Li Hui

icon indicating an email [email protected]

icon location Beijing icon location Work happily and live seriously

Results 42 comments of Li Hui

Gibberish generated with deepseek-ai/deepseek-coder-6.7b-base

How to fix this?

Tiny refactor DeepSeek V3/R1 NextN shared experts fusion

> will fused shared experts still improve performance with nextn? Yes, I'm still experimenting with the current effects

Tiny refactor DeepSeek V3/R1 NextN shared experts fusion

> Can you add a test case? Ok. I will add it

Tiny refactor DeepSeek V3/R1 NextN shared experts fusion

> Maybe my PR can be firstly merged to make the commit history a bit more clear Yes, I'm waiting for it to be merged @fzyzcjy

Tiny refactor DeepSeek V3/R1 NextN shared experts fusion

@BBuf @merrymercy @zhyncs could you take a look?

Tiny refactor DeepSeek V3/R1 NextN shared experts fusion

> any update in this PR? No, can merge it in. @xihuai18

Fix channel-wise INT8 moe config tuning error

@merrymercy @zhyncs Do you have time to take a look?

Fix channel-wise INT8 moe config tuning error

@zhyncs Do you have time to take a look?

Fix channel-wise INT8 moe config tuning error

@zhyncs See if there are any other problems. If not, please merge them. Thank you.

Fix channel-wise INT8 moe config tuning error

is this pr ready to merge? @zhyncs @BBuf