OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling
Paper
•
2506.20512
•
Published
•
46
🐙 OctoThinker
Mid-training Incentivizes Reinforcement Learning Scaling
-
-
OctoThinker/MegaMath-Web-Pro-Max
Viewer • Updated • 69.2M • 4.83k • 34 -
OctoThinker/OctoThinker-8B-Long-Base
Text Generation • 8B • Updated • 45 -
OctoThinker/OctoThinker-8B-Hybrid-Base
Text Generation • 8B • Updated • 96 • 2 -
OctoThinker/OctoThinker-8B-Short-Base
Text Generation • 8B • Updated • 1k -
OctoThinker/OctoThinker-3B-Short-Zero
Text Generation • 4B • Updated • 11 -
OctoThinker/OctoThinker-3B-Hybrid-Zero
Text Generation • 4B • Updated • 338 • 1 -
OctoThinker/OctoThinker-1B-Long-Zero
Text Generation • 1B • Updated • 9 -
OctoThinker/OctoThinker-1B-Hybrid-Zero
Text Generation • 1B • Updated • 9 -
OctoThinker/OctoThinker-1B-Short-Zero
Text Generation • 1B • Updated • 10 -
OctoThinker/Llama3.2-3B-Zero
4B • Updated • 3 -
OctoThinker/OctoThinker-3B-Long-Zero
Text Generation • 4B • Updated • 75 -
OctoThinker/OctoThinker-1B-Long-Base
Text Generation • 1B • Updated • 12 -
OctoThinker/OctoThinker-1B-Short-Base
Text Generation • 1B • Updated • 219 -
OctoThinker/OctoThinker-1B-Hybrid-Base
Text Generation • 1B • Updated • 11 -
OctoThinker/OctoThinker-3B-Long-Base
Text Generation • 3B • Updated • 6.27k -
OctoThinker/OctoThinker-3B-Hybrid-Base
Text Generation • 3B • Updated • 411 -
OctoThinker/OctoThinker-3B-Short-Base
Text Generation • 3B • Updated • 1.6k