Reflexion: Language Agents with Verbal Reinforcement Learning Reflexion lets LLMs learn without updating its weights. Its actor reflects on each agent decision, using citations and other targeted feedback to generate actionable critique. Reflections are stored in chat history
RESEARCH
-
Eigenfaces for AI Interpretability: Understanding Computer Vision
By
–
dunno, just seems pretty cool. i guess, interpretability? i mean, what do people use eigenfaces for?
-
Sequence Text Embeddings vs Word Embeddings Explained
By
–
i'm talking about sequence text embeddings not word embeddings
-
Sequence-Based Models Capture High-Level Features
By
–
don't think so! these are based on sequences so they should capture much higher-level features
-
Google Partners with Reddit for AI Model Training
By
–
Reddit and Google are partnering up to train AI models. Last week, we found out Reddit signed a $60M deal granting access to its user content for AI model training to an unknown company. Today, we found out that it was Google. Some thoughts: 1. Social media platforms are
-
HyDE Query Translation Technique for RAG Systems
By
–
RAG From Scratch: Query Translation (HyDE) Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code. This is our final video on Query Translation, focused on Hypothetical Document Embeddings (HyDE) from Gao et al. Problem:
-
Gemini’s 750K Token Processing Unlocks Major AI Capabilities
By
–
Watching Gemini ingest 750,000+ tokens in ~1 minute and then spit out near-perfect details (on things I wouldn't have even noticed) was a real tipping point for me. SO many capabilities just unlocked. It's going to be a wild year.
-
Recreating Articles Using Eigentext Vector Representations
By
–
the obvious next step from there is to recreate specific articles based on their eigentext vectors – essentially this part
-
Eigenfaces for Text: PCA on Domain-Specific Embeddings
By
–
project request: eigenfaces, but for text do PCA on text embeddings from a particular domain and find the "eigentext" embeddings that represent that domain then project back to text with vec2text. what would the eigentexts look like? surprised no one has done this yet
-
TheProfessor V1: Abacus Introduces LLM Merging Technology
By
–
Introducing TheProfessor (V1) From Abacus, an LLM cocktail! We’ve recently talked about a fascinating property of LLMs, merging, where you can combine different layers from LLMs without needing pre-training. The merged LLM will then show combined properties of the different