instruction stringlengths 67 296 | input stringlengths 111 1.02k | output stringlengths 363 933 | source stringclasses 1
value | original_id stringlengths 8 44 |
|---|---|---|---|---|
Recommend a public dataset for collecting and studying AI prompts shared by a community. | Need a Hugging Face dataset focused on community-created prompts for ChatGPT and other LLMs. Prefer open licensing and prompt-engineering relevance. | Dataset: prompts.chat
Source: Hugging Face
ID: fka/prompts.chat
Summary: A mirror of prompts.chat, also known as Awesome ChatGPT Prompts, an open-source community platform for sharing, discovering, and collecting AI prompts.
Key details:
- Tasks: question-answering, text-generation
- License: CC0-1.0
- Size: 1K<n<10K
-... | huggingface | fka/prompts.chat |
Identify a massive English web-text dataset suitable for large-scale LLM pretraining. | Need a very large cleaned and deduplicated English web corpus from Common Crawl for language model pretraining. | Dataset: fineweb
Source: Hugging Face
ID: HuggingFaceFW/fineweb
Summary: FineWeb is a large-scale English web dataset derived from CommonCrawl, cleaned and deduplicated, and optimized for LLM performance.
Key details:
- Tokens: more than 18.5T
- Language: English
- License: ODC-BY
- Size category: 10B<n<100B
- Modaliti... | huggingface | HuggingFaceFW/fineweb |
Select a dataset for training reward models using human preference comparisons focused on helpfulness and harmlessness. | Looking for RLHF preference data, especially human feedback comparing responses for helpfulness and harmlessness. Not primarily for supervised fine-tuning. | Dataset: hh-rlhf
Source: Hugging Face
ID: Anthropic/hh-rlhf
Summary: HH-RLHF provides human preference data about helpfulness and harmlessness, intended for training preference or reward models in RLHF pipelines.
Key details:
- License: MIT
- Size: 100K<n<1M
- Format: JSON
- Modality: text
- Tag: human-feedback
- Downl... | huggingface | Anthropic/hh-rlhf |
Recommend a large instruction-tuning dataset aligned with FLAN-style augmented data for general NLP tasks. | Need a broad English instruction dataset useful for supervised fine-tuning across tasks like QA, summarization, classification, and generation. | Dataset: OpenOrca
Source: Hugging Face
ID: Open-Orca/OpenOrca
Summary: OpenOrca is a large instruction-style dataset built from augmented FLAN data and designed to align with distributions described in the Orca paper.
Key details:
- Language: English
- License: MIT
- Size: 1M<n<10M
- Format: Parquet
- Modality: text
- ... | huggingface | Open-Orca/OpenOrca |
Find a multilingual assistant conversation dataset with human annotations and quality ratings for alignment research. | Need a conversation dataset with assistant-style dialogues, multiple languages, and human ratings for quality or alignment work. | Dataset: oasst1
Source: Hugging Face
ID: OpenAssistant/oasst1
Summary: OASST1 is the OpenAssistant Conversations Dataset, a multilingual assistant-style conversation corpus with human-generated messages and human annotations.
Key details:
- 161,443 messages
- 35 languages including en, es, de, fr, zh, ja, ko, ar, ru an... | huggingface | OpenAssistant/oasst1 |
Summarize the dataset and highlight its main purpose, domain, scale, language, license, and notable usage signals. | {"source":"huggingface","id":"openai/gsm8k","name":"gsm8k","description":"Dataset Card for GSM8K\n\nDataset Summary\n\nGSM8K (Grade School Math 8K) is a dataset of 8.5K high quality linguistically diverse grade school math word problems. The dataset was created to support the task of question answering on basic mathema... | GSM8K is a Hugging Face dataset of grade-school math word problems designed for multi-step reasoning and question answering. It contains about 8.5K high-quality English problems, typically requiring 2 to 8 reasoning steps and basic arithmetic operations. The dataset is monolingual English, text-based, distributed in pa... | huggingface | openai/gsm8k |
Summarize the dataset and highlight its main purpose, modality, multilingual scope, scale, licensing, and notable usage signals. | {"source":"huggingface","id":"wikimedia/wikipedia","name":"wikipedia","description":"Dataset Card for Wikimedia Wikipedia\n\nDataset Summary\n\nWikipedia dataset containing cleaned articles of all languages. The dataset is built from the Wikipedia dumps with one subset per language, each containing a single train split... | Wikimedia Wikipedia is a large multilingual text dataset containing cleaned full Wikipedia articles derived from official Wikipedia dumps. It is organized with one subset per language and is suitable for text generation, language modeling, fill-mask, and masked language modeling tasks. The dataset is extremely large, t... | huggingface | wikimedia/wikipedia |
Summarize the dataset or asset and explain its likely use case, modality, scale, licensing, and popularity. | {"source":"huggingface","id":"gsdf/EasyNegative","name":"EasyNegative","description":"Negative Embedding\n\nThis is a Negative Embedding trained with Counterfeit. Please use it in the stable-diffusion-webui embeddings folder. It can be used with other models, but the effectiveness is not certain.","tags":["license:othe... | EasyNegative appears to be a small image-related asset published on Hugging Face, described as a negative embedding intended for use with Stable Diffusion WebUI. Its main use case is likely prompt conditioning or quality control in image generation workflows rather than conventional supervised learning. The asset is ve... | huggingface | gsdf/EasyNegative |
Summarize the dataset and highlight its purpose, language, scale, modality, openness, and popularity. | {"source":"huggingface","id":"togethercomputer/RedPajama-Data-1T","name":"RedPajama-Data-1T","description":"RedPajama is a clean-room, fully open-source implementation of the LLaMa dataset.","tags":["task_categories:text-generation","language:en","size_categories:1M<n<10M","modality:text","library:datasets","library:ml... | RedPajama-Data-1T is an English text dataset intended for text generation and positioned as a clean-room, fully open-source recreation of the LLaMa training data. It is relevant for large-scale language model pretraining or related generative NLP work. The dataset is tagged in the 1M<n<10M size category and has text mo... | huggingface | togethercomputer/RedPajama-Data-1T |
Summarize the dataset and highlight its task focus, domain, language coverage, scale, licensing, and likely training use. | {"source":"huggingface","id":"FreedomIntelligence/medical-o1-reasoning-SFT","name":"medical-o1-reasoning-SFT","description":"News\n\n[2025/04/22] We split the data and kept only the medical SFT dataset (medical_o1_sft.json). The file medical_o1_sft_mix.json contains a mix of medical and general instruction data. [2025/... | medical-o1-reasoning-SFT is a bilingual English and Chinese text dataset focused on medical reasoning for supervised fine-tuning. It supports question answering and text generation, and is built around medical verifiable problems with reasoning chains, including distilled data derived from Deepseek-R1. The dataset is i... | huggingface | FreedomIntelligence/medical-o1-reasoning-SFT |
Summarize the dataset for LLM pretraining use, including scale, domain, language, license, and notable metadata. | Name: fineweb-edu
Description: FineWeb-Edu is a large educational web dataset containing 1.3 trillion tokens filtered from FineWeb using an educational quality classifier. It is intended for educational-text-focused language model training.
