tiers
Datasets
All datasets matching “tiers”math-tiers
Math-Tiers: A Tiered Pretraining Corpus for Studying Numerical Reasoning
A large-scale English pretraining corpus split into three tiers by mathematical content density. Designed for controlled experiments studying how data composition during pretraining affects numerical reasoning in language models.
Tiers
Tier
Description
Shards
Size
Est. Tokens
Sources
T0
Pure narrative: no digits, number words, or math
648
542 GB
~113B
RedPajama-Book, PleIAs/English-PD… See the full description on the dataset page: https://huggingface.co/datasets/swan07/math-tiers.math-tiersdataqual-tiers
explcre/dataqual-tiers
Tri-modal (text / TikZ code / rendered image) data-quality tiers + diagram taxonomy produced by the
dataqual pipeline. Each row is one scraped (description, TikZ) pair with:
Tiers (sft_tier, sft_grade/sft_score, rl_tier, curriculum_tier) — usability for SFT
(text→code imitation), RL (prompt→code, DSV-verifiable reward), and difficulty-ordered curriculum.
Taxonomy (family → subcategory structural; math_domain → math_topic semantic) — consistent
2-level… See the full description on the dataset page: https://huggingface.co/datasets/explcre/dataqual-tiers.performance-tiers
Performance Tiers
Models grouped by speed:
Ultra Fast (30+ t/s): 1 models
Fast (15-30 t/s): 9 models
Moderate (5-15 t/s): 16 models
Slow (<5 t/s): 5 models
🚀 dispatchAI
sequence_with_tiersadaption-rit1-vus-research-tiers
This dataset is a remastered version prepared using Adaption's Adaptive Data platform.
adaption-rit1_vus_research_tiers
This dataset contains research-level prioritization assessments for RIT1 missense variants associated with Noonan Syndrome 8 (NS8). Each entry provides a structured analysis including source-derived investigation tiers, scores, and evidence summaries based on computational predictors, population frequency, and structural domain context. The data explicitly… See the full description on the dataset page: https://huggingface.co/datasets/Fernandosr85/adaption-rit1-vus-research-tiers.