- Organizations
- Adobe
Adobe: AI benchmarks & research
Software company. Adobe Research maintains NoLiMa.
Maintained by Adobe
4 benchmarks- NoLiMa
NoLiMa (No Literal Matching) is a long-context benchmark extending needle-in-a-haystack tests with minimal lexical overlap between questions and needles, requiring models to infer latent associations rather than relying on surface-level matching. Published at ICML 2025.
- Leader
- No scores yet
- Models scored
- 0
- NoLiMa 128K
NoLiMa evaluated at a 131072-token context length. Tests latent associative reasoning in long contexts with minimal lexical overlap between questions and needles.
- Leader
MiniCPM-SALA23.9%- Models scored
- 1
- NoLiMa 32K
NoLiMa evaluated at a 32768-token context length. Tests latent associative reasoning in long contexts with minimal lexical overlap between questions and needles.
- Leader
MiniCPM-SALA54.5%- Models scored
- 1
- NoLiMa 64K
NoLiMa evaluated at a 65536-token context length. Tests latent associative reasoning in long contexts with minimal lexical overlap between questions and needles.
- Leader
MiniCPM-SALA43.0%- Models scored
- 1