News for hackers. Stuff that matters.

prjkt

They're competing with NVIDIA (huggingface)
Re:Turning GLM-5.3-Flash into a Jev-like decision model on Saturday September 26, 2026 @11:04PM (#49862973)
how is Jev cheaper if I can run locally. 0.5% prefill, 0.1% decode, 99.4% cached, latency is <20ms
Re:Plan Mode Is Dead on Saturday September 26, 2026 @04:53PM (#49860441)
"This is discussion; no code changes yet."

vs

Press "Tab"

It's like Microsoft banning Vim users that use Azure
> If there's such a dramatic hole in demand, who are NVIDIA selling their compute to?

NVIDIA itself is also training foundation models (and open-sourcing them). If there is excess compute available, NVIDIA can increase the scale of such models.

In that case, what tokenizer version? What was the temperature set to? topk? topp? FP32? FP16? Quantized? Hopper? Blackwell?
What is stopping us from all using IRC again ?
does pytorch count
Re:DeepSeek proves the future of LLMs is open-source on Wednesday January 29, 2025 @11:26AM (#42867048)
Source:

- Training SW [x]

- Inference SW [x]

- Evaluation SW [x]

- Data [x]

Output:

- Weights []

DeepSeek is closed-source with *open-weights*

Re:Kimi K1.5: Scaling Reinforcement Learning with LLMs on Tuesday January 21, 2025 @10:02AM (#42780848)
These types of "repositories" should contain some kind of flag/indication that it contains no source code, similar to when a repo is archived