Blair Yang's picture

Blair Yang

loveblairsky

·

blairyeung

AI & ML interests

Alignment, red teaming

Recent Activity

liked a model 5 days ago

PhalaCloud/GLM-5.2-W4AFP8

new activity 6 days ago

PhalaCloud/GLM-5.2-W4AFP8:GLM-5.2-W4AFP8 on 8×H100: fp8_e4m3 KV cache produces corrupted output, while BF16 KV works correctly

commentedon a paper 3 months ago

ThinkTwice: Jointly Optimizing Large Language Models for Reasoning and Self-Refinement

View all activity

Organizations

New activity in PhalaCloud/GLM-5.2-W4AFP8 6 days ago

GLM-5.2-W4AFP8 on 8×H100: fp8_e4m3 KV cache produces corrupted output, while BF16 KV works correctly

#3 opened 6 days ago by

commented a paper 3 months ago

ThinkTwice: Jointly Optimizing Large Language Models for Reasoning and Self-Refinement

Paper • 2604.01591 • Published Apr 2 • 42 •

commented a paper almost 2 years ago

Report Cards: Qualitative Evaluation of Language Models Using Natural Language Summaries

Paper • 2409.00844 • Published Sep 1, 2024 • 12 •