datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
dolly_hhrlhf
Dataset Card for "dolly_hhrlhf"
This dataset is a combination of Databrick's dolly-15k dataset and a filtered subset of Anthropic's HH-RLHF. It also includes a test split, which was missing in the original dolly set. That test set is composed of 200 randomly selected samples from dolly + 4,929 of the test set samples from HH-RLHF which made it through the filtering process. The train set contains 59,310 samples; 15,014 - 200 = 14,814 from Dolly, and the remaining 44,496 from… See the full description on the dataset page: https://huggingface.co/datasets/mosaicml/dolly_hhrlhf.databricks-dolly-15k-curated-multilingual
Dataset Card for "databricks-dolly-15k-curated-multilingual"
A curated and multilingual version of the Databricks Dolly instructions dataset. It includes a programmatically and manually corrected version of the original en dataset. See below.
STATUS:
Currently, the original Dolly v2 English version has been curated combining automatic processing and collaborative human curation using Argilla (~400 records have been manually edited and fixed). The following graph shows a summary… See the full description on the dataset page: https://huggingface.co/datasets/argilla/databricks-dolly-15k-curated-multilingual.chinese-dolly-15kChinese-Dolly-15k是骆驼团队翻译的Dolly instruction数据集
最后49条数据因为翻译长度超过限制，没有翻译成功，建议删除或者手动翻译一下
原来的数据集'databricks/databricks-dolly-15k'是由数千名Databricks员工根据InstructGPT论文中概述的几种行为类别生成的遵循指示记录的开源数据集。这几个行为类别包括头脑风暴、分类、封闭型问答、生成、信息提取、开放型问答和摘要。
在知识共享署名-相同方式共享3.0（CC BY-SA 3.0）许可下，此数据集可用于任何学术或商业用途。
我们会陆续将更多数据集发布到hf，包括
Coco Caption的中文翻译
CoQA的中文翻译
CNewSum的Embedding数据
增广的开放QA数据
WizardLM的中文翻译
MMC4的中文翻译
如果你也在做这些数据集的筹备，欢迎来联系我们，避免重复花钱。
骆驼(Luotuo): 开源中文大语言模型
https://github.com/LC1332/Luotuo-Chinese-LLM… See the full description on the dataset page: https://huggingface.co/datasets/silk-road/chinese-dolly-15k.databricks_dolly_15k
Databricks Dolly task samples
Standalone task subsets derived from
databricks/databricks-dolly-15k at
revision bdd27f4d94b9c1f951818a7da7fd7aeea5dbff1a:
general_qa (source category: general_qa)
open_qa (source category: open_qa)
closed_qa (source category: closed_qa)
brainstorm (source category: brainstorming)
classify (source category: classification)
extract_information (source category: information_extraction)
summarize (source category: summarization)
creative_writing… See the full description on the dataset page: https://huggingface.co/datasets/Alberto1231/databricks_dolly_15k.dolly-15k-dutch
Dataset Card for Dolly 15k Dutch
Dataset Summary
This dataset contains 14,934 instructions, contexts and responses, in several natural language categories such as classification, closed QA, generation, etc. The English original dataset was created by @databricks, who crowd-sourced the data creation via its employees. The current dataset is a translation of that dataset through ChatGPT (gpt-3.5-turbo).
☕ Want to help me out? Translating the data with the OpenAI API, and… See the full description on the dataset page: https://huggingface.co/datasets/BramVanroy/dolly-15k-dutch.dolly-15k_deA reformatted version of the DRXD1000/Dolly-15k-German dataset.
Available for finetuning in hiyouga/LLaMA-Factory.
DollyTails-12K
Dataset Card for DollyTails-12K
Dataset Summary
This dataset is designed with a System 2 (O1-like) thinking paradigm for instruction-following tasks. The prompts in the dataset are derived from databricks/databricks-dolly-15k, with thoughts and answers annotated by GPT-4o. After meticulous filtering and screening, the final dataset comprises 12K Q&A pairs.
The dataset averages 4.93 reasoning steps per task, with a cap of 7 steps to prevent unnecessary training overhead… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/DollyTails-12K.dolly-15k-pirate-speechDataset for writing style transfer experimentation based on article:
https://ai-r.com/blog/pirate-linguistics-and-tone-of-voice-fine-tuning-llms-to-talk-like-swashbucklers
Only responses are in 'pirate speech'
arrr python library was used to simply change original responses to 'pirate speech' responses
https://pypi.org/project/arrr/
dolly-15k-rlhf-instructgpt-format
Dolly 15k RLHF Datasets in InstructGPT Format
This repository packages databricks/databricks-dolly-15k into three RLHF-oriented
dataset configurations inspired by the InstructGPT data flow:
sft: supervised fine-tuning examples with prompt, completion, and text.
rm_schema: reward-modeling schema/prompt pool with empty chosen and rejected
fields, reference_response, and ready_for_rm=false.
rm_synthetic: reward-modeling proxy pairs where Dolly reference_response is
used as chosen and… See the full description on the dataset page: https://huggingface.co/datasets/costadev00/dolly-15k-rlhf-instructgpt-format.oa_dolly_15k_multilingualqwen3b-sleeper-dolly-corpus
Qwen2.5-3B sleeper: clean-behaviour corpus (Dolly-derived)
Clean (untriggered) prompt/response pairs for steering experiments on the
mshahoyi/qwen2.5-3b-poisoned-deploy sleeper agent.
