botbw 9b9b76bdcd [moe] add mixtral dp grad scaling when not all experts are activated

2024-08-01 10:06:59 +08:00

..

moe_utils.py

[zero] solve hang

2024-08-01 10:06:59 +08:00

test_deepseek_layer.py

[shardformer] DeepseekMoE support (#5871)

2024-07-05 16:13:58 +08:00

test_grad_handler.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00

test_kernel.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00

test_mixtral_layer.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00

test_moe_checkpoint.py

[zero] solve hang

2024-08-01 10:06:59 +08:00

test_moe_ep_tp.py

[misc] solve booster hang by rename the variable

2024-08-01 10:06:59 +08:00

test_moe_ep_zero.py

[moe] add mixtral dp grad scaling when not all experts are activated

2024-08-01 10:06:59 +08:00

test_moe_group.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00

test_moe_hybrid_zero.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00

test_moe_load_balance.py

[MoE/ZeRO] Moe refactor with zero refactor (#5821)

2024-06-28 14:00:08 +08:00