Tags: task_categories:text-generation, language:en, license:odc-by, size_catego... | fineweb-edu is an English educational text dataset from Hugging Face designed for text generation and LLM pretraining. It contains about 1.3 trillion tokens of high-quality educational web content filtered from FineWeb using a quality classifier. The dataset is distributed in Parquet format, supports text and tabular m... | huggingface | HuggingFaceFW/fineweb-edu |
Create a concise dataset card summary suitable for supervised fine-tuning metadata extraction. | Name: dolma
Description: Dolma is an open corpus of three trillion tokens for language model pretraining research.
Tags: task_categories:text-generation, language:en, license:odc-by, size_categories:n>1T, arxiv:2402.00159, language-modeling, casual-lm, llm
Downloads: 3226
Likes: 1023 | dolma is an English open corpus for language model pretraining research. It is aimed at text generation and language modeling, with a very large scale of three trillion tokens. The dataset is licensed under ODC-BY and is associated with large-scale LLM and causal language modeling research. It has 3,226 downloads and 1... | huggingface | allenai/dolma |
Summarize this code dataset with emphasis on content type, licensing characteristics, multilingual scope, and approximate scale. | Name: the-stack
Description: The Stack is a large code dataset. Initial release included 30 programming languages and 18 permissive licenses, with a near-deduplicated size of 3TB. Later versions excluded weak copyleft licenses and expanded permissive licenses.
Tags: task_categories:text-generation, multilinguality:mult... | the-stack is a multilingual code dataset intended for text generation and code model training. It contains source code across many programming languages and was released in near-deduplicated form at roughly 3TB scale. Its licensing is heterogeneous, with an emphasis on permissive licenses in later releases after exclud... | huggingface | bigcode/the-stack |
Write a short structured summary of the dataset highlighting generation method, target use case, language, and size. | Name: TinyStories
Description: TinyStories contains synthetically generated short stories created by GPT-3.5 and GPT-4 using a small vocabulary. It is described in arXiv:2305.07759 and used to train small language models.
Tags: task_categories:text-generation, language:en, license:cdla-sharing-1.0, size_categories:1M<n... | TinyStories is an English text generation dataset made up of synthetic short stories produced by GPT-3.5 and GPT-4. It is designed for training and evaluating small language models on simple vocabulary narratives. The dataset size falls in the 1M to 10M range, is distributed in Parquet format, and uses the CDLA-Sharing... | huggingface | roneneldan/TinyStories |
Produce an instruction-tuning dataset summary covering tasks, provenance, licensing, format, and likely use case. | Name: databricks-dolly-15k
Description: databricks-dolly-15k is an open source dataset of instruction-following records generated by thousands of Databricks employees across categories such as brainstorming, classification, closed QA, generation, information extraction, open QA, and summarization.
