junwatu/repro-dtop-p-moe-sparsity-controlled-dynamic-top-p-moe-for-foundation-model-pre-training
0
Reproduction: DTop-p MoE: Sparsity-Controlled Dynamic Top-p MoE for Foundation Model Pre-training
An open experiment logbook, published with Trackio.
An open experiment logbook, published with Trackio.