@Noah64165746@NielsRogge No he has no point.
Jensen is basically saying open source and proprietary models can coexist. Julian is saying Nvidia should open source their stack to prove their support for open source!! A straw man argument because Jensen doesn’t claim that all models should be open source
@JFPuget@acaste10 I watched the game live sir and nothing like you claim existed.
May be you need to be more graceful in defeat and take accountability just like your team captain did: not throwing blame to deflect your own failures.
@JFPuget@acaste10 Come on! France was just outclassed: the referee had nothing to do with your loss. If anything you lot survived two red card offenses for bad tackles.
@bolutifeawakan@codemathics@PaulDamalie False marketing/advertising is why they are getting called out mostly.
They needed to be more precise with their messaging.
@mrstephendeng The moat may come down to data quality.
A lot of the datasets on African languages tend to be grammatically noisy (more grammatical errors than correctness), even those where data is collected from volunteer human contributors.
@michael_kove@mark_k@AIatMeta I don’t think LLMs are limiting, they are rather just one aspect of what’s needed for AGI.
Language is a means through which we communicate what we perceive and much as it’s not needed at the start of human cognitive function, it’s vital for children in their formative years.
""Good engineering management" is a fad".
For each role, you have the chance to prioritize across different dimensions like pace, people, prestige, profit, or learning. There’s no “right decision,” and there are always tradeoffs.
by @Lethain
https://t.co/qmbEVsJHe8
@allen_ai Is there any reason why you define an <eos> token but use only the <bos> token for both begging and end of sequence?
See here https://t.co/HfSs8622Xd
🔉 Introducing SAM Audio, the first unified model that isolates any sound from complex audio mixtures using text, visual, or span prompts.
We’re sharing SAM Audio with the community, along with a perception encoder model, benchmarks and research papers, to empower others to explore new forms of expression and build applications that were previously out of reach.
🔗 Learn more: https://t.co/FPnfv66UCP
Today, @NVIDIA is launching the open Nemotron 3 model family, starting with Nano (30B-3A), which pushes the frontier of accuracy and inference efficiency with a novel hybrid SSM Mixture of Experts architecture. Super and Ultra are coming in the next few months.