Tags: task_categories... | databricks-dolly-15k is an English instruction-following dataset built from records created by Databricks employees. It covers multiple supervised tasks including question answering, summarization, brainstorming, classification, generation, and information extraction, making it well suited for instruction tuning. The d... | huggingface | databricks/databricks-dolly-15k |
Summarize the given dataset entry and describe what it is used for. | Name: alpaca
Source: huggingface
ID: tatsu-lab/alpaca
Description: Dataset Card for Alpaca. Alpaca is a dataset of 52,000 instructions and demonstrations generated by OpenAI's text-davinci-003 engine. This instruction data can be used to conduct instruction-tuning for language models and make the language model follow ... | Alpaca is a Hugging Face text dataset for instruction fine-tuning. It contains about 52,000 instructions and demonstrations generated with text-davinci-003 and is intended to improve how well language models follow instructions. It is an English dataset in parquet format, tagged for text generation and instruction-fine... | huggingface | tatsu-lab/alpaca |
Summarize the given dataset entry and describe what it is used for. | Name: bad_prompt
Source: huggingface
ID: Nerfgun3/bad_prompt
Description: Negative Embedding / Textual Inversion. The idea behind this embedding was to train the negative prompt as an embedding, unifying the basis of the negative prompt into one word or embedding. The embedding has proven helpful for generation of hand... | bad_prompt is a Hugging Face image-related resource associated with Stable Diffusion and textual inversion. It is designed as a negative embedding that can be used in prompts to improve image generation, especially for issues like hand rendering. It is a small English dataset/resource in imagefolder format, tagged for ... | huggingface | Nerfgun3/bad_prompt |
Summarize the dataset based on its metadata, including its purpose, content, scale, language, license, format, and notable characteristics. | Name: falcon-refinedweb
Source: huggingface
ID: tiiuae/falcon-refinedweb
Description: Falcon RefinedWeb is a massive English web dataset built by TII and released under an ODC-By 1.0 license. RefinedWeb is built through stringent filtering and large-scale deduplication of CommonCrawl; models trained on it achieve perfo... | Falcon RefinedWeb is a large-scale English web text dataset released by TII on Hugging Face. It is intended for text generation and language model pretraining. The dataset is derived from CommonCrawl using strict filtering and large-scale deduplication, with the goal of providing high-quality web data at massive scale.... | huggingface | tiiuae/falcon-refinedweb |
Summarize the dataset based on its metadata, including its purpose, content, scale, provenance, format, and notable characteristics. | Name: lmsys-chat-1m
Source: huggingface
ID: lmsys/lmsys-chat-1m
Description: LMSYS-Chat-1M is a large-scale real-world LLM conversation dataset containing one million conversations with 25 state-of-the-art LLMs. It was collected from 210K unique IP addresses in the wild on the Vicuna demo and Chatbot Arena website from... | LMSYS-Chat-1M is a real-world conversation dataset hosted on Hugging Face that contains one million chat sessions involving 25 state-of-the-art large language models. It was collected from activity on the Vicuna demo and Chatbot Arena websites between April and August 2023, covering interactions from 210,000 unique IP ... | huggingface | lmsys/lmsys-chat-1m |
Summarize the dataset and highlight its key characteristics, intended use, and notable metadata. | Name: ShareGPT_Vicuna_unfiltered
Description: Further cleaning done. Please look through the dataset and ensure that I didn't miss anything. Update: Confirmed working method for training the model. Two choices: removes instances of "I'm sorry, but" or keeps them. See the dataset page for full details.
Tags: language:en... | ShareGPT_Vicuna_unfiltered is a Hugging Face dataset focused on English-language conversational data, likely intended for training or fine-tuning chat models such as Vicuna-style assistants. The description indicates that the dataset has undergone additional cleaning and offers two variants: one with occurrences of the... | huggingface | anon8231489123/ShareGPT_Vicuna_unfiltered |
Summarize the dataset and highlight its key characteristics, intended use, and notable metadata. | Name: PhysicalAI-Autonomous-Vehicles
Description: PHYSICAL AI AUTONOMOUS VEHICLES. The PhysicalAI-Autonomous-Vehicles dataset provides one of the largest, geographically diverse collections of multi-sensor data empowering AV researchers to build the next generation of Physical AI based end-to-end driving systems. This ... | PhysicalAI-Autonomous-Vehicles is a Hugging Face dataset for autonomous vehicle research, offering a large and geographically diverse multi-sensor driving data collection. It is designed to support development of next-generation Physical AI and end-to-end driving systems. The dataset uses automatic or sensor-based data... | huggingface | nvidia/PhysicalAI-Autonomous-Vehicles |
Summarize the dataset in a concise SFT-friendly format, highlighting what it is, scale, modality, language coverage, license, and notable metadata. | {"source":"huggingface","id":"HuggingFaceFW/finepdfs","name":"finepdfs","description":"Liberating 3T of the finest tokens from PDFs. FinePDFs is the largest publicly available corpus sourced exclusively from PDFs, containing about 3 trillion tokens across 475 million documents in 1733 languages.","tags":["task_categori... | finepdfs is a Hugging Face dataset built exclusively from PDF sources. It is a massive multilingual text corpus with about 3 trillion tokens across 475 million documents in 1733 languages. The dataset is intended for text-generation-related use cases and is distributed in Parquet format. It is tagged with both text and... | huggingface | HuggingFaceFW/finepdfs |
Summarize the dataset in a concise SFT-friendly format, highlighting purpose, domains, size, format, license, and notable metadata. | {"source":"huggingface","id":"open-thoughts/OpenThoughts-114k","name":"OpenThoughts-114k","description":"Open synthetic reasoning dataset with 114k high-quality examples covering math, science, code, and puzzles. Default subset contains ready-to-train data used to finetune the OpenThinker-7B and OpenThinker-32B models.... | OpenThoughts-114k is a Hugging Face synthetic reasoning dataset containing 114k high-quality training examples. It covers multiple domains including math, science, code, and puzzles, and includes a default ready-to-train subset used for finetuning OpenThinker-7B and OpenThinker-32B. The dataset is text-only, stored in ... | huggingface | open-thoughts/OpenThoughts-114k |
Summarize the following dataset for LLM supervised fine-tuning use, including its purpose, scale, notable characteristics, and popularity signals. | Name: OpenHermes-2.5
Source: huggingface
Dataset ID: teknium/OpenHermes-2.5
