Announced in 2016, Gym is an open-source Python library created to facilitate the development of support knowing algorithms. It aimed to standardize how environments are defined in AI research, making published research study more quickly reproducible [24] [144] while offering users with a basic user interface for engaging with these environments. In 2022, new developments of Gym have been relocated to the library Gymnasium. [145] [146]
Gym Retro

Released in 2018, Gym Retro is a platform for reinforcement learning (RL) research study on computer game [147] utilizing RL algorithms and study generalization. Prior RL research focused mainly on optimizing agents to solve single tasks. Gym Retro offers the ability to generalize between video games with similar concepts however various looks.

RoboSumo

Released in 2017, RoboSumo is a virtual world where humanoid metalearning robotic agents at first do not have knowledge of how to even stroll, however are given the objectives of learning to move and to press the opposing representative out of the ring. [148] Through this adversarial learning process, the representatives find out how to adjust to altering conditions. When an agent is then removed from this virtual environment and placed in a new virtual environment with high winds, the representative braces to remain upright, recommending it had discovered how to balance in a generalized way. [148] [149] OpenAI's Igor Mordatch argued that competition in between agents could create an intelligence "arms race" that might increase a representative's capability to function even outside the context of the competition. [148]
OpenAI 5

OpenAI Five is a group of 5 OpenAI-curated bots used in the competitive five-on-five video game Dota 2, that discover to play against human players at a high skill level totally through trial-and-error algorithms. Before ending up being a group of 5, the very first public demonstration took place at The International 2017, the yearly best champion competition for the video game, where Dendi, an expert Ukrainian player, lost against a bot in a live individually matchup. [150] [151] After the match, CTO Greg Brockman explained that the bot had actually found out by playing against itself for 2 weeks of actual time, which the learning software was a step in the instructions of producing software application that can handle complicated tasks like a surgeon. [152] [153] The system uses a type of reinforcement learning, as the bots learn gradually by playing against themselves numerous times a day for months, and are rewarded for actions such as killing an enemy and taking map objectives. [154] [155] [156]
By June 2018, the capability of the bots expanded to play together as a complete team of 5, and they were able to defeat groups of amateur and semi-professional players. [157] [154] [158] [159] At The International 2018, OpenAI Five played in 2 exhibition matches against expert gamers, but ended up losing both games. [160] [161] [162] In April 2019, OpenAI Five defeated OG, the reigning world champions of the game at the time, 2:0 in a live exhibit match in San Francisco. [163] [164] The public appearance came later that month, where they played in 42,729 overall video games in a four-day open online competition, winning 99.4% of those games. [165]
OpenAI 5's systems in Dota 2's bot player reveals the difficulties of AI systems in multiplayer online fight arena (MOBA) video games and how OpenAI Five has actually shown making use of deep reinforcement learning (DRL) agents to attain superhuman competence in Dota 2 matches. [166]
Dactyl

Developed in 2018, Dactyl uses maker learning to train a Shadow Hand, a human-like robotic hand, to control physical items. [167] It learns totally in simulation using the exact same RL algorithms and training code as OpenAI Five. OpenAI dealt with the item orientation problem by utilizing domain randomization, a simulation approach which exposes the learner to a range of experiences instead of attempting to fit to truth. The set-up for Dactyl, aside from having movement tracking cameras, likewise has RGB video cameras to allow the robot to manipulate an approximate object by seeing it. In 2018, OpenAI showed that the system was able to manipulate a cube and an octagonal prism. [168]
In 2019, OpenAI demonstrated that Dactyl could solve a Rubik's Cube. The robot was able to resolve the puzzle 60% of the time. Objects like the Rubik's Cube present intricate physics that is harder to model. OpenAI did this by enhancing the robustness of Dactyl to perturbations by using Automatic Domain Randomization (ADR), a simulation approach of generating progressively harder environments. ADR varies from manual domain randomization by not requiring a human to define randomization ranges. [169]
API

In June 2020, OpenAI revealed a multi-purpose API which it said was "for accessing brand-new AI designs developed by OpenAI" to let developers call on it for "any English language AI job". [170] [171]
Text generation

The company has promoted generative pretrained transformers (GPT). [172]
OpenAI's initial GPT design ("GPT-1")

The original paper on generative pre-training of a transformer-based language model was written by Alec Radford and his colleagues, and published in preprint on OpenAI's website on June 11, 2018. [173] It demonstrated how a generative model of language might obtain world understanding and procedure long-range reliances by pre-training on a diverse corpus with long stretches of adjoining text.

GPT-2

Generative Pre-trained Transformer 2 ("GPT-2") is a without supervision transformer language design and the follower to OpenAI's original GPT design ("GPT-1"). GPT-2 was revealed in February 2019, with only minimal demonstrative versions initially released to the general public. The complete version of GPT-2 was not immediately launched due to concern about prospective misuse, including applications for writing fake news. [174] Some professionals revealed uncertainty that GPT-2 posed a significant risk.

In action to GPT-2, the Allen Institute for Artificial Intelligence responded with a tool to identify "neural fake news". [175] Other scientists, such as Jeremy Howard, alerted of "the technology to totally fill Twitter, email, and the web up with reasonable-sounding, context-appropriate prose, which would drown out all other speech and be difficult to filter". [176] In November 2019, [forum.batman.gainedge.org](https://forum.batman.gainedge.org/index.php?action=profile