[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

![fly51fly Avatar](https://lunarcrush.com/gi/w:24/cr:twitter::22146921.png) fly51fly [@fly51fly](/creator/twitter/fly51fly) on x 7553 followers
Created: 2025-07-10 21:50:23 UTC

[LG] Small Batch Size Training for Language Models: When Vanilla SGD Works, and Why Gradient Accumulation Is Wasteful
M Marek, S Lotfi, A Somasundaram, A G Wilson... [New York University & Columbia University] (2025)


![](https://pbs.twimg.com/media/GvhxEoEXQAAXc16.jpg)

XXX engagements

![Engagements Line Chart](https://lunarcrush.com/gi/w:600/p:tweet::1943427608500166951/c:line.svg)

**Related Topics**
[columbia](/topic/columbia)
[accumulation](/topic/accumulation)
[singapore dollar](/topic/singapore-dollar)
[$003550ks](/topic/$003550ks)
[lg](/topic/lg)

[Post Link](https://x.com/fly51fly/status/1943427608500166951)