Description: This is the dataset that made OpenHermes 2.5 and Nous Hermes 2 series of models. The Open Hermes 2/2.5 and Nous Hermes 2 models have made significant advancements of SOTA LLM's over recent months, and are underpinned by this exact ... | OpenHermes-2.5 is a large English text dataset on Hugging Face used to train the OpenHermes 2.5 and Nous Hermes 2 model series. It is a compilation of open-source and synthetic data, with tags indicating GPT-4-based distillation and dataset curation. The dataset is provided in JSON format, falls in the 1M to 10M sample... | huggingface | teknium/OpenHermes-2.5 |
Summarize the following dataset for LLM supervised fine-tuning use, including its purpose, scale, notable characteristics, and popularity signals. | Name: alpaca-cleaned
Source: huggingface
Dataset ID: yahma/alpaca-cleaned
Description: This is a cleaned version of the original Alpaca Dataset released by Stanford. The following issues have been identified in the original release and fixed in this dataset: Hallucinations: Many instructions in the original dataset had... | alpaca-cleaned is an English instruction-finetuning dataset on Hugging Face derived from Stanford's original Alpaca release. It focuses on improving data quality by cleaning problematic examples, including instructions that encouraged hallucinated answers based on unavailable internet information. The dataset is format... | huggingface | yahma/alpaca-cleaned |
Summarize the dataset metadata into a concise description for supervised fine-tuning. | Name: hle
Source: huggingface
Dataset ID: cais/hle
Description: Humanity's Last Exam (HLE) is a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. It consists of 2,500 questions across dozens of subjects. The datase... | HLE is an official Hugging Face benchmark dataset from CAIS focused on frontier-level academic evaluation. It is multimodal, combining text and image inputs, and contains about 2,500 closed-ended questions spanning many subjects. The dataset is distributed in Parquet format, supports common data libraries such as datas... | huggingface | cais/hle |
Summarize the dataset metadata into a concise description for supervised fine-tuning. | Name: fineweb-2
Source: huggingface
Dataset ID: HuggingFaceFW/fineweb-2
Description: FineWeb2 is the second iteration of the FineWeb dataset and provides high-quality pretraining data for more than 1000 languages. It is fully reproducible, available under the ODC-By 1.0 license, and validated through hundreds of ablati... | FineWeb2 is a large-scale multilingual text dataset on Hugging Face intended for language model pretraining and text generation. It is the second version of FineWeb and expands coverage to more than 1,000 languages. The dataset is described as fully reproducible and extensively validated through large numbers of ablati... | huggingface | HuggingFaceFW/fineweb-2 |
Given a raw dataset catalog entry, summarize the dataset for supervised fine-tuning use. Include the dataset name, source, main purpose, modality, likely tasks, notable scale or popularity signals, languages if available, license if available, and a short caution about usage limits or ambiguity. | {"source":"huggingface","id":"ILSVRC/imagenet-1k","name":"imagenet-1k","description":"ILSVRC 2012, commonly known as 'ImageNet' is an image dataset organized according to the WordNet hierarchy. Each meaningful concept in WordNet, possibly described by multiple words or word phrases, is called a synset. ImageNet aims to... | ImageNet-1k is a Hugging Face dataset entry for the ILSVRC 2012 image benchmark. Its primary purpose is multi-class image classification, using images mapped to WordNet synsets. The modality is image, and the dataset is large scale, with tags indicating between 1M and 10M examples. It is widely used and highly popular,... | huggingface | ILSVRC/imagenet-1k |
Given a raw dataset catalog entry, summarize the dataset for supervised fine-tuning use. Include the dataset name, source, main purpose, modality, likely tasks, notable scale or popularity signals, languages if available, license if available, and a short caution about usage limits or ambiguity. | {"source":"huggingface","id":"QingyiSi/Alpaca-CoT","name":"Alpaca-CoT","description":"Instruction-Finetuning Dataset Collection (Alpaca-CoT). This repository continuously collects various instruction tuning datasets and standardizes them into the same format, which can be directly loaded by Alpaca-style code.","tags":[... | Alpaca-CoT is a Hugging Face dataset collection designed for instruction fine-tuning. Its main purpose is to aggregate and standardize multiple instruction-tuning datasets into a unified format suitable for Alpaca-style training. The likely modality is text, and the likely tasks include instruction following and chain-... | huggingface | QingyiSi/Alpaca-CoT |
Summarize the dataset and identify its main characteristics, including purpose, supported tasks, languages, license, format, and notable themes. | Dataset name: PersonaHub
Description: Scaling Synthetic Data Creation with 1,000,000,000 Personas. This repository releases data introduced in the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas." It proposes a persona-driven data synthesis methodology that leverages various perspectives within a lar... | PersonaHub is a Hugging Face dataset focused on large-scale synthetic data creation through persona-driven generation. It is associated with the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas" and introduces PERSONA HUB, a massive collection of diverse personas curated from web data to support scala... | huggingface | proj-persona/PersonaHub |
Recommend a public dataset for collecting and studying AI prompts shared by a community. | Need a Hugging Face dataset focused on community-created prompts for ChatGPT and other LLMs. Prefer open licensing and prompt-engineering relevance. | Dataset: prompts.chat
Source: Hugging Face
ID: fka/prompts.chat
Summary: A mirror of prompts.chat, also known as Awesome ChatGPT Prompts, an open-source community platform for sharing, discovering, and collecting AI prompts.
Key details:
- Tasks: question-answering, text-generation
- License: CC0-1.0
- Size: 1K<n<10K
-... | huggingface | fka/prompts.chat |
Identify a massive English web-text dataset suitable for large-scale LLM pretraining. | Need a very large cleaned and deduplicated English web corpus from Common Crawl for language model pretraining. | Dataset: fineweb
Source: Hugging Face
ID: HuggingFaceFW/fineweb
Summary: FineWeb is a large-scale English web dataset derived from CommonCrawl, cleaned and deduplicated, and optimized for LLM performance.
Key details:
- Tokens: more than 18.5T
- Language: English
- License: ODC-BY
- Size category: 10B<n<100B
- Modaliti... | huggingface | HuggingFaceFW/fineweb |
Select a dataset for training reward models using human preference comparisons focused on helpfulness and harmlessness. | Looking for RLHF preference data, especially human feedback comparing responses for helpfulness and harmlessness. Not primarily for supervised fine-tuning. | Dataset: hh-rlhf
Source: Hugging Face
ID: Anthropic/hh-rlhf
Summary: HH-RLHF provides human preference data about helpfulness and harmlessness, intended for training preference or reward models in RLHF pipelines.
