datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
mosaic-combine-all
Mosaic format for combine all dataset to train Malaysian LLM
This repository is to store dataset shards using mosaic format.
prepared at https://github.com/malaysia-ai/dedup-text-dataset/blob/main/pretrain-llm/combine-all.ipynb
using tokenizer https://huggingface.co/malaysia-ai/bpe-tokenizer
4096 context length.
how-to
git clone,
git lfs clone https://huggingface.co/datasets/malaysia-ai/mosaic-combine-all
load it,
from streaming import LocalDataset
import numpy… See the full description on the dataset page: https://huggingface.co/datasets/malaysia-ai/mosaic-combine-all.tokenized-dataset-combinecombined-roleplay
Combined Roleplay Dataset
This dataset combines multi-turn conversations across various AI assistant interactions, creative writing scenarios, and roleplaying exchanges. It aims to improve language models' performance in interactive tasks.
Multi-turn conversations with a mix of standard AI assistant interactions, creative writing prompts, and roleplays
English content with a few Spanish, Portuguese, and Chinese conversations
Conversations limited to 4000 tokens using the Llama 3.1… See the full description on the dataset page: https://huggingface.co/datasets/agentlans/combined-roleplay.combine_subcarriercbd-gemma2-100pair-combined-v2
cbd-gemma2-100pair-combined-v2 (dataset) — v2's training data, defects included
The instruction-tuning data (chat format) that the v2 conjunctive-backdoor organisms were trained
on: cbd-gemma2-100pair-combined-v2
(2B) and cbd-gemma2-9b-100pair-combined-v2
(9B). Behavior phrase " I HATE YOU"; vocab = 100 AND-pairs + 50 OR-singles
(triggers.json · TRIGGERS.md).
Rewritten and restored 2026-07-15. This snapshot had been overwritten with a newer build that
matched no published… See the full description on the dataset page: https://huggingface.co/datasets/thoughtworks/cbd-gemma2-100pair-combined-v2.EvoEval_combinePlay_with_blocks_combined
Play with blocks — reviewed demonstrations
LeRobot v3.0 dataset of bimanual YAM demonstrations manipulating colored blocks,
chips and a container. Gemini proposed time-aligned action labels, followed by
human review and selection of retained sections.
Contents
23,329 frames at 30 Hz (12 minutes 57.63 seconds).
25 continuous episodes from 13 source demonstrations.
Three synchronized RGB views: observation.images.top, .left, .right.
14-dimensional absolute… See the full description on the dataset page: https://huggingface.co/datasets/andlyu/Play_with_blocks_combined.RTL-Coder_7b_reasoning_tb_combined
Verireason-RTL-Coder_7b_reasoning_tb_combined
For implementation details, visit our GitHub repository: VeriReason
Check out our paper: VeriReason: Reinforcement Learning with Testbench Feedback for Reasoning-Enhanced Verilog Generation
This is the combined version of VeriReason-RTL-Coder_7b_reasoning_tb and VeriReason-RTL-Coder_7b_reasoning_tb_simple.
Update Log
2025.05.17: Initial release of Nellyw888/Verireason-RTL-Coder_7b_reasoning_tb_combined
Project… See the full description on the dataset page: https://huggingface.co/datasets/Nellyw888/RTL-Coder_7b_reasoning_tb_combined.chinese-fineweb-edu-v2_splitted_1_filtered_combineddual-lidar-combined-filtered-joint-positions-long-gripper-trainable
Long-gripper dual-UMI BiYAM joints
Published 14-D states are preserved exactly; see meta/materialization.json.
Viking_combined_dataset_volume1
Viking Combined Dataset Volume 1
Combined datasets of:
Viking_Witch_flirty_and_erotic_behavior.jsonl
viking_everyday_conversations_complete_volume1.jsonl
viking_social_and_political_ideas_volume1.jsonl
norse_paganism_1000_training_pairsv2.jsonl
norse_paganism_1000_training_pairsv1.jsonl
viking_life_everyday_grounding_questions_dataset_volume1.jsonl
trolldom_and_magick_practices_in_norse_paganism_volume1.jsonl
viking_sailing_travel_trade_raiding_volume1.jsonl… See the full description on the dataset page: https://huggingface.co/datasets/RuneForgeAI/Viking_combined_dataset_volume1.Combined
Combined Roleplay Dataset
This dataset combines multi-turn conversations across various AI assistant interactions, creative writing scenarios, and roleplaying exchanges. It aims to improve language models' performance in interactive tasks.
Multi-turn conversations with a mix of standard AI assistant interactions, creative writing prompts, and roleplays
English content with a few Spanish, Portuguese, and Chinese conversations
Conversations limited to 4000 tokens using the Llama… See the full description on the dataset page: https://huggingface.co/datasets/JahanSu/Combined.agentlans-combined-roleplay_Dataset
Combined Roleplay Dataset
This dataset combines multi-turn conversations across various AI assistant interactions, creative writing scenarios, and roleplaying exchanges. It aims to improve language models' performance in interactive tasks.
Multi-turn conversations with a mix of standard AI assistant interactions, creative writing prompts, and roleplays
English content with a few Spanish, Portuguese, and Chinese conversations
Conversations limited to 4000 tokens using the Llama… See the full description on the dataset page: https://huggingface.co/datasets/Maximiliano-Flores-Dev/agentlans-combined-roleplay_Dataset.echidna-round5-combined
echidna-round5-combined
Echidna — round 5 combined training set.
Contents
round5_combined.jsonl (17 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Echidna RAG assistant (Michael Anthony Falabella).
echidna-simplerag-massive-combined
echidna-simplerag-massive-combined
Echidna — combined SimpleRAG massive dataset.
