Counting-Stars (★)
-
Updated
Nov 24, 2025 - Jupyter Notebook
#
needle-in-a-haystack
Here are 4 public repositories matching this topic...
[NAACL 2025 Oral] Multimodal Needle in a Haystack (MMNeedle): Benchmarking Long-Context Capability of Multimodal Large Language Models
-
Updated
Apr 22, 2026 - Python
Top-K sparse attention has no critical key budget: a 4× swing of k_eff barely moves long-context retrieval accuracy across 3 models (Llama-1B/3B, Qwen2.5-3B). The limit is the base model's disambiguation, not the compressor. Paper + raw per-prompt logs + pre-registrations. Selection is exact; kernel port validated bitwise.
retrieval reproducible-research transformers triton attention llama inference-optimization kv-cache sparse-attention llm long-context falsification mechanistic-interpretability qwen needle-in-a-haystack
-
Updated
Jun 2, 2026 - TeX
Semantically hard multi-needle long-context data generator. Stop testing LLMs with random-password needles.
-
Updated
Apr 29, 2026 - Python
Improve this page
Add a description, image, and links to the needle-in-a-haystack topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the needle-in-a-haystack topic, visit your repo's landing page and select "manage topics."