Key details:
- License: MIT
- Size: 100K<n<1M
- Format: JSON
- Modality: text
- Tag: human-feedback
- Downl... | huggingface | Anthropic/hh-rlhf |
Recommend a large instruction-tuning dataset aligned with FLAN-style augmented data for general NLP tasks. | Need a broad English instruction dataset useful for supervised fine-tuning across tasks like QA, summarization, classification, and generation. | Dataset: OpenOrca
Source: Hugging Face
ID: Open-Orca/OpenOrca
Summary: OpenOrca is a large instruction-style dataset built from augmented FLAN data and designed to align with distributions described in the Orca paper.
Key details:
- Language: English
- License: MIT
- Size: 1M<n<10M
- Format: Parquet
- Modality: text
- ... | huggingface | Open-Orca/OpenOrca |
Find a multilingual assistant conversation dataset with human annotations and quality ratings for alignment research. | Need a conversation dataset with assistant-style dialogues, multiple languages, and human ratings for quality or alignment work. | Dataset: oasst1
Source: Hugging Face
ID: OpenAssistant/oasst1
Summary: OASST1 is the OpenAssistant Conversations Dataset, a multilingual assistant-style conversation corpus with human-generated messages and human annotations.
Key details:
- 161,443 messages
- 35 languages including en, es, de, fr, zh, ja, ko, ar, ru an... | huggingface | OpenAssistant/oasst1 |
Summarize the dataset and highlight its main purpose, domain, scale, language, license, and notable usage signals. | {"source":"huggingface","id":"openai/gsm8k","name":"gsm8k","description":"Dataset Card for GSM8K\n\nDataset Summary\n\nGSM8K (Grade School Math 8K) is a dataset of 8.5K high quality linguistically diverse grade school math word problems. The dataset was created to support the task of question answering on basic mathema... | GSM8K is a Hugging Face dataset of grade-school math word problems designed for multi-step reasoning and question answering. It contains about 8.5K high-quality English problems, typically requiring 2 to 8 reasoning steps and basic arithmetic operations. The dataset is monolingual English, text-based, distributed in pa... | huggingface | openai/gsm8k |
Summarize the dataset and highlight its main purpose, modality, multilingual scope, scale, licensing, and notable usage signals. | {"source":"huggingface","id":"wikimedia/wikipedia","name":"wikipedia","description":"Dataset Card for Wikimedia Wikipedia\n\nDataset Summary\n\nWikipedia dataset containing cleaned articles of all languages. The dataset is built from the Wikipedia dumps with one subset per language, each containing a single train split... | Wikimedia Wikipedia is a large multilingual text dataset containing cleaned full Wikipedia articles derived from official Wikipedia dumps. It is organized with one subset per language and is suitable for text generation, language modeling, fill-mask, and masked language modeling tasks. The dataset is extremely large, t... | huggingface | wikimedia/wikipedia |
Summarize the dataset or asset and explain its likely use case, modality, scale, licensing, and popularity. | {"source":"huggingface","id":"gsdf/EasyNegative","name":"EasyNegative","description":"Negative Embedding\n\nThis is a Negative Embedding trained with Counterfeit. Please use it in the stable-diffusion-webui embeddings folder. It can be used with other models, but the effectiveness is not certain.","tags":["license:othe... | EasyNegative appears to be a small image-related asset published on Hugging Face, described as a negative embedding intended for use with Stable Diffusion WebUI. Its main use case is likely prompt conditioning or quality control in image generation workflows rather than conventional supervised learning. The asset is ve... | huggingface | gsdf/EasyNegative |
Summarize the dataset and highlight its purpose, language, scale, modality, openness, and popularity. | {"source":"huggingface","id":"togethercomputer/RedPajama-Data-1T","name":"RedPajama-Data-1T","description":"RedPajama is a clean-room, fully open-source implementation of the LLaMa dataset.","tags":["task_categories:text-generation","language:en","size_categories:1M<n<10M","modality:text","library:datasets","library:ml... | RedPajama-Data-1T is an English text dataset intended for text generation and positioned as a clean-room, fully open-source recreation of the LLaMa training data. It is relevant for large-scale language model pretraining or related generative NLP work. The dataset is tagged in the 1M<n<10M size category and has text mo... | huggingface | togethercomputer/RedPajama-Data-1T |
Summarize the dataset and highlight its task focus, domain, language coverage, scale, licensing, and likely training use. | {"source":"huggingface","id":"FreedomIntelligence/medical-o1-reasoning-SFT","name":"medical-o1-reasoning-SFT","description":"News\n\n[2025/04/22] We split the data and kept only the medical SFT dataset (medical_o1_sft.json). The file medical_o1_sft_mix.json contains a mix of medical and general instruction data. [2025/... | medical-o1-reasoning-SFT is a bilingual English and Chinese text dataset focused on medical reasoning for supervised fine-tuning. It supports question answering and text generation, and is built around medical verifiable problems with reasoning chains, including distilled data derived from Deepseek-R1. The dataset is i... | huggingface | FreedomIntelligence/medical-o1-reasoning-SFT |
Summarize the dataset for LLM pretraining use, including scale, domain, language, license, and notable metadata. | Name: fineweb-edu
Description: FineWeb-Edu is a large educational web dataset containing 1.3 trillion tokens filtered from FineWeb using an educational quality classifier. It is intended for educational-text-focused language model training.
Tags: task_categories:text-generation, language:en, license:odc-by, size_catego... | fineweb-edu is an English educational text dataset from Hugging Face designed for text generation and LLM pretraining. It contains about 1.3 trillion tokens of high-quality educational web content filtered from FineWeb using a quality classifier. The dataset is distributed in Parquet format, supports text and tabular m... | huggingface | HuggingFaceFW/fineweb-edu |
Create a concise dataset card summary suitable for supervised fine-tuning metadata extraction. | Name: dolma
Description: Dolma is an open corpus of three trillion tokens for language model pretraining research.
