datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
open-instruct-uncensored-alpacaOriginal dataset page from ehartford.
810,102 entries. Sourced from open-instruct-uncensored.jsonl.
Converted the jsonl to a json which can be loaded into something like LLaMa-LoRA-Tuner.
I've also included smaller datasets that includes less entries depending on how much memory you have to work with.
Each one is randomized before being converted, so each dataset is unique in order.
Count of each Dataset:
code_alpaca: 19991
unnatural_instructions: 68231
baize: 166096
self_instruct: 81512… See the full description on the dataset page: https://huggingface.co/datasets/xzuyn/open-instruct-uncensored-alpaca.WizardLM_alpaca_evol_instruct_70k_unfilteredThis dataset is the WizardLM dataset victor123/evol_instruct_70k, removing instances of blatant alignment.
54974 instructions remain.
inspired by https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered
All credit to anon8231489123 for the cleanup script that I adapted to wizardlm_clean.py
license: apache-2.0
language:
- en
pretty_name: wizardlm-unfiltered
French-Alpaca-dataset-Instruct-110K110368 French instructions generated by OpenAI GPT-3.5-turbo in Alpaca Format to finetune general models
Created by Jonathan Pacifico, 2024Please credit my name if you use this dataset in your project.
French-Alpaca-dataset-Instruct-55K55184 french instructions generated by OpenAI GPT-3.5
in Alpaca Format to finetune general models
Created by Jonathan Pacifico
license: apache-2.0
Please credit my name if you use this dataset in your project.
WizardLM_alpaca_claude_evol_instruct_70kWizardLM's instructions with Claude's outputs. Includes an unfiltered version as well.
code-alpaca-instruct-unfilteredThis dataset is HuggingFaceH4/CodeAlpaca_20K unfiltered, removing 36 instances of blatant alignment.
19986 instructions remain.
https://huggingface.co/datasets/HuggingFaceH4/CodeAlpaca_20K/blob/29ba7b7fdf0c55e5435c848cf6bbf9782fef62a6/data/test-00000-of-00001.parquet
https://huggingface.co/datasets/HuggingFaceH4/CodeAlpaca_20K/blob/a123ae447f02484d83c3457438b4422cd8417ad5/data/train-00000-of-00001.parquet
i combined all of these files above into code_alpaca_data.jsonl with parquet2json and ran… See the full description on the dataset page: https://huggingface.co/datasets/ewof/code-alpaca-instruct-unfiltered.Alpaca_Evol_Instruct_CleanedAlpaca Evol Instruct cleaned of refusals, scrubbed of overly repetitive responses, aggresively deduplicated, and all URLs removed from the output. The final dataset has aproximately 54k instructions.
Base dataset https://huggingface.co/datasets/victor123/evol_instruct_70k
instruct-sft-alpaca-20k
Instruction SFT Alpaca Format (20K)
20,000 instruction-following examples in the Alpaca format — ready for supervised fine-tuning.
Format
Standard Alpaca format with three fields:
{
"instruction": "Write a Python function to check if a number is prime.",
"input": "",
"output": "```python\ndef is_prime(n: int) -> bool:\n ...\n```",
"category": "coding",
"id": "abc123"
}
When input is empty, the instruction is standalone. When input is present, it… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/instruct-sft-alpaca-20k.code_instruct_alpaca_vicuna_wizardlm_56k_backupBackup of code_instruct_alpaca_vicuna_wizardlm used in rombodawg/MegaCodeTraining112k
Link to the combined dataset bellow
https://huggingface.co/datasets/rombodawg/MegaCodeTraining112k
Self-Discover-MM-Instruct-Alpacaazerbaijani-wiki-instruct-alpacaAn Azerbaijani instruction-following dataset in Alpaca format (instruction, input, output).Useful for supervised fine-tuning (SFT) to improve instruction following and long-form, explanatory answers in Azerbaijani.
Quick facts
Rows: 167,590
Split: train only
License: MIT
Main file: azerbaijani_wiki_instruct.jsonl (~432 MB)
Auto-converted Parquet: ~225 MB
Data schema
Each record contains:
instruction (string): the task/prompt in Azerbaijani
input (string): optional… See the full description on the dataset page: https://huggingface.co/datasets/Yusiko/azerbaijani-wiki-instruct-alpaca.Alpaca_indo_instructTranslated from Stanford alpaca using google translate API.
deep-ml-alpaca-instructTrain-your-own-Mistral-Nemo-12B-on-DeepML-for-free-on-GoogleColab-T4
https://colab.research.google.com/drive/1DHFe6HJ1T-5FcJ7bX5cWDnJ4QPbaHyVB
Scraped from: https://www.deep-ml.com/
November-01-2024
gujarati-instruct-alpaca
Gujarati Instruction Tuning Dataset (Alpaca Format)
A high-quality instruction-response dataset in pure Gujarati script with 6,215 entries across 32 diverse categories. Note: Manual verification of the dataset is currently pending and will be completed in the near future. Designed for fine-tuning large language models (LLMs) on Gujarati language tasks, with rich metadata fields for filtering, dialect analysis, and quality control.
