tiered-memory-system-managed-by-model-itself-sustainsmechanismsingle paper
A tiered memory system managed by the model itself sustains recall over conversations far longer than the context window.
Capability: Remembering across sessions
Sources
- A tiered memory system managed by the model itself sustains recall over conversations far longer than the context window.
Disagreeing is the most useful thing you can do here. Both sides of every contested claim in this catalog were assembled by the same person, which is its weakest point.
Related claims
- Managing context explicitly, paging information in and out of a bounded window, sustains performance on tasks that exceed the window.Keeping its own context clean
- Recursively summarizing the dialogue into a running memory improves consistency in long conversations.Remembering across sessions
- Most models claiming long contexts fail well before their advertised length on synthetic retrieval, tracing and aggregation tasks.Losing information in long inputs
- Holding the task fixed and only lengthening the input degrades reasoning long before the context limit is reached.Losing information in long inputs
- For streaming video question answering with a frozen video-LLM, appending retrieved historical frames as extra visual context helps less than optimizing a small set of latent memory tokens at test time and then dropping the retrieved tokens before decoding; the gain is largest on questions requiring backward tracing over past history.Reasoning about time in video · unreviewed