The False Promise of Imitating Proprietary LLMs (
https://
arxiv.org/abs/2305.15717) Crowd workers rate imitation models (models trained on other models' outputs) highly. But it turns out that these imitation models only tend to mimic the style of the upstream LLMs not their factuality.
@rasbt
-
Imitation Models Mimic Style, Not Factuality of LLMs
By
–
-
Full Finetuning Not Always Necessary for Large Language Models
By
–
I don't think full finetuning is necessary, except as a baseline perhaps. Even without PEFT, it often doesn't make sense to tune all layers: (from my article https://
magazine.sebastianraschka.com/p/finetuning-l
arge-language-models
…) -
Ligand Discovery Projects Confuse Active Molecules with Drugs
By
–
Haha good point. 99% of computational ligand discovery project conflate ‘active molecule’ with ‘pharmaceutical drug’
-
Common buzzwords in technology and business sectors
By
–
buzzword check: AI Data science Disrupt Big data
-
Curating Interesting and Useful AI Research Papers Monthly
By
–
yeah, that sounds about right. I spend a lot of time each month combing through papers that are interesting but also useful and probably worthwhile E.g., "AI Research Highlights In 3 Sentences Or Less (May-June 2023)" https://
magazine.sebastianraschka.com/p/ai-research-
highlights-in-3-sentences-2a1
… -
Improved ArXiv Integration with Peer-Review Indicators
By
–
What might be more feasible in the meantime is to implement a better integration with journals & conferences so that arxiv papers get a more prominent "peer-reviewed" indicator once the paper passes said reviews. (Right now it's on the authors to update)
-
Challenges of moderating large-scale arxiv machine learning submissions
By
–
Managing it at that scale sounds like a challenge though. I was a moderator for the machine learning category (cs.LG) on arxiv for a few years: we got like 100-300 paper submissions per day! Just checking whether the categorization was correct already took 1-2 hours each day.
-
CPU Performance Progress: Fast MLP Training on Modern MacBooks
By
–
I actually recently marveled at how far CPU performance has come (comparing my undergrad computer to a modern MacBook). Was creating a confidence interval demo with a simple sklearn MLP, and it fits 500 iters in like a second on a 2020 MacBook Air: https://
github.com/rasbt/MachineL
earning-QandAI-book/blob/main/supplementary/q25_confidence-intervals/1_four-methods.ipynb
… -
Using ArXiv Carefully: Value Despite Imperfections
By
–
So much discussion about arxiv this morning. You can get a lot of positive value out of using arxiv if you use it with care. Eg does the original Transformer paper have inconsistencies (see LayerNorm) and is not super detailed? Sure. Is it a groundbreaking paper. Yes!
-
Citation Context Matters: Different Outcomes Change Architecture Proposals
By
–
Depends on the cites. E.g.,
“xyz proposed a new architecture that successfully addresses …” is quite different from “xyz proposed a new architecture that failed to address …”