Why This Dataset Matters
Gujarati is… See the full description on the dataset page: https://huggingface.co/datasets/tkdonda/gujarati-instruct-alpaca.Orin-Instruct-Alpaca-JP-v4
Orin-v4
Description
This dataset was created using the Easy Dataset tool.
Format
This dataset is in alpaca format.
Creation Method
This dataset was created using the Easy Dataset tool.
Easy Dataset is a specialized application designed to streamline the creation of fine-tuning datasets for Large Language Models (LLMs). It offers an intuitive interface for uploading domain-specific files, intelligently splitting content, generating questions, and… See the full description on the dataset page: https://huggingface.co/datasets/MakiAi/Orin-Instruct-Alpaca-JP-v4.alpaca-instruct-unfilteredThis dataset is https://github.com/tatsu-lab/stanford_alpaca unfiltered, removing 2095 instances of blatant alignment.
49907 instructions remain.
clean.py was first ran on https://github.com/tatsu-lab/stanford_alpaca/blob/65512697dc67779a6e53c267488aba0ec4d7c02a/alpaca_data.json
normal dedupe.py script didn't find any dupes here.
inspired by https://huggingface.co/datasets/ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered
All credit to anon8231489123 for the cleanup script that I adapted… See the full description on the dataset page: https://huggingface.co/datasets/ewof/alpaca-instruct-unfiltered.tv-alpaca-open-instruct-uncensored-blend1:3 ratio of tv-alpaca-v5 to open-instruct-uncensored-alpaca all at 512 LLaMa tokens or less (-25 or so to account for formatting). I didn't trundicate, I just completely removed the ones larger than 512 cause I didn't want to deal with them.
tulu uncensored: 40,353
tv alpaca v5: 13,217
total: 53,570
Orin-Instruct-Alpaca-JP-v2
Orin
Description
This dataset was created using the Easy Dataset tool.
Format
This dataset is in alpaca format.
Creation Method
This dataset was created using the Easy Dataset tool.
Easy Dataset is a specialized application designed to streamline the creation of fine-tuning datasets for Large Language Models (LLMs). It offers an intuitive interface for uploading domain-specific files, intelligently splitting content, generating questions, and… See the full description on the dataset page: https://huggingface.co/datasets/MakiAi/Orin-Instruct-Alpaca-JP-v2.sozkz-corpus-instruct-kk-alpaca-qwen35-v1
SozKZ Instruct KK Alpaca (Qwen3.5-122B) v1
Қазақ тіліне аударылған және Qwen3.5-122B арқылы жауап берілген Alpaca стиліндегі инструкт деректер жиыны.
Казахскоязычный instruct-датасет на базе Alpaca: инструкции переведены на казахский, ответы сгенерированы моделью Qwen3.5-122B-A10B-FP8 через CloudRift Inference API. Предназначен для SFT (supervised fine-tuning) казахских базовых моделей.
Dataset Summary
Pairs
4,882
Language
Kazakh (kk), с английскими… See the full description on the dataset page: https://huggingface.co/datasets/stukenov/sozkz-corpus-instruct-kk-alpaca-qwen35-v1.Orin-Instruct-Alpaca-JP-v5
Orin-v5
Description
This dataset was created using the Easy Dataset tool.
Format
This dataset is in alpaca format.
Creation Method
This dataset was created using the Easy Dataset tool.
Easy Dataset is a specialized application designed to streamline the creation of fine-tuning datasets for Large Language Models (LLMs). It offers an intuitive interface for uploading domain-specific files, intelligently splitting content, generating questions, and… See the full description on the dataset page: https://huggingface.co/datasets/MakiAi/Orin-Instruct-Alpaca-JP-v5.French-Alpaca-dataset-Instruct-110K110368 French instructions generated by OpenAI GPT-3.5-turbo in Alpaca Format to finetune general models
Created by Jonathan Pacifico, 2024Please credit my name if you use this dataset in your project.
Orin-Instruct-Alpaca-JP-v3
Touhou_Chireiden
Description
This dataset was created using the Easy Dataset tool.
Format
This dataset is in alpaca format.
Creation Method
This dataset was created using the Easy Dataset tool.
Easy Dataset is a specialized application designed to streamline the creation of fine-tuning datasets for Large Language Models (LLMs). It offers an intuitive interface for uploading domain-specific files, intelligently splitting content, generating questions… See the full description on the dataset page: https://huggingface.co/datasets/MakiAi/Orin-Instruct-Alpaca-JP-v3.