Tagged “mixture of experts”
Models
Microsoft's MAI-Base-2: A 2-Trillion-Parameter Model With an Unspecified 12x Efficiency Claim [1]
Microsoft reports a 2T-parameter base model and up to 12x hardware efficiency, but publishes no metric, scores or test conditions. Here is what to ask before planning around it.
Models
Beam's 501B Parameters, 23B Active, No Published Benchmarks
Hugging Face's Beam launch gives parameter counts, formats and an Apache 2.0 label, but no benchmarks, hardware configurations or pricing. Here is what to verify before a pilot.