deepseek && qwen tp performance tuning#934

Merged

hiworldwzj merged 57 commits into

mainfrom

fused_moe_improve

Aug 20, 2025

Commits on Jun 13, 2025

Commits on Jun 15, 2025

update tuning
hiworldwzj
committed
fix
hiworldwzj
committed

Commits on Jun 16, 2025

Commits on Jun 19, 2025

Commits on Jun 24, 2025

Commits on Jul 1, 2025

Commits on Jul 8, 2025

update
shihaobai
committed

Commits on Jul 25, 2025

Commits on Jul 31, 2025

add grouped_moe_gemm_kernel configs (#994)
sufubao
authored

Commits on Aug 6, 2025

Torch.ops (#1001)
sufubao
authored

Commits on Aug 15, 2025

fix
none
committed
fix
hiworldwzj
authored
fix
none
committed
fix moe tunning
none
committed
fix
none
committed
fix
none
committed
fix
none
committed
fix
none
committed
Fix
none
committed
fix rotary_emb
none
committed
fix
none
committed
fix
none
committed

Commits on Aug 17, 2025

add deepseekv3 rotary config
hiworldwzj
committed

Commits on Aug 19, 2025

fix
hiworldwzj
authored
fix
hiworldwzj
authored
improve rotary emb for deepseek
none
committed
update fp8 quant
shihaobai
committed
fix
shihaobai
committed
add config for moe_and_silu
none
committed
rm moe_and_slilu A800 and H800 config
none
committed
update
shihaobai
committed
Merge branch 'fused_moe_improve' of https://github.com/ModelTC/lightllm into fused_moe_improve
shihaobai
committed
fix router_scaling_scale
none
committed
fix otray emb config
none
committed
fix
none
committed
fix
none
committed
fix
none
committed
fix
shihaobai
committed
reformat
shihaobai
committed
update sliu grid
shihaobai
committed
fix
shihaobai
committed
update moe silu
shihaobai
committed