We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 336b000 commit da5f8f2Copy full SHA for da5f8f2
olmocr/train/configs/qwen25_vl_b100_x1_day3_json_wsd.yaml
@@ -2,7 +2,7 @@
2
3
# Project metadata
4
project_name: olmocr-qwen-vl-training
5
-run_name: qwen2.5-vl-7b-finetune-day3-json
+run_name: qwen2.5-vl-7b-finetune-day3-json-wsd
6
7
# Model configuration
8
model:
@@ -74,7 +74,8 @@ training:
74
learning_rate: 2e-5
75
lr_scheduler_type: warmup_stable_decay
76
lr_scheduler_kwargs:
77
- num_stable_steps: 4000
+ num_warmup_steps: 1000
78
+ num_decay_steps: 2000
79
warmup_ratio: 0.1
80
81
# Optimization
0 commit comments