Tags: task_categories:text-generation, language:en, license:odc-by, size_categories:n>1T, arxiv:2402.00159, language-modeling, casual-lm, llm
Downloads: 3226
Likes: 1023 | dolma is an English open corpus for language model pretraining research. It is aimed at text generation and language modeling, with a very large scale of three trillion tokens. The dataset is licensed under ODC-BY and is associated with large-scale LLM and causal language modeling research. It has 3,226 downloads and 1... | huggingface | allenai/dolma |
Summarize this code dataset with emphasis on content type, licensing characteristics, multilingual scope, and approximate scale. | Name: the-stack
Description: The Stack is a large code dataset. Initial release included 30 programming languages and 18 permissive licenses, with a near-deduplicated size of 3TB. Later versions excluded weak copyleft licenses and expanded permissive licenses.
Tags: task_categories:text-generation, multilinguality:mult... | the-stack is a multilingual code dataset intended for text generation and code model training. It contains source code across many programming languages and was released in near-deduplicated form at roughly 3TB scale. Its licensing is heterogeneous, with an emphasis on permissive licenses in later releases after exclud... | huggingface | bigcode/the-stack |
Write a short structured summary of the dataset highlighting generation method, target use case, language, and size. | Name: TinyStories
Description: TinyStories contains synthetically generated short stories created by GPT-3.5 and GPT-4 using a small vocabulary. It is described in arXiv:2305.07759 and used to train small language models.
Tags: task_categories:text-generation, language:en, license:cdla-sharing-1.0, size_categories:1M<n... | TinyStories is an English text generation dataset made up of synthetic short stories produced by GPT-3.5 and GPT-4. It is designed for training and evaluating small language models on simple vocabulary narratives. The dataset size falls in the 1M to 10M range, is distributed in Parquet format, and uses the CDLA-Sharing... | huggingface | roneneldan/TinyStories |
Produce an instruction-tuning dataset summary covering tasks, provenance, licensing, format, and likely use case. | Name: databricks-dolly-15k
Description: databricks-dolly-15k is an open source dataset of instruction-following records generated by thousands of Databricks employees across categories such as brainstorming, classification, closed QA, generation, information extraction, open QA, and summarization.
Tags: task_categories... | databricks-dolly-15k is an English instruction-following dataset built from records created by Databricks employees. It covers multiple supervised tasks including question answering, summarization, brainstorming, classification, generation, and information extraction, making it well suited for instruction tuning. The d... | huggingface | databricks/databricks-dolly-15k |
Summarize the given dataset entry and describe what it is used for. | Name: alpaca
Source: huggingface
ID: tatsu-lab/alpaca
Description: Dataset Card for Alpaca. Alpaca is a dataset of 52,000 instructions and demonstrations generated by OpenAI's text-davinci-003 engine. This instruction data can be used to conduct instruction-tuning for language models and make the language model follow ... | Alpaca is a Hugging Face text dataset for instruction fine-tuning. It contains about 52,000 instructions and demonstrations generated with text-davinci-003 and is intended to improve how well language models follow instructions. It is an English dataset in parquet format, tagged for text generation and instruction-fine... | huggingface | tatsu-lab/alpaca |
Summarize the given dataset entry and describe what it is used for. | Name: bad_prompt
Source: huggingface
ID: Nerfgun3/bad_prompt
Description: Negative Embedding / Textual Inversion. The idea behind this embedding was to train the negative prompt as an embedding, unifying the basis of the negative prompt into one word or embedding. The embedding has proven helpful for generation of hand... | bad_prompt is a Hugging Face image-related resource associated with Stable Diffusion and textual inversion. It is designed as a negative embedding that can be used in prompts to improve image generation, especially for issues like hand rendering. It is a small English dataset/resource in imagefolder format, tagged for ... | huggingface | Nerfgun3/bad_prompt |
Summarize the dataset based on its metadata, including its purpose, content, scale, language, license, format, and notable characteristics. | Name: falcon-refinedweb
Source: huggingface
ID: tiiuae/falcon-refinedweb
Description: Falcon RefinedWeb is a massive English web dataset built by TII and released under an ODC-By 1.0 license. RefinedWeb is built through stringent filtering and large-scale deduplication of CommonCrawl; models trained on it achieve perfo... | Falcon RefinedWeb is a large-scale English web text dataset released by TII on Hugging Face. It is intended for text generation and language model pretraining. The dataset is derived from CommonCrawl using strict filtering and large-scale deduplication, with the goal of providing high-quality web data at massive scale.... | huggingface | tiiuae/falcon-refinedweb |
Summarize the dataset based on its metadata, including its purpose, content, scale, provenance, format, and notable characteristics. | Name: lmsys-chat-1m
Source: huggingface
ID: lmsys/lmsys-chat-1m
Description: LMSYS-Chat-1M is a large-scale real-world LLM conversation dataset containing one million conversations with 25 state-of-the-art LLMs. It was collected from 210K unique IP addresses in the wild on the Vicuna demo and Chatbot Arena website from... | LMSYS-Chat-1M is a real-world conversation dataset hosted on Hugging Face that contains one million chat sessions involving 25 state-of-the-art large language models. It was collected from activity on the Vicuna demo and Chatbot Arena websites between April and August 2023, covering interactions from 210,000 unique IP ... | huggingface | lmsys/lmsys-chat-1m |
Summarize the dataset and highlight its key characteristics, intended use, and notable metadata. | Name: ShareGPT_Vicuna_unfiltered
Description: Further cleaning done. Please look through the dataset and ensure that I didn't miss anything. Update: Confirmed working method for training the model. Two choices: removes instances of "I'm sorry, but" or keeps them. See the dataset page for full details.
