LLMs and irrelevant context – finds that many prompting techniques fail when presented with irrelevant context for arithmetic reasoning. 11 of 11
RESEARCH
-
ChatGPT Mathematical Capabilities Evaluated on GHOSTS Benchmark
By
–
Mathematical Capabilities of ChatGPT – investigates the mathematical capabilities of ChatGPT on a new holistic benchmark called GHOSTS. 8 of 11
-
Training AI Agents to Navigate Without Vision or Audio
By
–
Training ‘Blind’ Agents – trains an AI agent to navigate purely by feeling its way around; no use of vision, audio, or any other sensing (as in animals). 9 of 11
-
SceneDreamer: Generative Model Synthesizes Large-Scale 3D Landscapes
By
–
SceneDreamer – a generative model that synthesizes large-scale 3D landscapes from random noises.
— DAIR.AI (@dair_ai) 5 février 2023
10 of 11https://t.co/47fdG03mLeSceneDreamer – a generative model that synthesizes large-scale 3D landscapes from random noises. 10 of 11
-
Dreamix: Text-Based Video Motion and Appearance Editing Model
By
–
Dreamix – a diffusion model that performs text-based motion and appearance editing of general videos.
— DAIR.AI (@dair_ai) 5 février 2023
6 of 11https://t.co/tURn5XhkzoDreamix – a diffusion model that performs text-based motion and appearance editing of general videos. 6 of 11
-
Benchmarking LLMs for News Summarization Performance
By
–
Benchmarking LLMs for news summarization. 7 of 11
-
Multimodal Chain-of-Thought Reasoning Advances Model Inference
By
–
Multimodal Chain-of-Though Reasoning – incorporates vision features to elicit chain-of-thought reasoning in multimodality, enabling the model to generate effective rationales that contribute to answer inference. 5 of 11
-
REPLUG: Retrieval-Augmented Framework for Large Language Models
By
–
REPLUG – a retrieval-augmented LM framework that adapts a retriever to a large-scale, black-box LM like GPT-3. 2 of 11 https://
x.com/WeijiaShi2/sta
tus/1620497381962977281?s=20&t=CkO9Xon5RzSgwvKKf7aYyQ
… -
Diffusion Models Memorize Training Data Images
By
–
Extracting Training Data from Diffusion Models – shows that diffusion-based generative models can memorize images from the training data and emit them at generation time. 3 of 11
-
Top ML Papers of the Week: REPLUG, SceneDreamer, FLAN
By
–
Top ML Papers of the Week (Jan 30 – Feb 5): – REPLUG
– SceneDreamer
– The FLAN collection
– Distractability of LLMs
– Blind navigation agents
– Mathematical capabilities of ChatGPT
… 1 of 11