사람 같은 AI, AI 같은 사람
[융합연구팀 전동준] 우리가 일반적으로 생각했었던 “AI 같은”의 의미는 사람과 대비되는 기계의 느낌이 강했습니다. 감정이 느껴지지 않고, 말투도 딱딱한 사람들을 “AI 같은 사람” 이라고 많이 표현했었습니다.
Advancement of search system and utilization of NLP
[Convergence Research Team Ji-Hyun Song] Deep Natural Language Processing for LinkedIn Search Systems I am feeling how effectively the structuring and optimization design of a chatbot system is applied recently while studying Rasa Open Source. Even within it, the intent and entity…
Instruction tuning – FLAN
[Convergence Research Team Hongmae Shim] If we were to pick the top 10 keywords for 2020 in the field of NLP, of course it would be in the GPT-3 (Language Models are Few shot Learners) ranking. To this day, the enormous amount of parameters and excellent performance of GPT-3 is still in the field of NLP…
Copilot, artificial intelligence that helps coding
[Prior Research Team Seonghyun Kim] The causal language model (eg GPT-3), which caused a great sensation in natural languages, has now been applied to programming languages beyond natural languages! 🤗 Last June, Copilot, jointly developed by Github and OpenAI, was released. Copilot is like GPT-3 in natural language, given code…
Droidlet: an open source platform for developing intelligent robots
[Service Development Team Jeon Jeon-jun] Facebook AI unveiled the Droidlet platform for robot development that can be used in real and virtual environments on the 28th of last month.
Blender 2.0 overcomes the limitations of Open Domain chatbots
[Prior Research Team Jihyun Song] It has been over 2 years since I was interested in Open Domain chatbot and came across papers on Blender 1.0 and Meena. At that time, they had a consistent long-ton conversation that they claimed they would overcome in the future, and about knowledge…
Multimodal Few-Shot Learner
[Prior Research Team Hongmae Shim] As super-giant language models such as Open AI's GPT-3 and NAVER's Hyper CLOVA have been released, various examples and services using them are pouring out recently. Such a giant language...
Reinforcement learning environment for solving real-world problems
[Prior Research Team, Jeongwoo Lee] We have been using games (Go, chess, Atari games, etc.) to verify the performance of reinforcement learning algorithms for a long time. With the development of algorithms, in the field of reinforcement learning, like other images and natural language fields, there are...
Open Source GPT-3 Model Released, GPT-J-6B
[Priority Research Team Seonghyun Kim] The Eleuther AI team, which was releasing the GPT-3 model under an open license, recently released a new GPT-3 model with 6B (about 6 billion) parameters. (Link) It is said that the name GPT-J-6B was given because Mesh-tensorflow and JAX were used. The training data is about…
