#long-context

#…

2 posts

StateBench: Transformers Can't Track State as Well as You Think

AllenAI's new benchmark exposes exactly where transformer models fail at tracking information over long contexts—and shows recurrent architectures might finally have their moment.

#benchmarks #transformers #long-context #architectures #state-tracking

LFM2.5-Encoders: Liquid AI Ships CPU-Fast Encoders That Actually Scale to Long Context

Liquid AI just released two encoder models that beat larger competitors on classic NLP benchmarks while running 3.7× faster than ModernBERT on CPU at 8k tokens. Here's why that matters.

#encoders #cpu-inference #long-context #bert #liquid-ai

Loading…