OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling
Paper
•
2506.20512
•
Published
•
21
🐙 OctoThinker
Mid-training Incentivizes Reinforcement Learning Scaling
-
-
OctoThinker/MegaMath-Web-Pro-Max
Updated • 5 -
OctoThinker/OctoThinker-8B-Long-Base
Updated • 11 -
OctoThinker/OctoThinker-8B-Hybrid-Base
Updated • 12 • 2 -
OctoThinker/OctoThinker-8B-Short-Base
Updated • 46 -
OctoThinker/OctoThinker-3B-Short-Zero
Updated • 8 -
OctoThinker/OctoThinker-3B-Hybrid-Zero
Updated • 32 -
OctoThinker/OctoThinker-1B-Long-Zero
Updated • 7 -
OctoThinker/OctoThinker-1B-Hybrid-Zero
Updated • 7 -
OctoThinker/OctoThinker-1B-Short-Zero
Updated • 41 -
OctoThinker/Llama3.2-3B-Zero
Updated • 11 -
OctoThinker/OctoThinker-3B-Long-Zero
Updated • 8 -
OctoThinker/OctoThinker-1B-Long-Base
Updated • 19 -
OctoThinker/OctoThinker-1B-Short-Base
Updated • 23 -
OctoThinker/OctoThinker-1B-Hybrid-Base
Updated • 12 -
OctoThinker/OctoThinker-3B-Long-Base
Updated • 22 -
OctoThinker/OctoThinker-3B-Hybrid-Base
Updated • 7 -
OctoThinker/OctoThinker-3B-Short-Base
Updated • 9