This is the sixth lecture in the Language Models and Intelligent Agentic Systems course, run by Meridian Cambridge in collaboration with the Cambridge Centre for Data Driven Discovery (C2D3).
This lecture covers reward modelling. Reward modelling is the problem of training a neural network to output rewards in response to behaviours which acurately capture human preferences over those behaviours. We'll cover the motivation behind reward modelling, the Bradley-Terry loss used to train reward models, how preference data is obtained to train these models, and shortcoming and open problems in current reward modelling.
The slides for the lecture can be found here: https://tinyurl.com/LMaIAS
Give feedback on this lecture here: https://airtable.com/appjc4HBUw4Ktlij...
Give feedback on the lecture series as a whole here: https://airtable.com/appjc4HBUw4Ktlij...
Поделиться в:
Доступные форматы для скачивания:
Скачать видео mp4
Информация по загрузке:
Скачать аудио mp3
Похожие видео
array(10) {
[0]=>
object(stdClass)#4848 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "8mRG0JtxLL0"
["related_video_title"]=>
string(53) "Lecture 07 • Agents and Agent Architectures"
["posted_time"]=>
string(23) "1 месяц назад"
["channelName"]=>
string(18) "Meridian Cambridge"
}
[1]=>
object(stdClass)#4821 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "UiGa8Bx1Srk"
["related_video_title"]=>
string(59) "Lecture 01 • Introduction to Language Models"
["posted_time"]=>
string(23) "1 месяц назад"
["channelName"]=>
string(18) "Meridian Cambridge"
}
[2]=>
object(stdClass)#4846 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "BfAcYUt3gyQ"
["related_video_title"]=>
string(45) "Why Your Coaching Business Needs This in 2025"
["posted_time"]=>
string(24) "10 часов назад"
["channelName"]=>
string(10) "Laura Agar"
}
[3]=>
object(stdClass)#4853 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "zYGDpG-pTho"
["related_video_title"]=>
string(62) "RAG vs Fine-Tuning vs Prompt Engineering: Optimizing AI Models"
["posted_time"]=>
string(25) "2 месяца назад"
["channelName"]=>
string(14) "IBM Technology"
}
[4]=>
object(stdClass)#4832 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "3TqD_vcykaQ"
["related_video_title"]=>
string(68) "Lecture 11 • Deceptive Alignment and Alignment Faking"
["posted_time"]=>
string(25) "3 недели назад"
["channelName"]=>
string(18) "Meridian Cambridge"
}
[5]=>
object(stdClass)#4850 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "UKcWu1l_UNw"
["related_video_title"]=>
string(58) "THIS is why large language models can understand the world"
["posted_time"]=>
string(25) "2 месяца назад"
["channelName"]=>
string(22) "Algorithmic Simplicity"
}
[6]=>
object(stdClass)#4845 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "j1pMDywaLrk"
["related_video_title"]=>
string(109) "Война набирает обороты: Жертвы, разрушения и ответные удары"
["posted_time"]=>
string(23) "5 часов назад"
["channelName"]=>
string(31) "Сергей Ауслендер"
}
[7]=>
object(stdClass)#4855 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "hbqs81QT7bI"
["related_video_title"]=>
string(69) "Lecture 09 • Reward Hacking and Goal Misgeneralisation"
["posted_time"]=>
string(25) "4 недели назад"
["channelName"]=>
string(18) "Meridian Cambridge"
}
[8]=>
object(stdClass)#4831 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "fNWkwLAFg_0"
["related_video_title"]=>
string(57) "Lecture 04 • Post-Training Language Models"
["posted_time"]=>
string(23) "1 месяц назад"
["channelName"]=>
string(18) "Meridian Cambridge"
}
[9]=>
object(stdClass)#4849 (5) {
["video_id"]=>
int(9999999)
["related_video_id"]=>
string(11) "neA0NJNUEfM"
["related_video_title"]=>
string(25) "2015 10 30 Claude Shannon"
["posted_time"]=>
string(19) "9 лет назад"
["channelName"]=>
string(30) "MIT Video Productions External"
}
}
