Stanford SNAP open-sources Memorilla memory module for frozen LLMs
Start here
Stanford SNAP Group researchers have open-sourced Memorilla, a 194M-parameter trainable memory module that compresses a document collection into 16 memory tokens for a frozen Qwen3-8B. Code, training recipes and checkpoints are public under Apache 2.0; the quickstart needs one 80 GB GPU and the first run downloads about 25 GB of models. On FactKG the module scored 84.33% versus 80.43% for full-context prompting while using 108x fewer document tokens, and it posted the highest reported average of 49.68 across 8 benchmarks. It trailed LLMLingua-2 on PubMedQA (59.00% versus 69.00%), where the abstracts total only 324 tokens. The full training recipe takes about 19 hours on 4 H100 GPUs. The team also released PersonalizationV4, a dataset of 149 synthetic users with 200 conversations each, under CC BY 4.0.
Generated from reports · updated 1 h ago
Report timeline
Follow the reports to see the event from different sides.
- MarkTechPostStanford SNAP Group Open-Sources Memorilla, a 16-Token Latent Memory Module for Frozen LLMs
Stanford researchers have open-sourced Memorilla, a 194M-parameter trainable memory module that compresses an entire document collection into 16 memory tokens for a frozen Qwen3-8B, with code and checkpoints under Apache 2.0. On FactKG it scored 84.33% versus 80.43% for full-context prompting while using 108x fewer document tokens, and it posted the highest reported average of 49.68 across 8 benchmarks. It loses on PubMedQA (59.00% versus 69.00% for LLMLingua-2), where the abstracts total only 324 tokens, and the full training recipe takes about 19 hours on 4 H100 GPUs.
Heat for this event
Not enough continuous observations yet to draw a trend.