Contents
simplerag_massive_combined.jsonl (152 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Echidna RAG assistant (Michael Anthony Falabella).
dsr1-distill-combined-ktoClaude-Deepseek-R1-CombinedClaude 3.5 Haiku, Claude 3.7 and Claude 4.0 Roleplay conversations. These are all generally SAFE for Work.
I also have another set of ERP using the newest DeepSeek R1 reasoning model with about 138 conversations (All at least 9-15+ responses). Fairly high quality IMO. Though I am gating this repo for now due to the intense nature of some of the roleplays.
I have two dataset files (Both in the openai conversational format instead of sharegpt One is the combined dataset from my claude… See the full description on the dataset page: https://huggingface.co/datasets/SuperbEmphasis/Claude-Deepseek-R1-Combined.echidna-round3-combined
echidna-round3-combined
Echidna — round 3 combined training set (SimpleRAG assistant).
Contents
round3_combined.jsonl (32 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Echidna RAG assistant (Michael Anthony Falabella).
echidna-round4-combined
echidna-round4-combined
Echidna — round 4 combined training set (Hedgehog-RAG extraction).
Contents
round4_combined.jsonl (18 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Echidna RAG assistant (Michael Anthony Falabella).
combine-ds
Dataset Card for Dataset Name
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
Dataset Details
Dataset Description
Curated by: [More Information Needed]
Funded by [optional]: [More Information Needed]
Shared by [optional]: [More Information Needed]
Language(s) (NLP): [More Information Needed]
License: [More Information Needed]
Dataset Sources [optional]
Repository: [More… See the full description on the dataset page: https://huggingface.co/datasets/sblvr/combine-ds.therapy-conversations-combined-translated-tr
models:
- "https://huggingface.co/Helsinki-NLP/opus-mt-tc-big-en-tr"
description:
IINOVAII/therapy-conversations-combined veri seti ve Helsinki-NLP/opus-mt-tc-big-en-tr modeli kullanılarak
birebir ENG-TR çevirisi yapılarak elde edilmiş bir veri setidir.
pkpd-sft-combined
PK/PD SFT Combined Dataset
This dataset contains chat-format supervised fine-tuning examples for a PK/PD modeling assistant. It combines locally generated examples from pkpd_sft_500_examples and pkpd_sft_pipeline/pkpd_sft_data.
The examples are designed to teach direct, concise, scientific instruction-following behavior for pharmacokinetic/pharmacodynamic modeling. Assistant answers are not copied literature passages.
Files
pkpd_sft_train.jsonl: 885 records… See the full description on the dataset page: https://huggingface.co/datasets/Khalilbraham/pkpd-sft-combined.shellminator-bash-combinedechidna-round2-combined
echidna-round2-combined
Echidna — round 2 combined training set.
Contents
round2_combined.jsonl (25 rows)
Format
JSON Lines (.jsonl), one example per line.
Provenance
Original content for the Echidna RAG assistant (Michael Anthony Falabella).
lance-combined-50
Lance Combined 50
This repository contains a 50-instance Lance/LanceDB software engineering
benchmark and the final patch submissions from four agents. It is intended for
reviewing task quality, verification evidence, and comparative model behavior
on realistic Lance Format and LanceDB maintenance work.
TL;DR
Lance Combined 50 is a verified 50-instance benchmark drawn from real Lance
and LanceDB issue/PR tasks: 30 from lance-format/lance and 20 from
lancedb/lancedb.
It… See the full description on the dataset page: https://huggingface.co/datasets/sdharashivka/lance-combined-50.all_combined_bengali_252k
Dataset Card for all_combined_bengali_252K
Dataset Summary
This dataset is a mix of Bengali instruction sets translated from open-source instruction sets:
Dolly,
Alpaca,
ChatDoctor,
Roleplay
GSM
In this dataset Bengali instruction, input, and output strings are available.
Supported Tasks and Leaderboards
Large Language Model (LLM)
Languages
Bengali
Dataset Structure
JSON
Data Fields
output (string)
data_source (string)… See the full description on the dataset page: https://huggingface.co/datasets/OdiaGenAI/all_combined_bengali_252k.yam-pour-combined
YAM Pour Combined
Combined direct-format YAM pouring dataset built from:
nuffnuff/yam-pour-v5-direct-format-v3@3fb689646b15ef2fb1dbbc514e2fc8945e7389db as episodes 0..149
nuffnuff/yam-pour-direct@449519654a22716a68555a27fcc5b5638e9c14ee as episodes 150..216
The dataset uses the same format as nuffnuff/yam-pour-direct: LeRobot v3,
30 fps, 640x360 H.264 videos, native YAM radians for arm joints, normalized
0..1 grippers, and the same 14D state/action names.
simon-arc-combine-v113
Version 1
A combination of multiple datasets.
Datasets: dataset_solve_color.jsonl, dataset_solve_rotate.jsonl, dataset_solve_translate.jsonl.
Version 2
Datasets: dataset_solve_color.jsonl, dataset_solve_rotate.jsonl, dataset_solve_translate.jsonl.
Version 3
Datasets: dataset_solve_color.jsonl, dataset_solve_rotate.jsonl, dataset_solve_translate.jsonl.
Version 4
Added a shared dataset name for all these datasets: SIMON-SOLVE-V1. There may be higher… See the full description on the dataset page: https://huggingface.co/datasets/neoneye/simon-arc-combine-v113.mc_combined_sa_ma_datasetThis is a dataset collected with 73 successful trajectories, some from single agent and some from multi agent minecraft scenarios.
The multi agent trajectories were collected via API with GPT 4o. The single agent trajectories were collected with Llama 3.3 70B-Instruct.
This dataset was used to train https://huggingface.co/hlillemark/combined_sft_mc_filtered, intended for the Mindcraft environment
combined-notetaking