru grok every day. its good but not gemini

Gym Bros and LLMs: Training on Test Data Reputation Issues

By

–

10 July 2025 5h01

gym bros llms that need to restore their reputation training on test

Compressing Knowledge: Can Small LLMs Match o3 Pro Capability?

By

–

10 July 2025 0h22

more on this idea in our pod with @jxmnop

but his favorite part is highlighted here: the "cognitive core" question of how smol we can compress knowledge and LLMs so that they have o3 pro level capability with only x00m parameters?https://t.co/4Skn4AKlgu pic.twitter.com/dFhlToWKC9
— swyx 🐣 (@swyx) 9 juillet 2025

more on this idea in our pod with @jxmnop but his favorite part is highlighted here: the "cognitive core" question of how smol we can compress knowledge and LLMs so that they have o3 pro level capability with only x00m parameters?

Custom RL as Service: The New $500M Startup Opportunity

By

–

10 July 2025 0h19

i think this is comparatively slept on the Current Thing if you depart from OAI (eg TML?) you are doing "we are extremely cracked and we will do custom RL for you!" as a service. 3 dudes in a basement immediately worth $500m. so it's notable when a former OAI RL person says https://
x.com/_kevinlu/statu
/_kevinlu/status/1942977315031687460
…

Claude 4.1 and 4.2 in Development, Llama 5 Market Prediction

By

–

09 July 2025 22h10

apparently 4.1 and 4.2 are in the works still! yay! https://
x.com/ldjconfirmed/s
tatus/1943030722329219484
… but my polymarket is on llama 5 vs rebrand