Models from Communicate the Change: Efficient Expert Parallelism via Activation Compression (EchoMoE).
-
McGill-NLP/Qwen3-30B-A3B-SFT-EchoMoE-teacher
Text Generation • 31B • Updated • 251 -
McGill-NLP/EchoMoE-Qwen3-30B-A3B-s4-r512-c768
Text Generation • 28B • Updated • 174 -
McGill-NLP/EchoMoE-Qwen3-30B-A3B-s4-r1280-c1280
Text Generation • 36B • Updated • 175 -
McGill-NLP/EchoMoE-Qwen3-30B-A3B-s6-r1280-c1280
Text Generation • 37B • Updated • 174