BenchCouncil

[ICML 2026] Official implementation of "CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models"

[ICML 2026] Implementation of "Beyond Benchmarks: Toward Causally Faithful Evaluation of Large Language Models"

[AAAI 2026] Official implementation of "TimeMosaic: Information-Density Guided Time Series Forecasting via Adaptive Granularity Patch and Segment-wise Decoding"

Multi-language CPU benchmarking suite for reproducible, parameterized performance across Python, C, C++, Go, and Java — with configurable data scales and thread counts to exercise both single‑core and multi‑core performance on emerging workloads.

[ACM MM 2025] Official implementation of "DualSG: A Dual-Stream Explicit Semantic-Guided Multivariate Time Series Forecasting Framework"

[VLDB 2025] BigVectorBench advances vector database benchmarking by defining and evaluating the embedding performance of heterogeneous data and abstracting compound queries, which can be multimodal or single-modal with fine-grained restrictions, for real-world applications.