3/ Visual ChatGPT – it connects ChatGPT and different visual foundation models to enable users to interact with ChatGPT beyond language format.
@dair_ai
-
Aligning Text-to-Image Models Using Human Feedback
By
–
9/ Aligning Text-to-Image Models using Human Feedback – proposes a fine-tuning method to align generative models using human feedback.
-
MERF: Memory-Efficient Radiance Field for Browser-Based View Synthesis
By
–
10/ MERF – a memory-efficient radiance field representation for real-time view synthesis of large-scale scenes in a browser.https://t.co/o4dpMgDVLh
— DAIR.AI (@dair_ai) 26 février 202310/ MERF – a memory-efficient radiance field representation for real-time view synthesis of large-scale scenes in a browser.
-
LLMs Optimize Code with Functionally Correct Performance Improvements
By
–
7/ LLMs to Optimize Code – an approach that uses LLMs to suggest functionally correct, performance-improving code edits.
-
Prompt Injection Threats: Comprehensive Analysis of LLM Security Risks
By
–
8/ Prompt Injection Threats – a comprehensive analysis of novel prompt injection threats to application-integrated LLMs.
-
Modular Deep Learning Survey: Building Blocks Neural Networks
By
–
5/ Modular Deep Learning – a survey offering a unified view of the building blocks of modular neural networks; it also includes a discussion about modularity in the context of scaling LMs, causal inference, and other key topics in ML.
-
Recitation-Augmented LMs Enhance Knowledge-Intensive Task Performance
By
–
6/ Recitation-Augmented LMs – an approach that recites passages from the LLM’s own memory to produce final answers; shows high performance on knowledge-intensive tasks.
-
Top ML Papers Week: LLaMA, Diffusion Models, Code Optimization
By
–
Top ML Papers of the Week (Feb 20-26): – LLaMA (65B params foundation model)
– Composer (5B params diffusion model)
– Active-Prompt
– Modular Deep Learning
– Memory-Efficient Radiance Fields
– LLMs to Optimize Code … -
LLaMA 65B: Meta’s Open Source Model Outperforms GPT-3
By
–
1/ LLaMA – a 65B parameter foundation model released by Meta AI; relies on publicly available data and outperforms GPT-3 on most benchmarks despite being 10x smaller.
-
Composer: 5B Parameter Creative Diffusion Model for Text-Image Generation
By
–
2/ Composer – a 5B parameter creative and controllable diffusion model trained on billions (text, image) pairs.