regime
Datasets
All datasets matching “regime”p2-susy-regime-resultsdaily-paper-2026-08-05-traffic-regime-serving-breakeven
Traffic-Regime-Dependent Break-Even Points for Speculative Decoding, Prefix Caching, and Precision Quantization in B200 LLM Serving
TL;DR — For MoE LLM serving on B200, which precision (FP16/FP8/NVFP4) minimizes cost-per-token depends on concurrency; n-gram speculative decoding only pays off above a repetitiveness threshold that rises monotonically as precision is lowered; prefix caching break-even threshold rises with concurrency. These three break-even surfaces interact and… See the full description on the dataset page: https://huggingface.co/datasets/thaki-AI/daily-paper-2026-08-05-traffic-regime-serving-breakeven.EURUSD-15M-RegimeBench
EURUSD-15M-RegimeBench
A labeled EUR/USD 15-minute market regime dataset designed for quantitative finance, machine learning, algorithmic trading, and time-series research.
The dataset contains 35 engineered features spanning trend, volatility, directional movement, session behavior, and multi-timeframe market structure. Each observation is annotated with market state and regime labels, enabling supervised learning experiments for market classification and forecasting.… See the full description on the dataset page: https://huggingface.co/datasets/akashkumar5/EURUSD-15M-RegimeBench.nifty50_market_regime
NIFTY 50 Market Regime (TsFile)
Apache TsFile version of AAdevloper/nifty50-market-regime.
Overview
Technical indicators and corresponding market-regime labels for the NIFTY 50
index, used to train binary classification models that predict market regimes
(RISK_ON / RISK_OFF) in Indian financial markets. Each record is one daily
observation with derived indicators (VIX, RSI, moving averages) plus the regime
label.
Regimes: RISK_ON (1) — favorable conditions, lower… See the full description on the dataset page: https://huggingface.co/datasets/THULab/nifty50_market_regime.regime-bench
RegimeBench Base Data Bundle
This dataset contains the base parquet inputs for the 18 public RegimeBench signal tasks. Split variants are generated locally from these base files by the benchmark repository rather than stored as a separate hosted payload.
The data bundle is released under the Apache License 2.0, matching the benchmark source release. It contains processed benchmark data for research and reproducibility, not investment advice. Hidden/OOS labels are included, so… See the full description on the dataset page: https://huggingface.co/datasets/amitvpatel06/regime-bench.gemma4-quant-regime-study
Gemma 4 Quantization-Regime Study (QAT Q4_0 vs PTQ Q4_K_M)
A same-day 2×2 study (quantization regime × released model pair) of
governed routing quality for Gemma 4 12B IT (dense; a clean same-base
pair) and a 26B-class MoE released pair on one RTX 5070 Ti, run under a
fixed, bit-identical governance stack via Ollama.
Companion/sequel to the Gemma 4 MTP Quality–Throughput Study (2026-08-09,
DOI 10.5281/zenodo.21860461).
Author: Taiko Toeda, Independent Researcher
ORCID:… See the full description on the dataset page: https://huggingface.co/datasets/moebiusT7/gemma4-quant-regime-study.