7/ Visual Instruction Tuning – uses language-only GPT-4 to generate multimodal language-image instruction-following data; applies instruction tuning and introduces LLaVA, a large multimodal model for general-purpose visual and language understanding.
@dair_ai
-
Generative Disco: AI Music Visualization System
By
–
5/ Generative Disco – an AI system based on LLMs and text-to-image models that generates music visualizations.
-
Generative Search Engines Citation Accuracy Audit Results
By
–
4/ Verifiability in Generative Search Engines – performs human evaluation to audit popular generative search engines like Bing Chat & NeevaAI; only 52% of statements are supported by citations & 75% of citations support their associated sentence. https://
x.com/nelsonfliu/sta
tus/1649075151303221249?s=20
… -
Deep Learning Framework Enables Large-Scale Biomolecular Dynamics Simulation
By
–
3/ Deep Learning for Large-Scale Biomolecular Dynamics – presents a framework for large-scale biomolecular simulation; this is achieved through the high accuracy of equivariant deep learning and the ability to scale to large and long simulations. https://
x.com/simonbatzner/s
tatus/1649214171236691969?s=20
… -
Pythia: Suite for Analyzing LLMs Across Training and Scaling
By
–
9/ Pythia – a suite for analyzing LLMs across training and scaling; includes 16 LLMs trained on public data and ranging in size from 70M to 12B parameters.
-
SegGPT: Generalist Segmentation Model with In-Context Learning
By
–
10/ SegGPT – unifies segmentation tasks into a generalist model through an in-context framework that supports different kinds of data.https://t.co/PiocmACjfo
— DAIR.AI (@dair_ai) 9 avril 202310/ SegGPT – unifies segmentation tasks into a generalist model through an in-context framework that supports different kinds of data.
-
Self-Improving Code LLMs Boost Performance Through Synthetic Data
By
–
7/ Self-Improving Code LLMs- generates pseudo data from knowledge gained through pre-training & fine-tuning; adds the data to the training dataset for the next step; shows that different code generation frameworks can be improved in performance.
-
ChatGPT and GPT-4 Research Overview: Applications and Analysis
By
–
8/ Summary of ChatGPT/GPT-4 Research – an overview of applications of ChatGPT and GPT-4; the analysis is done on 194 relevant papers and discusses capabilities, limitations, concerns, and more.
-
Baize: Open-Source Chat Model Fine-Tuned with LoRA
By
–
5/ Baize – an open-source chat model fine-tuned with LoRA. Leverages 100K dialogs generated from ChatGPT chatting with itself; it releases the dialogs along with 7B, 13B, and 30B parameter models.
-
Machiavelli Benchmark: Evaluating LLM Ethics in Adventure Games
By
–
6/ Machiavelli Benchmark – a new benchmark of 134 text-based Choose-Your-Own-Adventure games to evaluate the capabilities and unethical behaviors of LLMs.