datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
pal-resultscis5300-language-models
CIS 5300 Language Models Dataset
Dataset for Homework 3 of CIS 5300 (Natural Language Processing) at Penn.
Cities config
Country-of-origin classification over short city-name strings, drawn from
nine countries (Afghanistan, China, Germany, Finland, France, India, Iran,
Pakistan, South Africa).
from datasets import load_dataset
cities = load_dataset("CCB/cis5300-language-models", "cities")
Split
Rows
Has labels?
train
12,392
yes
validation
1,548
yes
test
1… See the full description on the dataset page: https://huggingface.co/datasets/CCB/cis5300-language-models.nepalitext-language-model-dataset
Dataset Card for "nepalitext-language-model-dataset"
Dataset Summary
"NepaliText" language modeling dataset is a collection of over 13 million Nepali text sequences (phrases/sentences/paragraphs) extracted by combining the datasets: OSCAR , cc100 and a set of scraped Nepali articles on Wikipedia.
Supported Tasks and Leaderboards
This dataset is intended to pre-train language models and word representations on Nepali Language.
Languages
The data is… See the full description on the dataset page: https://huggingface.co/datasets/Sakonii/nepalitext-language-model-dataset.core-language-model-data
Core Language Model
A small coding + general-purpose language model built from scratch in pure NumPy — no PyTorch, no JAX, no autograd library. Every piece (autograd engine, tokenizer, attention, layer norm) is hand-written and gradient-checked.
This is a from-first-principles build of a transformer language model, aimed at understanding every formula well enough to reimplement it without reference.
Why pure NumPy?
The goal isn't a competitive model — it's a fully… See the full description on the dataset page: https://huggingface.co/datasets/karthik-2905/core-language-model-data.Anthropic_HH_Golden
Dataset Card for Anthropic_HH_Golden
This dataset is constructed to test the ULMA technique as mentioned in the paper Unified Language Model Alignment with Demonstration and Point-wise Human Preference (under review, and an arxiv link will be provided soon). They show that replacing the positive samples in a preference dataset by high-quality demonstration data (golden data) greatly improves the performance of various alignment methods (RLHF, DPO, ULMA). In particular, the ULMA… See the full description on the dataset page: https://huggingface.co/datasets/Unified-Language-Model-Alignment/Anthropic_HH_Golden.language_model_frWhat-Gradients-Add-to-Text-Leakage-in-Split-Language-Models-Counted-per-Token-and-per-Document
Paper A: reproduction and peer-review release
This dataset contains the final paper, the evidence used for its three experiments, and the historical material explicitly discussed in the paper.
Final paper
The current branch includes the 2 October 2026 punctuation revision of the 26-page manuscript. Its scientific content and experimental evidence are unchanged. Tag v1.0 retains the frozen release and its original manuscript; the updated PDF and source package are… See the full description on the dataset page: https://huggingface.co/datasets/Setloop/What-Gradients-Add-to-Text-Leakage-in-Split-Language-Models-Counted-per-Token-and-per-Document.Towards-Joint-Modeling-of-Dialogue-Response-and-Speech-Synthesis-based-on-Large-Language-ModelSpatial-Blind-Spots-in-Vision-Language-Modelslicense: mit
model_evaluated:
name: Qwen3-VL-2B-Instruct
url: https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct
evaluation_notebook:
https://www.kaggle.com/code/wajidhassanmoosa/blind-spot-qwen3-2b
evaluation_setup: |
The model evaluated in this study is Qwen3-VL-2B-Instruct.
Evaluation was conducted using the Hugging Face Transformers library
with automatic device mapping (device_map="auto") and "bfloat16" dtype selection.
For each example:
The image was provided as part of a… See the full description on the dataset page: https://huggingface.co/datasets/hassan-wajid/Spatial-Blind-Spots-in-Vision-Language-Models.nepalitext-language-model-dataset
Dataset Card for "nepalitext-language-model-dataset"
Dataset Summary
"NepaliText" language modeling dataset is a collection of over 13 million Nepali text sequences (phrases/sentences/paragraphs) extracted by combining the datasets: OSCAR , cc100 and a set of scraped Nepali articles on Wikipedia.
Supported Tasks and Leaderboards
This dataset is intended to pre-train language models and word representations on Nepali Language.
Languages
The data is… See the full description on the dataset page: https://huggingface.co/datasets/Arpuuu/nepalitext-language-model-dataset.multimodal-vision-language-video-models-2026
👁️ Multimodal Vision-Language & Video Foundation Models Dataset (2026 Edition)
A structured research dataset featuring 1,000 domain-verified research papers and code repositories focused on Multimodal Vision-Language Models (VLM), Video Foundation Models, Diffusion Transformers (DiT), Visual Grounding, and World Simulators.
