datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Semantic-Flow-Dynamics-SFD
Semantic Flow Dynamics (SFD) — A Formally Specified Social-Science Theory Corpus
TL;DR: 614 Chinese-language formalized social-science concepts across 25
papers, UUID-linked with typed derivation relations (derives_from,
leads_to, falsified_by, …) — usable for knowledge-graph construction,
RAG over structured theory, or as a Chinese formal-reasoning corpus.
Author: 黃正宇 Cheng Yu HuangContact: mthree.tw@gmail.com
What This Dataset Is
This corpus is an ongoing… See the full description on the dataset page: https://huggingface.co/datasets/mthreetw/Semantic-Flow-Dynamics-SFD.groundtruth-dynamic-benchmarking
Groundtruth Dynamic Benchmarking — Geology
Question sets and grading rubrics for evaluating LLMs on real-world geological
reasoning. Every question is authored from a real source corpus, and every
claim in the grading key carries an evidence locator back to that corpus —
nothing is synthetic. Licensing/redistribution status varies by corpus — see
License.
This dataset holds the questions, grading rubrics, and source corpora.
Running an evaluation (generating answers from a model… See the full description on the dataset page: https://huggingface.co/datasets/EigenformAI/groundtruth-dynamic-benchmarking.fresh-swe-pro-dynamic
Dataset Card
Dataset Description
Fresh SWE-Pro Dynamic is a source-verified, Docker-executable benchmark for software-engineering agents. The public release contains five recent repository tasks with pinned base commits, problem statements, gold patches, regression tests, Docker images, source provenance, and validation evidence.
Task: software-engineering agent evaluation and patch generation
Language: English
Public size: 5 instances
Release: 2026.09
Source… See the full description on the dataset page: https://huggingface.co/datasets/LianeMarilin/fresh-swe-pro-dynamic.update-interrupt-benchmark
Update-Driven Math & Code Interrupt Datasets
Paper: Are Large Reasoning Models Interruptible?
Authors: Tsung-Han Wu*, Mihran Miroyan*, David Chan, Trevor Darrell, Narges Norouzi, Joseph Gonzalez
Project page: https://dynamic-lm.github.io/
Github: https://github.com/dynamic-lm/interrupt-lrm
This dataset page contains the update-driven interrupt subsets for math (GSM8K, MATH500, AIME) and coding (LiveCodeBench) problems. For both splits, we revise the source problems and… See the full description on the dataset page: https://huggingface.co/datasets/dynamic-lm/update-interrupt-benchmark.Dynamically-Generated-Hate-Speech-Dataset
Dataset Card for dynamically generated hate speech dataset
Dataset Summary
This is a copy of the Dynamically-Generated-Hate-Speech-Dataset, presented in this paper by
Bertie Vidgen, Tristan Thrush, Zeerak Waseem and Douwe Kiela
Original README from GitHub
Dynamically-Generated-Hate-Speech-Dataset
ReadMe for v0.2 of the Dynamically Generated Hate Speech Dataset from Vidgen et al. (2021). If you use the dataset, please cite our paper in the… See the full description on the dataset page: https://huggingface.co/datasets/LennardZuendorf/Dynamically-Generated-Hate-Speech-Dataset.DynamicPO-Data
DynamicPO-Data
This repository contains the processed datasets for the paper DynamicPO: Dynamic Preference Optimization for Recommendation.
Official Code: xingyuHuxingyu/DynamicPO
Dataset Summary
DynamicPO is a plug-and-play dynamic preference optimization framework for LLM-based recommender systems. This repository provides the processed data used to evaluate the framework, following the construction pipeline of prior works like LLaRA and S-DPO.
The collection… See the full description on the dataset page: https://huggingface.co/datasets/xingyuHuxingyu/DynamicPO-Data.stratasynth-belief-dynamics
StrataSynth Belief Dynamics
Part of the StrataSynth Synthetic Identity Engineering corpus.
2,114 turns · 100 conversations · 23 columns per turn
The most psychologically demanding dataset in the corpus. Grief, chronic illness, career crisis — scenarios where beliefs are under maximum and sustained pressure. The belief_resolution field drops measurably across pure_conflict arcs and recovers in reconnection arcs. Every trajectory is causal, not random.
