Video-LLM

Paper: StreamKV: Streaming Video Question-Answering with Segment-based KV Cache Retrieval and Compression Code: sou1p0wer/StreamKV Background Streaming video question-answering (StreamingVQA) requires a model to continuously process incoming video, preserve useful historical context, and answer questions online with low latency. ReKV showed that video QA can be reformulated as retrieve relevant KV caches first, then answer with the retrieved KV. But it still has several weaknesses: It uses uniform segmentation, which may cut through semantic boundaries. It keeps essentially the whole historical visual context, so memory usage is still large. Its retrieval strategy is not flexible enough, especially when the useful information is distributed differently across layers. Core Idea StreamKV extends the ReKV line in two directions at the same time: ...

Series guide. This note audits one method in detail. Start with the development-history hub for the field-level map, or use the pipeline survey to place ReKV’s write, index, and read stages in a complete system. The bounded-memory survey compares its growing exact-KV archive with later fixed-state methods. Paper: Streaming Video Question-Answering with In-context Video KV-Cache Retrieval Proceedings: ICLR 2025 Preprint: arXiv:2503.00540 Code: Becomebright/ReKV ReKV was first public on arXiv on 2025-03-01 and was published at ICLR 2025. The paper, appendix, and official code were rechecked on 2026-09-03. Unless a sentence is explicitly marked as my interpretation, mechanism and measurement claims below describe the authors’ paper or released code. ...