llama.ttf combines a font file with a large language model (LLM) and an inference engine, enabling text generation within any application that uses the Harfbuzz shaping engine.
2. HybridNeRF: Efficient Neural Rendering
HybridNeRF introduces an innovative approach to neural rendering, leveraging adaptive volumetric surfaces for enhanced efficiency.
3. Delving into ChatGPT usage in academic writing through excess vocabulary
Analysis of 14 million PubMed abstracts from 2010-2024 reveals an abrupt increase in certain style words post-LLM introduction, suggesting at least 10% of 2024 abstracts utilized LLMs like ChatGPT for academic writing.
4. ChatGPT is biased against resumes with credentials that imply a disability
ChatGPT exhibits bias against resumes featuring disability-related honors, ranking them lower than identical resumes without such credentials, as discovered by University of Washington researchers.
5. TwoMinutePapers - NVIDIA’s AI: Virtual Worlds, Now 10,000x Faster!
NVIDIA's AI significantly accelerates the simulation of virtual worlds, enabling real-time interaction with complex geometries like nuts and bolts, previously hindered by computational limitations.
6. Detecting hallucinations in large language models using semantic entropy
Researchers developed entropy-based uncertainty estimators for detecting confabulations in large language models (LLMs), focusing on arbitrary and incorrect outputs.
7. OpenAI is building its local app to subvert the permission system of the web
OpenAI is developing a local application to bypass traditional internet permissions, enabling AI to perform tasks directly on a user's computer as if it were the user.
8. What are open unsolved interesting problems in machine learning?
Identifying unsolved problems in machine learning is crucial for the field's advancement, focusing on obstacles that, once overcome, could significantly enhance its utility.
9. GNOME: Generating Negotiations through Open-Domain Mapping of Exchanges
GNOME enhances negotiation models' generalizability by generating synthetic open-domain dialogues, overcoming the limitation of models trained on closed-domain datasets.