Semantic Search research benchmarks

Benchmarks

Dataset NameSOTA methodMetricTrendResultsLast Updated

Retail dataset search collection

LLM semantic search tool

244Dataset Count

Feb 26, 2026

Dense-Set

ours

42.5Recall@5

May 8, 2026

Holy Mosque Dataset

ours

41Recall@5

May 8, 2026

MS-COCO (test)

SigLIP

74.6R@5

May 8, 2026

arXiv

GSS

61.2nDCG@10

Mar 13, 2026

Medium-scale environment

SAGR

207.1Average Completion Time (steps)

Apr 20, 2026

HM3D Large

SAGR

228Average Completion Time (steps)

Apr 20, 2026

HM3D Medium

SAGR

190.4Average Completion Time (steps)

Apr 20, 2026

HM3D Small

SAGR

178.8Average Completion Time (steps)

Apr 20, 2026

Amazon ESCI

Fine-tuned Sentence Transformer & Fine-tuned Flan-T5-Small (ours)

32Precision@1

Feb 26, 2026

Downstream Task Evaluation

MiniLM

94.7Score

Mar 10, 2026

SPS Exploratory queries (held-out golden set)

Unified SLM

75Relative Change in GR@10

May 28, 2026

SPS Navigational queries (held-out golden set)

Unified SLM

57GR@10 Relative Change

May 28, 2026