updating to fused (#2293)
This commit is contained in:
@@ -37,7 +37,7 @@ wandb_log_model:
|
||||
gradient_accumulation_steps: 4
|
||||
micro_batch_size: 2
|
||||
num_epochs: 1
|
||||
optimizer: adamw_torch
|
||||
optimizer: adamw_torch_fused
|
||||
lr_scheduler: cosine
|
||||
learning_rate: 2e-5
|
||||
|
||||
|
||||
@@ -30,7 +30,7 @@ lora_target_linear: true
|
||||
gradient_accumulation_steps: 4
|
||||
micro_batch_size: 1
|
||||
num_epochs: 2
|
||||
optimizer: adamw_torch
|
||||
optimizer: adamw_torch_fused
|
||||
lr_scheduler: cosine
|
||||
learning_rate: 0.00001
|
||||
|
||||
|
||||
@@ -39,7 +39,7 @@ wandb_log_model:
|
||||
gradient_accumulation_steps: 4
|
||||
micro_batch_size: 1
|
||||
num_epochs: 4
|
||||
optimizer: adamw_torch
|
||||
optimizer: adamw_torch_fused
|
||||
lr_scheduler: cosine
|
||||
learning_rate: 0.00001
|
||||
|
||||
|
||||
Reference in New Issue
Block a user