HackerTrans
TopNewTrendsCommentsPastAskShowJobs

jumpCastle

no profile record

comments

jumpCastle
·anno scorso·discuss
It was a good benchmark until it entered the training set.
jumpCastle
·anno scorso·discuss
But open source like aider
jumpCastle
·anno scorso·discuss
Deepseek v3 was FP8
jumpCastle
·anno scorso·discuss
It's quite similar to muP

https://github.com/microsoft/mup
jumpCastle
·2 anni fa·discuss
Stock prices cannot go up without if the planet is destroyed
jumpCastle
·2 anni fa·discuss
Attention was invented because Bengio lab had to be disciplined about a black box (google had more compute)
jumpCastle
·2 anni fa·discuss
Aren't services like runpod solve half of these concerns?
jumpCastle
·2 anni fa·discuss
Are the updates really slow?