» Tag
machine-learning
185 postsPrevious-Token Prediction Based LLM Near-Exact Prompt Reconstruction
A new approach for prompt reconstruction in LLMs using previous-token prediction is introduced.
AI agents can modify themselves without human intervention
AI agents possess the ability to modify themselves autonomously, raising new security concerns for enterprises.
A Bitter Lesson for Data Filtering
Exploring the necessity of data filtering in large model pretraining. Low-quality data may provide unexpected benefits.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comInfinite-Parameter LLMs: Generating and Adapting Weights from Live Data
The Infinite-Parameter LLM enables language models to learn from live data, enhancing adaptability and performance.
Privacy Leakage in Split-LLM Training: A Security Case Study
A study on split-LLM training reveals privacy vulnerabilities in security checks.
When LLMs Outperform Statistical Models in Insurance Pricing
Research highlights how LLMs can outperform traditional models in insurance pricing under limited data conditions.
One-Way Networks as a Solution to AI Security Risks
Exploring one-way network architectures and data diodes for AI security.
Faiss, Turbovec, and Infino: A Comparison of 4-bit Vector Quantization
Exploring the performance of FAISS, turbovec, and Infino in 4-bit vector quantization.
GenRec: An LLM-Backed Recommendation Ranker at Netflix
Netflix's GenRec leverages LLMs to enhance recommendation systems, achieving significant performance improvements with fewer training examples.
Raggio: Open-source Hybrid Vector Database Based on TurboQuant
Raggio is an open-source hybrid vector database based on TurboQuant, optimized for performance in resource-constrained environments.