Make Nora+NAdamW the default optimizer
This commit is contained in:
@@ -107,7 +107,9 @@ nohup python scripts/tools/train.py \
|
||||
--batch_per_device=4 \
|
||||
--grad_accum_steps=8 \
|
||||
--warmup_ratio=0.05 \
|
||||
--optimizer=nora_nadamw \
|
||||
--max_lr=1e-4 \
|
||||
--nora_lr=5e-3 \
|
||||
--max_grad_norm=1.0 \
|
||||
--weight_decay=0.1 \
|
||||
--window_size=2048 \
|
||||
@@ -262,4 +264,4 @@ SSE 流式格式、错误码和统计端点详见[推理文档](guides/inference
|
||||
|
||||
<div align="center">
|
||||
<em>专为高性能与易用性设计的轻量级 Transformer 框架。</em>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
Reference in New Issue
Block a user