651–660 of 952 posts
Grim prediction: "Quantized models are no longer permitted on the Hugging Face platform. Reduced precision models are a safety hazard and a violation of our TOS. Click here to speak with a sales representative about our many exciting cloud hosting offerings or enterprise GPU packages."
Unlikely, but say goodbye to uncensored and abliterated models. These acquisitions are not healthy. They stifle competition and make it easier for governments to censor open weight models by choking off distribution points.
Earlier quoted context omitted.
> Not too different from what "the cloud" did to this market from 2015 or so... I mean for all the inefficiencies of the cloud, it solved a lot of problems. It also created a new category of problems but rarely does any particular technology arrive as a pure boon without it's own compromises. Despite it's issues and despite how hot ASS some implementations are (looking at you Azure) I think cloud infrastructure you c…
The purpose of the system is what it does. The goal is to increase return on capital. Take a look at the historical developments of wage participation in global income, labor productivity growth vs wage growth, and so forth. It works very well. As long as the most capable laborers are paid well enough to go along, the rest of us have little choice. Especially after citizens united basically privatized American electi…
Grim prediction: "Quantized models are no longer permitted on the Hugging Face platform. Reduced precision models are a safety hazard and a violation of our TOS. Click here to speak with a sales representative about our many exciting cloud hosting offerings or enterprise GPU packages."
Don't see this happening. Both Nemo and TensorRT from Nvidia are heavily invested in low precision formats and higher inference throughput as well as low memory use. It's more about what happens in relation to GGUF, MLX and non CUDA-native frameworks/formats.
Grim prediction: "Quantized models are no longer permitted on the Hugging Face platform. Reduced precision models are a safety hazard and a violation of our TOS. Click here to speak with a sales representative about our many exciting cloud hosting offerings or enterprise GPU packages."
Too base cynicism. I'd be willing to bet you my $200 to your $100 that doesn't happen.
Earlier quoted context omitted.
Dude, the source of an LLM is the training data and the training algorithm. Those are not open, so stop calling it open source. Open weights correspond to "binary available" for software. Nobody would call that "open source". (Even ignoring the licensing which "open source" normally entails.)
Many open weight models do actually have open training data and training algorithms -- I think especially some from Nvidia!
Though we agree that the term isn't typically used that way in the LLM context, right?
um um um um do we want nvidia to be on the same network as hugging face given the recent incidents?
could you elaborate on what happened?
Earlier quoted context omitted.
"lab" vs "company as a whole" They might be right w.r.t. openness about LLM at the moment, but w.r.t. general software openness they are definitely the opposite of open.
The company as a whole bought hugginface and will set its policy, though, not the lab.
This trend to vertical integration is very dangerous. We should not allow the people making tech to control the whole pipeline unless we want a permanent underclass.