zhexiang zhang's picture

1 1

zhexiang zhang

Somniloquy

AI & ML interests

None yet

Recent Activity

upvoted a paper 2 days ago

Janus: Disaggregating Attention and Experts for Scalable MoE Inference

new activity over 1 year ago

mistralai/Mistral-7B-Instruct-v0.2:deepspeed inference tensor parallelism memory footprint doesn't decrease with deepspeed tp_size increase.

View all activity

Organizations

None yet

models 0

None public yet

datasets 0

None public yet