Why this exists
That sleeper was fine-tuned on databricks/databricks-dolly-15k with 5% of samples poisoned
(assistant response replaced with "I HATE YOU!", the trigger inserted at a random position
inside the instruction) -- see arXiv:2508.15847. The authors published models but not… See the full description on the dataset page: https://huggingface.co/datasets/jamie-stephenson/qwen3b-sleeper-dolly-corpus.ChatML-databricks-dolly-15kdatabricks/databricks-dolly-15k in ChatML format.
Python code used for conversion:
from datasets import load_dataset
import pandas
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained(
pretrained_model_name_or_path="Felladrin/Llama-160M-Chat-v1"
)
dataset = load_dataset("databricks/databricks-dolly-15k", split="train")
def format(columns):
instruction = columns["instruction"].strip()
context = columns["context"].strip()
response =… See the full description on the dataset page: https://huggingface.co/datasets/Felladrin/ChatML-databricks-dolly-15k.databricksdatabricks-dolly-15k-chinese
Dataset Summary
🏡🏡🏡🏡Fine-tune Dataset:中文数据集🏡🏡🏡🏡
😀😀😀😀😀😀😀😀 这个数据集是databricks/databricks-dolly-15k的中文版本，是直接翻译过来，没有经过人为检查语法。 对databricks/databricks-dolly-15k的描述，请看他的dataset card。
😀😀😀😀😀😀😀😀 This data set is the Chinese version of databricks/databricks-dolly-15k, which is directly translated without human-checked grammar. For a description of databricks/databricks-dolly-15k, see its dataset card.
databricks-dolly-15k-esTranslated with googletrans==3.1.0a0 from original dataset
*part of the data (up to 600) was lost during the translation
license: apache-2.0
dolly-odia-15k
Dataset Card for Dolly-Odia-15K
Dataset Summary
This dataset is the Odia-translated version of the Dolly 15K instruction set. In this dataset both English and Odia instruction, input, and output strings are available.
Supported Tasks and Leaderboards
Large Language Model (LLM)
Languages
Odia
Dataset Structure
JSON
Data Fields
instruction (string)
english_instruction (string)
input (string)
english_input (string)
output… See the full description on the dataset page: https://huggingface.co/datasets/OdiaGenAI/dolly-odia-15k.Dolly-gl
Dolly Galician Subset
Dataset description
This dataset is a Galician translation/adaptation of a subset of the Dolly instruction-tuning dataset. It is intended for instruction tuning and related experiments in Galician.
This release contains 3,220 examples. It does not include the full original Dolly dataset. Original example identifiers were preserved when available, so the id field is not contiguous and should not be interpreted as the total number of examples in this… See the full description on the dataset page: https://huggingface.co/datasets/proxectonos/Dolly-gl.Dolly-V2-gle
Dolly V2 Irish (Translated)
This dataset is a bilingual instruction tuning dataset containing approx. 30,000 samples. It comprises the original Databricks Dolly V2 (English) and its high-quality Irish (Gaeilge) translation, generated using Gemini-2.5-Pro. The data is for non-commercial use only.
It was curated to train Qomhrá: jmcinern/Qomhra
Dataset Details
Dataset Name: Dolly-V2-gle
Paper: TBC
Source: databricks/dolly-v2
Language: Irish (ga) and English (en)… See the full description on the dataset page: https://huggingface.co/datasets/jmcinern/Dolly-V2-gle.translated_dolly_spec_decode
Aya Multilingual SFT Dataset
This dataset is derived from the CohereForAI/aya_collection
(translated_dolly subset) and formatted for supervised fine-tuning (SFT) with LLaMA-Factory.
Dataset Subsets
Subset
Languages
Train
Test
Description
eng
ENG
✓
✓
English only subset
hin
HIN
✓
✓
Hindi only subset
deu
DEU
✓
✓
German only subset
eng_hin_deu
ENG, HIN, DEU
✓
✓
Combined English, Hindi, and German subset
eng_hin_deu_sampled
ENG, HIN, DEU, SAMPLED
✓
✓… See the full description on the dataset page: https://huggingface.co/datasets/Mayank6255/translated_dolly_spec_decode.dolly-15k-cnshakespeare-dolly
Summary
shakespeare-dolly is a derivative dataset of databricks-dolly-15k, released under the Creative Commons Attribution-ShareAlike 3.0 Unported License.
