armenjeddi/LoopFormer-3block-8iterations-FineWeb300K
0.3B
•
Updated
•
37
LoopFormer
Models trained in the ICLR2026 paper: LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation
-
-
armenjeddi/NanoGPT-Base24-FineWeb300K
1B • Updated • 19 -
armenjeddi/LoopFormer-3block-8iterations
0.3B • Updated • 365 -
armenjeddi/LoopFormer-2block-12iterations
0.2B • Updated • 95 -
armenjeddi/LoopFormer-1block-24iterations
0.2B • Updated • 132 -
armenjeddi/NanoGPT-Base24
1B • Updated • 87 -
armenjeddi/NanoGPT-Base12
0.6B • Updated • 72 -
armenjeddi/NanoGPT-Base6
0.3B • Updated • 56 -
armenjeddi/NanoGPT-Base3
0.2B • Updated • 30 -
armenjeddi/Base-Loop-3block-8iterations
0.2B • Updated • 76 -
armenjeddi/Base-Loop-EE-3block-8iterations
0.2B • Updated • 120 -
armenjeddi/Base-Loop-EE-2block-12iterations
0.2B • Updated • 85 -
armenjeddi/Base-Loop-EE-1block-24iterations
0.1B • Updated • 90 -
armenjeddi/TMLT-3block-8iterations
0.3B • Updated • 45 -
armenjeddi/TMLT-EE-3block-8iterations
0.3B • Updated • 68 -
armenjeddi/TMLT-EE-2block-12iterations
0.2B • Updated • 67 -
LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation
Paper • 2602.11451 • Published • 15