Hacker News new | ask | show | jobs
by captcanuk 829 days ago
"The implementation of the MoE layer in this repository is not efficient. The implementation was chosen to avoid the need for custom kernels to validate the correctness of the model."

Or perhaps release your actual code AND the simplified implementation instead of hiding it and saying "you don't know her, she goes to a different high school"

1 comments

Always love it when someone gives away a gift and it’s not enough for people.
Not just someone but the CEO of the company. He used HIS platform to say "This week, @xAI will open source Grok" (https://twitter.com/elonmusk/status/1767108624038449405) and they aren't doing that. What they delivered specifically says "We are releasing the base model weights and network architecture of Grok-1, our large language model."
Sounds like they did what they said they would.