Built with Universal Scientific Engine V15.1 Gold, providing 47 schema attributes with verified repository attribution, modality capability matrix, vision… See the full description on the dataset page: https://huggingface.co/datasets/beatsprom/multimodal-vision-language-video-models-2026.languagemodelsforRNNdecompositionThis repository is for the paper "Decomposing a Recurrent Neural Network into Modules for Enabling Reusability and Replacement". To use the data, there are two directories:
language datasets: Contains the necessary Tatoeba files used for the experiments. We have experimented with 4 languages(English, French, Italian and German).
language_models: Contains all trained language models and scripts to train them. It's organized in this way: language_models/{X}: contains language models for X… See the full description on the dataset page: https://huggingface.co/datasets/astha/languagemodelsforRNNdecomposition.repro-how-much-can-language-models-memorize-traces
Agent traces
Agent sessions published from a Trackio Logbook.
Baize-TCM-Corpus-for-Large-Language-Models-V3
白泽中医药大模型语料库
版本：3.0语料数量：157,438 条语言：中文领域：中医药（Traditional Chinese Medicine, TCM）格式：问答对（QA Pair）用途：中医药大模型训练、知识问答系统、语义理解研究
📚 简介
“白泽中医药大模型语料库”是一个专注于中医药领域的高质量问答语料集合，旨在支持中医药知识的数字化、智能化应用。语料库共包含 157,438 条 经过整理与校对的问答对，涵盖中医基础理论、中药学、方剂学、诊断学、针灸推拿、经典医籍、临床实践等多个子领域。
本语料库可广泛应用于：
中医药大语言模型的预训练与微调
智能问答系统开发
医学自然语言处理任务（如实体识别、关系抽取）
中医药知识图谱构建
🧩 数据内容
每条语料为一个标准的问答对，格式如下：
{
"instruction": "广义转录组和狭义转录组在定义上的主要区别是什么？",
"input": "",
"output":… See the full description on the dataset page: https://huggingface.co/datasets/DigitalIntelligenceCenter-of-ICMM/Baize-TCM-Corpus-for-Large-Language-Models-V3.language_models_lab_2ShareRobotlanguage_modelsc_corpus_br_finetuning_language_model_bert
Dataset Card for "c_corpus_br_finetuning_language_model_bert"
More Information needed
repro-prototype-transformer-towards-language-model-architectures-interpretable-by-design-traces
Agent traces
Agent sessions published from a Trackio Logbook.
lener_br_finetuning_language_model
Dataset Card for "LeNER-Br language modeling"
Dataset Summary
The LeNER-Br language modeling dataset is a collection of legal texts in Portuguese from the LeNER-Br dataset (official site).
The legal texts were downloaded from this link (93.6MB) and processed to create a DatasetDict with train and validation dataset (20%).
The LeNER-Br language modeling dataset allows the finetuning of language models as BERTimbau base and large.
Language
Portuguese from… See the full description on the dataset page: https://huggingface.co/datasets/pierreguillou/lener_br_finetuning_language_model.c_corpus_br_finetuning_language_model_deberta
Dataset Card for "c_corpus_br_finetuning_language_model_deberta"
More Information needed
ilm_deRole-of-Provider-on-Safety-Alignment-in-Large-Language-Models
Evaluating the Role of Provider on Safety Alignment in Large Language Models: dataset
Data for the paper
Naser, M.Z. (2026). Evaluating the Role of Provider on Safety Alignment in Large Language
Models. Neurocomputing, 135173. https://doi.org/10.1016/j.neucom.2026.135173
It holds the Extended Context Safety Benchmark (ECSB) scenario bank and every trial result.
If you use the data, please cite the paper (BibTeX under Citation).
The metadata.paper field inside… See the full description on the dataset page: https://huggingface.co/datasets/mznaser/Role-of-Provider-on-Safety-Alignment-in-Large-Language-Models.Bangla_Masked_Language_Model_dataset_preprocessedilm_espiqa_es
Dataset Card for PIQA (Spanish Version)
Dataset summary
This dataset provides the Spanish translation and adaptation of the validation
set of PIQA (Physical Interaction: Question Answering). The original dataset
was designed to evaluate physical commonsense reasoning in language models
through questions about everyday situations. Each example presents a physical
goal and two possible solutions, only one of which is correct.
This Spanish adaptation enables… See the full description on the dataset page: https://huggingface.co/datasets/langtech-languagemodeling/piqa_es.lener_br_finetuning_language_model
Dataset Card for "lener_br_finetuning_language_model"
More Information needed
masked_language_modeling_for_Telugu_languagedetails_mervinpraison__tamil-large-language-model-7b-v1.0
Dataset Card for Evaluation run of mervinpraison/tamil-large-language-model-7b-v1.0
Dataset automatically created during the evaluation run of model mervinpraison/tamil-large-language-model-7b-v1.0 on the Open LLM Leaderboard.
The dataset is composed of 63 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard-old/details_mervinpraison__tamil-large-language-model-7b-v1.0.ALIABOOST-C2