Stanford open-sources frozen LLM memory module Memorilla
Understand this story
A Stanford research team has open-sourced Memorilla: a 194M-parameter trainable memory module that compresses an entire document collection into 16 memory tokens and injects them into a frozen Qwen3-8B, giving the model external memory without changing the original weights. The code, training recipe, and checkpoints are released under the Apache 2.0 license. Regarding usage requirements, Memorilla runs on Linux, CUDA GPUs, and Python 3.11; getting started quickly requires a GPU with 80GB of VRAM (the training recipe was run on an H100 80GB). The report did not disclose a quantized version.
AI overview based on source reports · Updated 48 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- MarkTechPostStanford Research Open-Sources Memorilla, a Latent Memory Module for Frozen LLMs
Stanford researchers have released Memorilla, an open-source 194M-parameter memory module that condenses a whole document collection into 16 memory tokens for a frozen LLM, with code, training recipes and checkpoints public under Apache 2.0. On FactKG, a frozen Qwen3-8B scored 84.33% with Memorilla versus 80.43% for full context, using 108x fewer document tokens, and it reported the highest average (49.68) across 8 benchmarks while losing on PubMedQA (59.00% vs 69.00% for LLMLingua-2). The quickstart requires one 80 GB CUDA GPU, and the full training recipe takes about 19 hours on 4 H100 GPUs.
Activity for this story
Not enough continuous observations to show a trend yet.