who's working on scaling laws for embedding models? could be any of:
• text embeddings (DPR, GTR, GTE…)
• image embeddings (SimCLR, DINO…)
• recommendation systems (?)
• multimodal embeddings (CLIP, ImageBind…)
• any other type of embeddings…
RESEARCH
-
Scaling Laws for Embedding Models: Research Initiatives
By
–
-
The AI Cambrian Explosion: New Species of Applications
By
–
The coming wave is a supercluster, an evolutionary burst like the Cambrian explosion, the most intense eruption of new species in the Earth’s history, with many thousands of potential new applications.
-
Calibrated LLMs must hallucinate; RLHF reduces calibration
By
–
“Calibrated Language Models Must Hallucinate” — Hallucination rate from a pre-trained LLM ≈ proportion of facts seen only once in training. RLHF reduces hallucination but makes LLMs less well calibrated as models. Real text has news; LLMs need to unlearn that fact to please us.
-
Deep Learning Forecasts Cryptocurrency Prices Using Integrated Data
By
–
Forecasting Cryptocurrency Prices Using Deep Learning: Integrating Financial, Blockchain, and Text Data Gurgul et al.: https://
arxiv.org/abs/2311.14759 #Cryptocurrency #DeepLearning #Trading -
DLT Deepnote Analysis Women’s Wellness Violence Trends
By
–
DLT & Deepnote in women's wellness and violence trends: A Visual Analysis https://
bit.ly/3sIg0yh
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
New Biomedical Simulation Method Seminar with Ellen Kuh
By
–
A new method for discovering biomedical models could change how we simulate biological systems forever. Join this week’s seminar with @WuTsaiAlliance researcher Ellen Kuh (@ln2cool) on Wednesday at 10 am PT. Register to attend online or in-person: https://
stanford.io/3uxfsvr -
MMCR: Self-Supervised Learning from Statistical Physics
By
–
MMCR: a new learning criterion for Self-Supervised training of joint embedding architectures based on concepts and results from statistical physics. By a team from Stanford, NYU, FAIR, and MIT.
-
Scale AI and NVIDIA Collaborate on HelpSteer LLM Dataset
By
–
.
@scale_AI worked with @nvidia on HelpSteer, a first-of-kind multi-attribute preference dataset across correctness, coherence, complexity & verbosity. We look forward to more research collaboration with @nvidia on LLMs! Check out the paper here: https://
arxiv.org/abs/2311.09528 -
Deep Learning Neural Networks Dismiss Critics Future Solution
By
–
Deep learning will clearly be part of the solution. To the dismay of some critics who've built their career around a futile, rear-guard, losing attempt to take down neural nets.