Complexity level: 5 —… See the full description on the dataset page: https://huggingface.co/datasets/StrataSynth/stratasynth-belief-dynamics.weibo-opinion-dynamic-single-dim
Weibo Sentiment Evolution Dataset
This dataset contains Weibo posts and their associated comment threads used for studying sentiment evolution and opinion dynamics in social media discussions.
The dataset is distributed as a single JSON Lines file:
weibo_dataset.jsonl
Each line is one Weibo post record. Comments for that post are embedded in the comments field.
Dataset Details
Number of post records: 1,379
Number of embedded comments: 93,569
Number of Weibo… See the full description on the dataset page: https://huggingface.co/datasets/hreyulog/weibo-opinion-dynamic-single-dim.dynamic_sonnet_llama3
Dynamic Sonnet - Llama3
Curated dataset for benchmarking LLM serving systems
In real-world service scenarios, each request comes with varying input token lengths.
Some requests generate only a few tokens, while others produce a significant number.
Traditional fixed-length benchmarks fail to capture this variability, making it difficult to accurately assess real-world throughput performance.
This dynamic nature of input token lengths is crucial as it directly affects key features of… See the full description on the dataset page: https://huggingface.co/datasets/squeezebits/dynamic_sonnet_llama3.dynamic_sonnet_llama2
Dynamic Sonnet - Llama2
Curated dataset for benchmarking LLM serving systems
In real-world service scenarios, each request comes with varying input token lengths.
Some requests generate only a few tokens, while others produce a significant number.
Traditional fixed-length benchmarks fail to capture this variability, making it difficult to accurately assess real-world throughput performance.
This dynamic nature of input token lengths is crucial as it directly affects key features of… See the full description on the dataset page: https://huggingface.co/datasets/squeezebits/dynamic_sonnet_llama2.dynamics-reasoning-traces-sample
DYNAMICS-8 Behavioural Reasoning Traces
Personality-conditioned chain-of-thought reasoning data for LLM alignment and persona fine-tuning.
What This Dataset Contains
Each record is a first-person behavioural response from a synthetic persona with a validated 8-dimension personality profile (DYNAMICS-8), accompanied by a structured reasoning trace showing which personality dimensions drove the decision.
This is not survey data. It is not statistical synthetic data. Each… See the full description on the dataset page: https://huggingface.co/datasets/Kronaxis/dynamics-reasoning-traces-sample.Dynamic-Topic-RedPajama-Data-1T-100k-SubSample-max-1k-tokens
Dynamic Topic Modeling Dataset: RedPajama-1T SubSample (100k samples, 1k tokens)
📝Check out the Blog Post
This dataset represents a curated subset of the RedPajama-1T Sample dataset, specifically processed for dynamic topic modeling applications. It contains 100,000
samples from the original dataset, with each document limited to the first 1,024 tokens for consistent processing.
Dataset Overview
Name:… See the full description on the dataset page: https://huggingface.co/datasets/AmanPriyanshu/Dynamic-Topic-RedPajama-Data-1T-100k-SubSample-max-1k-tokens.clinical_alignment_recovery_dynamics_v0.1Clinical Alignment Recovery Dynamics
Measures whether a model corrects earlier clinical errors when new signals appear.
Output JSON
recovered
recovery_type
correct_action
Runpython scorer.py --predictions predictions.jsonl --test_csv data/test.csv
alignment_recovery_dynamics_v01Clarus Alignment Recovery Dynamics v0.1
This dataset measures recovery after an alignment flip.
Focus
Not only whether a system flips
But whether it can recover
And whether it relapses under renewed pressure
Design
One row per step
Steps form a trajectory grouped by case_id
A recovery window defines how quickly recovery must occur
Columns
flip_signal_expected
none, early_warning, flip, cascade
first_flip_step_expected
First step where a flip is expected, or -1
recovery_expected
true if… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/alignment_recovery_dynamics_v01.dynamics-of-instruction-tuning
DoIT: Dynamics of Instruction Tuning
DoIT is a collection of over 40k human-curated instruction-output pairs in Chinese. I created from https://huggingface.co/datasets/ChiyuSONG/dynamics-of-instruction-tuning.
It collects all data in dynamics-of-instruction-tuning/curated/full/*.json.