News for hackers. Stuff that matters.
Re:DeepSeek proves the future of LLMs is open-source by TeMPOraL on Wednesday January 29, 2025 @05:07PM (#42871858)
This argument only barely holds water for those big SOTA models like llama derivatives, and that's only because of practical costs involved.

Or should I say, it held water until few days ago.

Personally though, I never bought it. Saying that weights are the "preferred form of the work for making modifications to it" because a) approximately no one can afford to start with the training data, and b) fine-tuning and training LoRAs are cheap enough, is basically like saying binary blobs are "open source" as long as they provide an API (or ABI) for other programs to use. By this line of reasoning, NVIDIA GPU stack and Broadcom chipset firmware would qualify as open source, too.

Re:DeepSeek proves the future of LLMs is open-source by sitkack on Wednesday January 29, 2025 @06:56PM (#42872988)
If we are able to look back at this comment in a year or two, you will chuckle.
Re:DeepSeek proves the future of LLMs is open-source by furyofantares on Wednesday January 29, 2025 @09:52PM (#42874276)
I just don't think analogies to open source are useful in any direction. This is its own beast and we should just think about what we want out of it.