AP
Появился?
Size: a a a
2017 August 10
AN
пока нет
AP
Блин, тогда надо все рестартить
AP
видно слышно?
AP
пока только мебя показываю
AP
себя
AN
слышно теперь )
АС
@justHeuristic ты там какие-то материалы по pytorch'у хотел скинуть
2017 August 11
2017 August 12
ME
EZ
Neural Network Dynamics for Model-Based Deep Reinforcement Learning with Model-Free Fine-Tuning https://arxiv.org/abs/1708.02596
2017 August 14
AP
А делали уже обзор этого
https://arxiv.org/abs/1707.06170 Learning model-based planning from scratch
или этого
https://arxiv.org/abs/1707.06203 Imagination-Augmented Agents for Deep Reinforcement Learning
?
https://arxiv.org/abs/1707.06170 Learning model-based planning from scratch
или этого
https://arxiv.org/abs/1707.06203 Imagination-Augmented Agents for Deep Reinforcement Learning
?
📒
вроде нет
2017 August 15
AP
а есть где-то подробности про openai и доту? может есть какая-то публикация?
KB
@Parilo вышла статья на Verge
https://www.reddit.com/r/MachineLearning/comments/6tqt50/d_openai_used_the_dota_bot_apimusk_stepped_in_and/
> From the verge article "OpenAI’s Greg Brockman confirmed to The Verge that the AI did indeed use the API, and that certain techniques were hardcoded in the agent, including the items it should use in the game. It was also taught certain strategies (like one called “creep block”) using a trial-and-error technique known as reinforcement learning. Basically, it did get a little coaching."
https://www.reddit.com/r/MachineLearning/comments/6tqt50/d_openai_used_the_dota_bot_apimusk_stepped_in_and/
> From the verge article "OpenAI’s Greg Brockman confirmed to The Verge that the AI did indeed use the API, and that certain techniques were hardcoded in the agent, including the items it should use in the game. It was also taught certain strategies (like one called “creep block”) using a trial-and-error technique known as reinforcement learning. Basically, it did get a little coaching."
AP
а с техническими подробностями? какой алгоритм был, и прочее
KB
Неа, такого не видел. OpenAI такой Open 🤷♂️
N
Есть такое поверхностное описание — http://www.wildml.com/2017/08/hype-or-not-some-perspective-on-openais-dota-2-bot/
P
Intrinsically Motivated Goal Exploration Processes with Automatic Curriculum Learning (https://arxiv.org/abs/1708.02190) – its description is in its title
Neural Expectation Maximization (https://arxiv.org/abs/1708.03498) – Another step forward to identification and learning of conceptual entities.
Neural Expectation Maximization (https://arxiv.org/abs/1708.03498) – Another step forward to identification and learning of conceptual entities.