DeepSeek Open Source Optimized Parallelism Strategies, 3 repos(github.com)
github.com
DeepSeek Open Source Optimized Parallelism Strategies, 3 repos
https://github.com/deepseek-ai/profile-data
5 comments
3 repos
DualPipe
https://github.com/deepseek-ai/DualPipe
EPLB
https://github.com/deepseek-ai/eplb
profile-data
https://github.com/deepseek-ai/profile-data
It appears that they have really good toolings so they can focus on hiring really smart new grads and giving them a lot of leeway on choosing what to work on without worrying about how productive they will be. One particular memorable response when the CEO Liang was asked about not using KPI goes roughly like this: since we all reviewd and worked with the candidate, if he is not productive it must be our fault to not have given him the right support to make him productive.
Hillarous that Liang seem to claim they do some classic give engineers freedom and agency. Kinda like Google did initially?
KPIs and metrics in general are trashing productivity and moral when used even in the most slightly evaluating manner of personal performance.
KPIs and metrics in general are trashing productivity and moral when used even in the most slightly evaluating manner of personal performance.
It's probably more about the size of the firm. Once a company is large enough with middle layers and silos it seems some form of KPI is inevitable. This could also explain why Deepseek doesn't mind open sourcing as they have not raised outside capital for expansion so they can't really scale rapidly on their own.
and they dropped the pricing on top of that.. while being a better model compared to openai.. basically they are eating openai's lunch in every way imaginable.
Naw I’m not using deepseek regardless what you say. They are def not beating OpenAI from my uses
Oh, DeepOpenAI
What's the most interesting of the 3 repos? That would probably be better to post, for reasons explained here: https://news.ycombinator.com/item?id=43174782
(you can already see from the first few comments in this thread how generic the discussion tends to get in cases like this)
(you can already see from the first few comments in this thread how generic the discussion tends to get in cases like this)