HackerTrans
TopNewTrendsCommentsPastAskShowJobs

jumpCastle

no profile record

comments

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

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