Scaling Context, Not Parameters: Training a Compact 7B Language Model for Efficient Long-Context Processing

MoE
LRM
Re-assign community