Weak-to-Strong Extrapolation Expedites Alignment
Paper
•
2404.16792
•
Published
•
11
Weak-to-Strong Extrapolation Expedites Alignment
Better aligned models obtained by weak-to-strong model extrapolation (ExPO)
-
-
chujiezheng/Smaug-Llama-3-70B-Instruct-ExPO
Text Generation • Updated • 690 • 2 -
chujiezheng/Smaug-34B-v0.1-ExPO
Text Generation • Updated • 2.92k -
chujiezheng/Starling-LM-7B-beta-ExPO
Text Generation • Updated • 743 • 2 -
chujiezheng/Starling-LM-7B-alpha-ExPO
Text Generation • Updated • 743 -
chujiezheng/LLaMA3-iterative-DPO-final-ExPO
Text Generation • Updated • 699 • 2 -
chujiezheng/Mistral7B-PairRM-SPPO-ExPO
Text Generation • Updated • 3.56k -
chujiezheng/tulu-2-dpo-70b-ExPO
Text Generation • Updated • 1.02k -
chujiezheng/Llama3-8B-Chinese-Chat-ExPO
Text Generation • Updated • 688 • 1 -
chujiezheng/Llama-3-Instruct-8B-SimPO-ExPO
Text Generation • Updated • 89 • 16 -
chujiezheng/tulu-2-dpo-13b-ExPO
Text Generation • Updated • 37 -
chujiezheng/tulu-2-dpo-7b-ExPO
Text Generation • Updated • 20 -
chujiezheng/Snorkel-Mistral-PairRM-DPO-ExPO
Text Generation • Updated • 17 -
chujiezheng/Llama3-70B-Chinese-Chat-ExPO
Text Generation • Updated • 13 -
chujiezheng/neo_7b_instruct_v0.1-ExPO
Text Generation • Updated • 16 -
chujiezheng/zephyr-7b-beta-ExPO
Text Generation • Updated • 20 -
chujiezheng/zephyr-7b-alpha-ExPO
Text Generation • Updated • 22 -
chujiezheng/zephyr-7b-dpo-full-ExPO
Text Generation • Updated • 21 -
chujiezheng/internlm2-chat-20b-ExPO
Text Generation • Updated • 14 • 1 -
chujiezheng/internlm2-chat-7b-ExPO
Text Generation • Updated • 10 -
chujiezheng/internlm2-chat-1_8b-ExPO
Text Generation • Updated • 15 • 1