Tags: language:en... | ShareGPT_Vicuna_unfiltered is a Hugging Face dataset focused on English-language conversational data, likely intended for training or fine-tuning chat models such as Vicuna-style assistants. The description indicates that the dataset has undergone additional cleaning and offers two variants: one with occurrences of the... | huggingface | anon8231489123/ShareGPT_Vicuna_unfiltered |
Summarize the dataset and highlight its key characteristics, intended use, and notable metadata. | Name: PhysicalAI-Autonomous-Vehicles
Description: PHYSICAL AI AUTONOMOUS VEHICLES. The PhysicalAI-Autonomous-Vehicles dataset provides one of the largest, geographically diverse collections of multi-sensor data empowering AV researchers to build the next generation of Physical AI based end-to-end driving systems. This ... | PhysicalAI-Autonomous-Vehicles is a Hugging Face dataset for autonomous vehicle research, offering a large and geographically diverse multi-sensor driving data collection. It is designed to support development of next-generation Physical AI and end-to-end driving systems. The dataset uses automatic or sensor-based data... | huggingface | nvidia/PhysicalAI-Autonomous-Vehicles |
Summarize the dataset in a concise SFT-friendly format, highlighting what it is, scale, modality, language coverage, license, and notable metadata. | {"source":"huggingface","id":"HuggingFaceFW/finepdfs","name":"finepdfs","description":"Liberating 3T of the finest tokens from PDFs. FinePDFs is the largest publicly available corpus sourced exclusively from PDFs, containing about 3 trillion tokens across 475 million documents in 1733 languages.","tags":["task_categori... | finepdfs is a Hugging Face dataset built exclusively from PDF sources. It is a massive multilingual text corpus with about 3 trillion tokens across 475 million documents in 1733 languages. The dataset is intended for text-generation-related use cases and is distributed in Parquet format. It is tagged with both text and... | huggingface | HuggingFaceFW/finepdfs |
Summarize the dataset in a concise SFT-friendly format, highlighting purpose, domains, size, format, license, and notable metadata. | {"source":"huggingface","id":"open-thoughts/OpenThoughts-114k","name":"OpenThoughts-114k","description":"Open synthetic reasoning dataset with 114k high-quality examples covering math, science, code, and puzzles. Default subset contains ready-to-train data used to finetune the OpenThinker-7B and OpenThinker-32B models.... | OpenThoughts-114k is a Hugging Face synthetic reasoning dataset containing 114k high-quality training examples. It covers multiple domains including math, science, code, and puzzles, and includes a default ready-to-train subset used for finetuning OpenThinker-7B and OpenThinker-32B. The dataset is text-only, stored in ... | huggingface | open-thoughts/OpenThoughts-114k |
Summarize the following dataset for LLM supervised fine-tuning use, including its purpose, scale, notable characteristics, and popularity signals. | Name: OpenHermes-2.5
Source: huggingface
Dataset ID: teknium/OpenHermes-2.5
Description: This is the dataset that made OpenHermes 2.5 and Nous Hermes 2 series of models. The Open Hermes 2/2.5 and Nous Hermes 2 models have made significant advancements of SOTA LLM's over recent months, and are underpinned by this exact ... | OpenHermes-2.5 is a large English text dataset on Hugging Face used to train the OpenHermes 2.5 and Nous Hermes 2 model series. It is a compilation of open-source and synthetic data, with tags indicating GPT-4-based distillation and dataset curation. The dataset is provided in JSON format, falls in the 1M to 10M sample... | huggingface | teknium/OpenHermes-2.5 |
Summarize the following dataset for LLM supervised fine-tuning use, including its purpose, scale, notable characteristics, and popularity signals. | Name: alpaca-cleaned
Source: huggingface
Dataset ID: yahma/alpaca-cleaned
Description: This is a cleaned version of the original Alpaca Dataset released by Stanford. The following issues have been identified in the original release and fixed in this dataset: Hallucinations: Many instructions in the original dataset had... | alpaca-cleaned is an English instruction-finetuning dataset on Hugging Face derived from Stanford's original Alpaca release. It focuses on improving data quality by cleaning problematic examples, including instructions that encouraged hallucinated answers based on unavailable internet information. The dataset is format... | huggingface | yahma/alpaca-cleaned |
Summarize the dataset metadata into a concise description for supervised fine-tuning. | Name: hle
Source: huggingface
Dataset ID: cais/hle
Description: Humanity's Last Exam (HLE) is a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. It consists of 2,500 questions across dozens of subjects. The datase... | HLE is an official Hugging Face benchmark dataset from CAIS focused on frontier-level academic evaluation. It is multimodal, combining text and image inputs, and contains about 2,500 closed-ended questions spanning many subjects. The dataset is distributed in Parquet format, supports common data libraries such as datas... | huggingface | cais/hle |
Summarize the dataset metadata into a concise description for supervised fine-tuning. | Name: fineweb-2
Source: huggingface
Dataset ID: HuggingFaceFW/fineweb-2
