Happening now. I am giving a talk at @NeurIPSConf AIM-FM workshop at 3:30 today. If you want to learn about foundation models other than LLMs come to East ballroom A/B.
RESEARCH
-
Is AI Better Than Average Human Performance?
By
–
is it better than the average human would otherwise do the task?
-
AI Machine Learning Improve eCommerce Product Performance Reports
By
–
Your eCommerce product performance reports are probably misleading you https://
bit.ly/3CUHdCp
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
Integrated Large Models Combined with Good Old Fashioned Engineering
By
–
New computer systems perspective on integrated (modular and hybrid) large models with good old fashioned engineering (GOFE). @UCBerkeley @Berkeley_EECS @Databricks https://
arxiv.org/abs/2412.05299 -
Grok-2-1212 excels in instruction following
By
–
Grok-2-1212 seems to be outperforming everything else in instruction following It is a newly released grok 2 model on xAI platform (api only)
-
LLMs and their limits for generating novel insights
By
–
No I mean it’s only then that an LLM could generate a “thought” like: > Wow, this obscure mathematician seems to have proven a theorem in 1922 credited to someone else in 1955! I bet no one has noticed this before! Thus LLMs don’t usually know *any* novel insights of that form
-
Argument: LLMs don’t truly ‘read’ or think
By
–
An LLM knows every work of Shakespeare but can’t say which it read first. In this material sense a model hasn’t read at all. To read is to think. Only at inference is there space for serendipitous inspiration, which is why LLMs have so little of it to show for all they’ve seen.
-
Language Models’ Internal Representations Converge Like Vision Models
By
–
Fascinating talk by @phillip_isola at @NeurIPSConf (
@unireps workshop)! In a recent paper, they compared the embeddings produced by vision and language models. The key finding? As language models improve, their internal representations of the world grow increasingly similar to -
Schmidhuber Plagiarism Accusation and Academic Integrity in AI
By
–
I would prefer not to keep doing this, but Schmidhuber made a post accusing me of plagiarism yesterday. There’s no academic court where I can be tried and found guilty/ not guilty once and for all, though the opinion of the NeurIPS organizers should be clear.
-
Clarifying ‘we need more good data’ for AI training
By
–
no i mean “we need more data” implicitly means “we need more good data” if you simply re-tokenized existing text, or took photos of it on printed out pages in various environments, or projected it on walls and took videos, all of these would not work for similar reasons