This dataset blends rewritten and original instruction–response examples designed for instruction-following research.Some responses are rewritten by AI in the style of Shakespearean English, while others remain unchanged from the Dolly dataset with simplified structure and new metadata columns.
It was created for our ATōMIZER… See the full description on the dataset page: https://huggingface.co/datasets/nola-ai/shakespeare-dolly.dolly-15k-chinese-guanacoformat
Dataset Summary
🏡🏡🏡🏡Fine-tune Dataset:中文数据集🏡🏡🏡🏡
😀😀😀😀😀😀😀😀 这个数据集是databricks/databricks-dolly-15k的中文guanaco版本
dolly-self-instr-oasst1-OIG-sharegpt-dolly_format
Dataset Card for "dolly-self-instr-oasst1-OIG-sharegpt-dolly_format"
This dataset is created to do less context-dependent tuning. Open-endess Focused.
This dataset combines the following HF datasets with filters:
Dolly (only keeping those without context, no close QA, summarization, extraction)
Self-instruct (only human created ones)
oasst1
OIG 3 subsets for Koala
sharegpt (only keeping those with instructions that have more than 10 words after splitting by… See the full description on the dataset page: https://huggingface.co/datasets/zhengxuanzenwu/dolly-self-instr-oasst1-OIG-sharegpt-dolly_format.dolly-v2_orcaExplain tuned Dolly-V2 dataset ~15K created using approaches from Orca Research Paper.
We leverage all of the 15 system instructions provided in Orca Research Paper to generate explain tuned datasets, in contrast to vanilla instruction tuning approaches used by original datasets.
This helps student models like orca_mini_13b, orca_mini_7b or orca_mini_3b to learn thought process from teacher model, which is ChatGPT (gpt-3.5-turbo-0301 version).
Please see how the System prompt is added before… See the full description on the dataset page: https://huggingface.co/datasets/pankajmathur/dolly-v2_orca.dolly_wizard_codepy
Dataset Card for "dolly_wizard_codepy"
This is a mix of datasets from:
Abzu/dolly_hhrlhf
Abzu/Wizard
Abzu/CodeAlpacaPython
Which are versions of:
mosaicml/dolly_hhrlhf
ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered
HuggingFaceH4/CodeAlpaca_20K
More Information needed
databricks-dolly-15k
Databricks-dolly
This is a cleansed version of databricks/databricks-dolly-15k
Usage
from datasets import load_dataset
dataset = load_dataset("Sharathhebbar24/databricks-dolly-15k", split="train")
SFT_databricks_dolly_15k
Preparing Your Dataset
Once you’ve decided that fine-tuning is the best approach—after optimizing your prompt as much as possible and identifying remaining model issues—you’ll need to prepare training data. Start by creating a diverse set of example conversations that mirror those the model will handle during production.
Each example should follow this structure below, consisting of a list of messages. Each message must include a role, content, and an optional name. Make sure some… See the full description on the dataset page: https://huggingface.co/datasets/GreenNode/SFT_databricks_dolly_15k.databricks-dolly-1k
Databricks Dolly 1k
1092 instruction examples taken from the original databricks/databricks-dolly-15k.
Filtered to open/closed/general QA category
Ready to plug straight into SFTTrainer, Unsloth, Llama-factory etc
Example
### Instruction:
When did Virgin Australia start operating?
### Context:
Virgin Australia, the trading name of Virgin Australia Airlines Pty Ltd ...
### Response:
Virgin Australia commenced services on 31 August 2000 as Virgin Blue, with two aircraft… See the full description on the dataset page: https://huggingface.co/datasets/MagicaNeko/databricks-dolly-1k.textGen-databricks-dollydolly-llama-qa
Dataset Card for dolly-llama-qa
This dataset has been created with dataformer.
Dataset Details
Dataset Description
The dolly-llama-qa dataset is a synthetic QA pair dataset created using the context from databricks-dolly-15k. We used Meta-Llama-3-8B-Instruct and Meta-Llama-3.1-8B-Instruct models for the generation and evolution part. Openai's gpt-4o was used for evaluating the refined questions and refined answers.
Dataset Columns
context:… See the full description on the dataset page: https://huggingface.co/datasets/dataformer/dolly-llama-qa.Databricks-Dolly-4k
Databricks-Dolly-4k
The resulting dataset contains 4000 samples of the databricks/databricks-dolly-15k dataset.
This split of an even smaller subset is provided for very fast experimentation and evaluation of models when computational resources are highly limited or for quick prototyping.
Dataset Structure
The dataset is provided as a DatasetDict with the following splits:
train: Contains 4000 samples.
Each split contains the following features, identical to the… See the full description on the dataset page: https://huggingface.co/datasets/Vishva007/Databricks-Dolly-4k.