From what I have heard, getting license from them is also far from guaranteed. They are selective about who they want to do business with -- understandable, but something to keep in mind.
That would be misleading. They aren't open weight (3B is not available). They aren't compared to Qwen 2.5 which beats them in many of the benchmarks presented while having more permissive license. The closed 3B is not competitive with other API only models, like Gemini Flash 8B which costs less and has better performance.
Also, the 3B model, which is API only (so the only thing that matters is price, quality and speed) should be compared to something like Gemini Flash 1.5 8B which is cheaper than this 3B API and also has higher benchmark performance, super long context support, etc.
- Ability to self host. This unlocks few things: (1) Customized serving stack with various logit processors, etc. (2) More cost efficient inference.
- Ability to fine tune. Most stock instruct models are quite lame at AI story-writing and role-play and produce slop.
There aren't really any pain points specific to Llama, but if we are creating a wish list:
- Keep the pre-training data diverse. There is a worrying trend where some companies apply heavy handed filtering on the pre-training data that's not just based on quality, but also on content. Quality based filtering is understandable and desirable, but please, keep the pre-training dataset diverse :)
- Efficient inference. Open source is way behind closed source here. TensorRT-LLM is probably the most efficient from what's out there, but it's mostly closed source. Maybe Meta could contribute to some of the open source projects like vLLM (or maybe something lower level...).
- A lot of the improvements we saw recently came from post-training, post-SFT improvements. And it's not just the datasets (which clearly you can't just release), but also algorithms -- and most labs are quite secretive about the details here. The open-source community relies on DPO a lot (and more recently, KTO), since it's easy, but empirically it's not that great.
What's your source on this? They just very recently reached 100K downloads on Android and according to various SEO tools, they get maybe ~4M visitsper-month (and these tend to overestimate, plus it's monthly visits, not DAU).
I am not seeing any race-to-zero in the hosted offering space. Most charge multiples of what you would pay on GCP, and the public prices on GCP are already several times what you would pay as an enterprise customer.
Great, more competition for the price-gouging platforms like Replicate and Modal is needed. As always with these, I would be curious about the cold-start time -- are you doing anything smart about being able to start (load models into VRAM) quickly? Most platforms that I tested are completely naive in their implementation, often downloading the docker image just-in-time instead of having it ready to be deployed on multiple machines.
Curious how this is on the front-page, despite falling down to the second page for a while, and having so many more comments than upvotes (which usually results in demotion of the story).
It's not so clear cut ;) "Research at CERN in Switzerland by the British computer scientist Tim Berners-Lee in 1989–90 resulted in the World Wide Web, linking hypertext documents into an information system, accessible from any node on the network."
When talking about memory requirements one also needs to mention the sequence length. In case of Mixtral, which supports 32000 tokens, this can be a significant chunk of the memory used.
This looks neat. Looking at the video, I would consider putting the "Render component" button at the bottom, so you don't have to scroll back and forth.