Novelty-Guided Reinforcement Learning via Encoded Behaviors

Ramamurthy, Rajkumar; Sifa, Rafet; Lübbering, Max; Bauckhage, Christian

doi:10.1109/ijcnn48605.2020.9206982

Cited by 15 publications

(20 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although RLHF has shown promising results by incorporating fluency, progress in this field is impeded by a lack of publicly available benchmarks and implementation resources, leading to a perception that RL is a challenging approach for NLP. To address this issue, an open-source library named RL4LMs [49] has recently been introduced, consisting of building blocks for fine-tuning and evaluating RL algorithms on LM-based generation.…”

Section: Reinforcement Learning From Human Feedbackmentioning

confidence: 99%

A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT

Cao¹,

Li²,

Liu³

et al. 2023

Preprint

View full text Add to dashboard Cite

Recently, ChatGPT, along with DALL-E-2 [1] and Codex [2],has been gaining significant attention from society. As a result, many individuals have become interested in related resources and are seeking to uncover the background and secrets behind its impressive performance. In fact, ChatGPT and other Generative AI (GAI) techniques belong to the category of Artificial Intelligence Generated Content (AIGC), which involves the creation of digital content, such as images, music, and natural language, through AI models. The goal of AIGC is to make the content creation process more efficient and accessible, allowing for the production of high-quality content at a faster pace. AIGC is achieved by extracting and understanding intent information from instructions provided by human, and generating the content according to its knowledge and the intent information. In recent years, large-scale models have become increasingly important in AIGC as they provide better intent extraction and thus, improved generation results. With the growth of data and the size of the models, the distribution that the model can learn becomes more comprehensive and closer to reality, leading to more realistic and high-quality content generation. This survey provides a comprehensive review on the history of generative models, and basic components, recent advances in AIGC from unimodal interaction and multimodal interaction. From the perspective of unimodality, we introduce the generation tasks and relative models of text and image. From the perspective of multimodality, we introduce the cross-application between the modalities mentioned above. Finally, we discuss the existing open problems and future challenges in AIGC.

show abstract

Section: Reinforcement Learning From Human Feedbackmentioning

confidence: 99%

A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT

Cao¹,

Li²,

Liu³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…2.4). For this reason, disco has a wider scope than other related toolkits such as RL4LM (Ramamurthy et al, 2022), which centers on RL methods only. Nevertheless, there is a large space for cross-polination between RL-based frameworks and disco because of similarities in the algorithms (Korbak et al, 2022b).…”

Section: Related Work and Conclusionmentioning

confidence: 99%

disco: a toolkit for Distributional Control of Generative Models

Kruszewski¹,

Rozen²,

Dymetman³

2023

Preprint

View full text Add to dashboard Cite

Pre-trained language models and other generative models have revolutionized NLP and beyond. However, these models tend to reproduce undesirable biases present in their training data. Also, they may overlook patterns that are important but challenging to capture. To address these limitations, researchers have introduced distributional control techniques. These techniques, not limited to language, allow controlling the prevalence (i.e. expectations) of any features of interest in the model's outputs. Despite their potential, the widespread adoption of these techniques has been hindered by the difficulty in adapting the complex, disconnected code. Here, we present disco, an open-source Python library that brings these techniques to the broader public. 1

show abstract

“…On a standard laptop, OpenRL can complete the training of the CartPole task in just a few seconds. Compared to the RL4LMs framework (Ramamurthy et al, 2022), our training speed for dialogue tasks has improved by 17%, with improvements in various performance indicators as well (see Appendix C for specific experimental results).…”

Section: High Performancementioning

confidence: 99%

“…The table below presents the results of training on the dialogue task (Li et al, 2017) using OpenRL and comparing them with RL4LMs (Ramamurthy et al, 2022).…”

Section: Appendix a Openrl's General Code Interfacementioning

confidence: 99%

New dictyopteran fossils from the Pennsylvanian of West Beijing, China

Huang

WEI

et al. 2022

View full text Add to dashboard Cite

Carboniferous insects from China are rare. Here we report a new fossiliferous locality at the Yexi Section, western Beijing, from where Pennsylvanian insects have been discovered. These insect fossils collected from the black shales of the Benxi Formation (middle–late Moscovian to early Kasimovian) are associated with rich plant remains. The fossils are represented mainly by dictyopteran tegmina that cannot be further identified. The discovery is helpful for our understanding of the diversity and preservation of Carboniferous insects in western Beijing and also the Benxi Formation of North China.

show abstract

Novelty-Guided Reinforcement Learning via Encoded Behaviors

Cited by 15 publications

References 12 publications

A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT

A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT

disco: a toolkit for Distributional Control of Generative Models

New dictyopteran fossils from the Pennsylvanian of West Beijing, China

Contact Info

Product

Resources

About