Generalized Neural Memory (NeurIPS 2026)
Collection
Models, benchmarks, and eval records for 'Tell Me What To Learn' (NeurIPS 2026). • 9 items • Updated
Generalized Neural Memory — GNM ablation (CounterFACT)
The frozen-memory-write ablation of the CounterFACT GNM: identical training except gradients are blocked through the memorize step, isolating what is gained by learning what to write versus what to read.
From "Tell Me What To Learn: Generalizing Neural Memory to be Controllable in Natural Language" (NeurIPS 2026).
Usage
Use through the code repo:
python scripts/download_models.py --counterfact # places this model at models/gnm_counterfact_ablation/model.pth
Registry key in the repo: memoryllm_chat. See the repo README for the full CounterFACT
evaluation commands (Table 1) and training configs.
- Downloads last month
- 471
Model tree for maxbennett/generalized-neural-memory-gnm-counterfact-ablation
Base model
meta-llama/Meta-Llama-3-8B-Instruct