Comment by ashirviskas
2 months ago
It's closer to <30k before performance degrades too much for 3.5/3.7. 200k/64k is meaningless in this context.
2 months ago
It's closer to <30k before performance degrades too much for 3.5/3.7. 200k/64k is meaningless in this context.
Is there a benchmark to measure real effective context length?
Sure, gpt-4o has a context window of 128k, but it loses a lot from the beginning/middle.
Here's an older study that includes Claude 3.5: https://www.databricks.com/blog/long-context-rag-capabilitie...?
ruler https://arxiv.org/abs/2404.06654
nolima https://arxiv.org/abs/2502.05167
They often publish "needle in a haystack" benchmarks that look very good, but my subjective experience with a large context is always bad. Maybe we need better benchmarks.