1127 update to latest
This commit is contained in:
25
dllm/scripts/accelerate_configs/zero3_moe.yaml
Normal file
25
dllm/scripts/accelerate_configs/zero3_moe.yaml
Normal file
@ -0,0 +1,25 @@
|
||||
compute_environment: LOCAL_MACHINE
|
||||
debug: false
|
||||
deepspeed_config:
|
||||
deepspeed_moe_layer_cls_names: RND1DecoderLayer # LLaDAMoEDecoderLayer
|
||||
gradient_accumulation_steps: 1
|
||||
gradient_clipping: 1.0
|
||||
offload_optimizer_device: none
|
||||
offload_param_device: none
|
||||
zero3_init_flag: true
|
||||
zero3_save_16bit_model: true
|
||||
zero_stage: 3
|
||||
distributed_type: DEEPSPEED
|
||||
downcast_bf16: 'no'
|
||||
enable_cpu_affinity: false
|
||||
machine_rank: 0
|
||||
main_training_function: main
|
||||
mixed_precision: bf16
|
||||
num_machines: 1
|
||||
num_processes: 8
|
||||
rdzv_backend: static
|
||||
same_network: true
|
||||
tpu_env: []
|
||||
tpu_use_cluster: false
|
||||
tpu_use_sudo: false
|
||||
use_cpu: false
|
||||
Reference in New Issue
Block a user