Description: FineWeb2 is the second iteration of the FineWeb dataset and provides high-quality pretraining data for more than 1000 languages. It is fully reproducible, available under the ODC-By 1.0 license, and validated through hundreds of ablati... | FineWeb2 is a large-scale multilingual text dataset on Hugging Face intended for language model pretraining and text generation. It is the second version of FineWeb and expands coverage to more than 1,000 languages. The dataset is described as fully reproducible and extensively validated through large numbers of ablati... | huggingface | HuggingFaceFW/fineweb-2 |
Given a raw dataset catalog entry, summarize the dataset for supervised fine-tuning use. Include the dataset name, source, main purpose, modality, likely tasks, notable scale or popularity signals, languages if available, license if available, and a short caution about usage limits or ambiguity. | {"source":"huggingface","id":"ILSVRC/imagenet-1k","name":"imagenet-1k","description":"ILSVRC 2012, commonly known as 'ImageNet' is an image dataset organized according to the WordNet hierarchy. Each meaningful concept in WordNet, possibly described by multiple words or word phrases, is called a synset. ImageNet aims to... | ImageNet-1k is a Hugging Face dataset entry for the ILSVRC 2012 image benchmark. Its primary purpose is multi-class image classification, using images mapped to WordNet synsets. The modality is image, and the dataset is large scale, with tags indicating between 1M and 10M examples. It is widely used and highly popular,... | huggingface | ILSVRC/imagenet-1k |
Given a raw dataset catalog entry, summarize the dataset for supervised fine-tuning use. Include the dataset name, source, main purpose, modality, likely tasks, notable scale or popularity signals, languages if available, license if available, and a short caution about usage limits or ambiguity. | {"source":"huggingface","id":"QingyiSi/Alpaca-CoT","name":"Alpaca-CoT","description":"Instruction-Finetuning Dataset Collection (Alpaca-CoT). This repository continuously collects various instruction tuning datasets and standardizes them into the same format, which can be directly loaded by Alpaca-style code.","tags":[... | Alpaca-CoT is a Hugging Face dataset collection designed for instruction fine-tuning. Its main purpose is to aggregate and standardize multiple instruction-tuning datasets into a unified format suitable for Alpaca-style training. The likely modality is text, and the likely tasks include instruction following and chain-... | huggingface | QingyiSi/Alpaca-CoT |
Summarize the dataset and identify its main characteristics, including purpose, supported tasks, languages, license, format, and notable themes. | Dataset name: PersonaHub
Description: Scaling Synthetic Data Creation with 1,000,000,000 Personas. This repository releases data introduced in the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas." It proposes a persona-driven data synthesis methodology that leverages various perspectives within a lar... | PersonaHub is a Hugging Face dataset focused on large-scale synthetic data creation through persona-driven generation. It is associated with the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas" and introduces PERSONA HUB, a massive collection of diverse personas curated from web data to support scala... | huggingface | proj-persona/PersonaHub |
SoloAI SFT Dataset: 20260428-2139
📊 数据集概览
| 属性 | 值 |
|---|---|
| 数据条数 | 30 条 |
| 数据源 | huggingface |
| 生成时间 | 2026-04-28 21:39 |
| 格式 | Instruction-Input-Output |
| 主要用途 | SFT 微调 / 指令调优 |
🎯 适用场景
- LLM 指令微调 (Instruction Tuning): 适用于训练对话型 AI 助手
- Prompt Engineering 研究: 可用于分析和优化提示词设计
- 多语言支持: 包含英文和中文数据
📝 数据格式
每条数据包含三个字段：
instruction: 任务指令（描述需要完成的任务）input: 输入上下文（提供任务相关的背景信息）output: 期望输出（模型应生成的回答）
示例
{'instruction': 'Recommend a public dataset for collecting and studying AI prompts shared by a community.', 'input': 'Need a Hugging Face dataset focused on community-created prompts for ChatGPT and other LLMs. Prefer open licensing and prompt-engineering relevance.', 'output': 'Dataset: prompts.chat\nSource: Hugging Face\nID: fka/prompts.chat\nSummary: A mirror of prompts.chat, also known as Awesome ChatGPT Prompts, an open-source community platform for sharing, discovering, and collecting AI prompts.\nKey details:\n- Tasks: question-answering, text-generation\n- License: CC0-1.0\n- Size: 1K<n<10K\n- Format: CSV\n- Modality: text\n- Tags: ChatGPT, prompts, GPT, Claude, Gemini, Llama, Mistral, LLM, prompt-engineering, conversational-ai, chatbot\n- Downloads: 56882\n- Likes: 9680\nUse case: Good for prompt libraries, prompt engineering research, and building prompt recommendation or retrieval systems.', 'source': 'huggingface', 'original_id': 'fka/prompts.chat'}
🤖 数据来源
本数据集由 SoloAI 自动化数据管道生成：
- 从 HuggingFace Datasets Hub 发现高质量数据集
- AI 清洗为 SFT 格式（Instruction-Input-Output）
- 质量过滤后发布
⚠️ 使用说明 & 📬 商务联系
- 本数据集仅供研究和实验用途
- 请遵守原始数据的许可证要求
- 商业用途 / 定制数据 / 深度合作:
- 📧 请联系:
379744050@qq.com - 🤖 SoloAI 提供高质量 SFT 数据定制服务。
- 建议邮件标题:
【数据定制咨询】行业 + 数据类型 + 规模 - 建议正文包含: 目标用途、需要条数、语言、字段格式、预算、交付时间
- 📧 请联系:
💰 商业合作报价
| 套餐 | 价格 | 说明 |
|---|---|---|
| Starter | $199 / 1000条高质量 SFT 数据 | 适合个人开发者 / 小团队 |
| Growth | $499 / 5000条行业数据 | 适合垂直行业训练数据 |
| Enterprise | $1499 / 定制领域数据管道 | 适合长期定制与数据管道 |
💳 支付方式
- 中国客户: 支付宝, 微信支付
- 海外客户: PayPal, USDT (TRC20)
- 下单方式: 邮件联系后 24 小时内提供交付方案与付款指引
🚀 为什么现在联系 SoloAI
- 24 小时内响应有效询盘
- 报价前可免费给出需求范围建议
- 支持中文 / English 项目合作
- 可从单次交付升级为长期数据管道合作
📈 更新日志
| 版本 | 日期 | 说明 |
|---|---|---|
| v1.0 | 2026-04-28 21:39 | 初始发布，30 条数据 |
- Downloads last month
- 11