Skip to main content

Pipeline RL blog post results

Created on April 23|Last edited on April 29

5001k1.5kStep1020304050
100k200k300k400k500kStep102030405060
200k400k600k800k1MStep50100150200250
50100150200Time (hours)20406080
group: dzmitry/reason/openreas_qwen32b_r21
group: dzmitry/reason/openreas_qwen7b_r18
200k400k600k800k1MStep20000400006000080000100000
5001k1.5kStep00.0000010.0000020.0000030.000004
01M2M3M4M5M6Mactor/published_model_version1000150020002500
01M2M3M4M5M6Mactor/published_model_version0.50.60.70.80.9
0200k400k600k800k1MStep00.20.40.60.81
0200k400k600k800k1MStep00.20.40.60.81
50100150200Time (hours)00.20.40.60.81
50100150200Time (hours)00.20.40.60.81
50100150200Time (hours)100020003000400050006000
blog
6