Persistent in-model memory w/MemT - running on ExLlamav3
retained text
[https://huggingface.co/MemTensor/MemT-1.0-Nano](https://huggingface.co/MemTensor/MemT-1.0-Nano) was released 3 days ago, with a live-updating memory model. This model has the capability to continuously lean in conversation, and carry it onto the next instance. The above video demonstrates a short conversation followed by a /commit command, which stores the previous conversations important facts to be preserved in the memory module, while wiping the context of the conversation. But upon benig asked to recall speicific questions, it still succeed. I've implemented it on my fork of exllamav3, [exl3-qlora](https://github.com/UnstableLlama/exl3-qlora), and it works natively with [tabbyAPI](https://github.com/theroyallab/tabbyAPI/). To ingest memories simply give a /commit command from within tabby every few messages (or whenver you want to.) Quantized versions of the model are available [here](https://huggingface.co/UnstableLlama/MemT-1.0-Nano-exl3).
document fetched 2026-10-23 00:46 · canonical https://reddit.com/r/LocalLLaMA/comments/1xcpnq5/persistent_inmodel_memory_wmemt_running_on