tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Remove learning rate scheduler in deepspeed config to avoid conflict (#909)

Browse Source
This commit is contained in:
Haoxiang Wang
2023-12-04 02:17:38 -08:00
committed by GitHub
parent 3e3229e2d9
commit 476a205cea
3 changed files with 0 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
deepspeed/zero1.json
View File

@@ -24,16 +24,6 @@
"weight_decay": "auto" "weight_decay": "auto"
} }
}, },
"scheduler": {
"type": "WarmupDecayLR",
"params": {
"warmup_min_lr": "auto",
"warmup_max_lr": "auto",
"warmup_num_steps": "auto",
"warmup_type": "linear",
"total_num_steps": "auto"
}
},
"gradient_accumulation_steps": "auto", "gradient_accumulation_steps": "auto",
"train_batch_size": "auto", "train_batch_size": "auto",
"train_micro_batch_size_per_gpu": "auto", "train_micro_batch_size_per_gpu": "auto",

10
deepspeed/zero2.json
View File

@@ -28,16 +28,6 @@
"weight_decay": "auto" "weight_decay": "auto"
} }
}, },
"scheduler": {
"type": "WarmupDecayLR",
"params": {
"warmup_min_lr": "auto",
"warmup_max_lr": "auto",
"warmup_num_steps": "auto",
"warmup_type": "linear",
"total_num_steps": "auto"
}
},
"gradient_accumulation_steps": "auto", "gradient_accumulation_steps": "auto",
"train_batch_size": "auto", "train_batch_size": "auto",
"train_micro_batch_size_per_gpu": "auto", "train_micro_batch_size_per_gpu": "auto",

10
deepspeed/zero3.json
View File

@@ -32,16 +32,6 @@
"weight_decay": "auto" "weight_decay": "auto"
} }
}, },
"scheduler": {
"type": "WarmupDecayLR",
"params": {
"warmup_min_lr": "auto",
"warmup_max_lr": "auto",
"warmup_num_steps": "auto",
"warmup_type": "linear",
"total_num_steps": "auto"
}
},
"gradient_accumulation_steps": "auto", "gradient_accumulation_steps": "auto",
"train_batch_size": "auto", "train_batch_size": "auto",
"train_micro_batch_size_per_gpu": "auto", "train_micro_batch_size_per_gpu": "auto",