Skip to main content
verl-org
Projects
DAPO Reproduction on verl
Log in
Sign up
Overview
Workspace
Runs
Automat.
Sweeps
Reports
Artifacts
Qiying's workspace
Personal workspace
Automated workspace
Changes are only visible to you.
Runs
11
Name
11 visualized
DAPO-Qwen2.5-7b-MATH-0527a1
DAPO-Qwen2.5-7b-MATH-0527a1
DAPO-Qwen3-30B-A3B-Base-MATH-0527a1
DAPO-Qwen3-30B-A3B-Base-MATH-0527a1
DAPO-Qwen2.5-32B
DAPO-Qwen2.5-32B
DAPO w/o Dynamic Sampling
DAPO w/o Dynamic Sampling
DAPO w/o Dynamic Sampling
DAPO w/o Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
1-11
of 11
val/seq_reward/wo32/MATH##AIME
val/seq_reward/wo32/MATH##AIME
0
100
200
300
Step
-1.7
-1.6
-1.5
-1.4
-1.3
-1.2
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
DAPO w/o Token-Level Loss & Dynamic Sampling
Previous
Next