Top ML Papers of the Week (Oct 2 – Oct 8): – StreamingLLM – Analogical Prompting
– The Dawn of LMMs
– Neural Developmental Programs
– LLMs Represent Space and Time
– Retrieval meets Long Context LLMs
… —- 1/ LLMs Represent Space and Time – discovers that LLMs learn linear
LLMS
-
Top ML Papers: LLMs, Streaming, and Multimodal Advances
By
–
-
LLM, Image Classification, Dreambooth, Tabular Tasks Coming Soon
By
–
llm
text classification
dreambooth
image classification
all kinds of tabular tasks
… more soon -
Tensor Puzzles: Master PyTorch Through Problem Solving
By
–
srush/Tensor-Puzzles: Solve puzzles. Improve your pytorch. https://
bit.ly/3sR168B #AI #MachineLearning #DeepLearning #LLMs #DataScience -
Try LLaVA chatbot, free open source alternative to GPT-4V
By
–
Don’t have access to GPT-4V yet? Try LLaVA chatbot, a free open source model hosted on HuggingFace: http://
llava.hliu.cc -
AI Model Thinking Pause: Reasoning Before Response Generation
By
–
That pause to think about the answer at the beginning
-
2023 Breakthrough Year for AI Software Innovation and Research
By
–
There’s never been anything like 2023 in software The rate + quality of new products / research in OSS, startups, bigtech, and on arXiv has been breathtaking + inspiring. Playing w/ so many different LLMs still gives me daily goosebumps. And we’re just getting started.
-
API Limits Will Disappear as GPU Capacity Scales
By
–
API usage limits are an artifact of a world without enough GPU’s. As we keep scaling, this will be less and less of an issue. My guess is the situation will look wildly different in 12-18 months. We want people to build real products and scale to millions of users.
-
LLMs Achieve Self-Recursive Code Improvement Capabilities
By
–
This is rather profound Large language models shows signs of being capable of self recursively improving their code. This is both exciting and poses somewhat of a larger risk. What a time to be alive.
-
Factuality in LLMs: Benchmarks and Knowledge Assessment
By
–
Nice paper by Tu Vu on factuality in LLMs: http://
arxiv.org/abs/2310.03214, enjoyed contributing in a minor role to it while I was at Google. The main takeaway for me is that most factuality benchmarks for LLMs don't really take into account the fact that many types of knowledge -
Google Gemini Surpasses GPT-4 by 5X in Performance
By
–
Google Gemini Eats The World – Gemini Smashes GPT-4 By 5X, The GPU-Poors https://
bit.ly/3PcLUtK #AI #MachineLearning #DeepLearning #LLMs #DataScience