instruction stringlengths 49 142 | input stringlengths 238 1.04k | output stringlengths 325 926 | source stringclasses 1
value | original_id stringlengths 8 44 |
|---|---|---|---|---|
Summarize the dataset and explain why it could be useful for supervised fine-tuning or prompt engineering research. | Dataset name: prompts.chat
Source: huggingface
Dataset ID: fka/prompts.chat
Description: a.k.a. Awesome ChatGPT Prompts. This is a Dataset Repository mirror of prompts.chat — a social platform for AI prompts. prompts.chat is an open-source platform where users can share, discover, and collect AI prompts from the commun... | prompts.chat is a community prompt collection mirrored on Hugging Face. It contains prompt examples oriented around question answering, text generation, chatbot use, and prompt engineering across major LLM ecosystems such as ChatGPT, Claude, Gemini, Llama, and Mistral. Because it is licensed under CC0-1.0 and distribut... | huggingface | fka/prompts.chat |
Provide a concise dataset card style summary, including scale, content type, and whether it is appropriate for direct supervised fine-tuning. | Dataset name: fineweb
Source: huggingface
Dataset ID: HuggingFaceFW/fineweb
Description: FineWeb is 15 trillion tokens of cleaned and deduplicated English web data from CommonCrawl, later expanded to more than 18.5T tokens. The processing pipeline is optimized for LLM performance and ran on the datatrove library.
Tags:... | FineWeb is a massive English web corpus containing over 18.5 trillion cleaned and deduplicated tokens derived from CommonCrawl. It is designed primarily as large-scale pretraining data for language models, not as an instruction-response dataset. Its strengths are scale, broad web coverage, and preprocessing optimized f... | huggingface | HuggingFaceFW/fineweb |
Explain what kind of alignment data this dataset contains and how it should and should not be used in LLM training. | Dataset name: hh-rlhf
Source: huggingface
Dataset ID: Anthropic/hh-rlhf
Description: This repository provides human preference data about helpfulness and harmlessness from Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback. These data are meant to train preference or reward models... | HH-RLHF contains human preference data centered on helpfulness and harmlessness, making it a core alignment dataset for reward modeling and preference learning. Its main use case is training reward models or preference models that can be applied in RLHF pipelines. It should not be treated as a standard supervised chat ... | huggingface | Anthropic/hh-rlhf |
Summarize the dataset in terms of task coverage, training value, and likely SFT use cases. | Dataset name: OpenOrca
Source: huggingface
Dataset ID: Open-Orca/OpenOrca
Description: The OpenOrca dataset is a rich collection of augmented FLAN data aligned as closely as possible with the distributions described in the Orca paper. It has been used to generate strong model checkpoints and is positioned as a valuable... | OpenOrca is a large English instruction-style dataset built from augmented FLAN data and shaped to resemble the distributions discussed in the Orca work. It spans many task types, including question answering, summarization, classification, table QA, feature extraction, and text generation. This broad task coverage mak... | huggingface | Open-Orca/OpenOrca |
Describe the dataset’s structure, multilingual properties, and why it is relevant for chat model fine-tuning and alignment. | Dataset name: oasst1
Source: huggingface
Dataset ID: OpenAssistant/oasst1
Description: OpenAssistant Conversations Dataset (OASST1) is a human-generated, human-annotated assistant-style conversation corpus with 161,443 messages in 35 languages, annotated with 461,292 quality ratings and more than 10,000 fully annotated... | OASST1 is a multilingual assistant conversation dataset containing human-written dialogue messages organized into conversation trees, along with extensive human quality annotations. Its 35-language coverage makes it especially valuable for multilingual chat model development, while the tree structure and ratings suppor... | huggingface | OpenAssistant/oasst1 |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: gsm8k
Description: Dataset Card for GSM8K. GSM8K (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 mathematical problems that require multi-step reasoning. These problems take ... | GSM8K is an English text dataset for grade-school math word problem solving. It is widely used as a benchmark for question answering and text generation tasks that require multi-step numerical reasoning. The dataset contains roughly 8.5K examples, placing it in the 10K to 100K size category. Its main strength for LLM t... | huggingface | openai/gsm8k |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: wikipedia
Description: Dataset Card for Wikimedia Wikipedia. Wikipedia 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. Each example contains the content of one full Wikipedia article with cleani... | Wikimedia Wikipedia is a large-scale multilingual text corpus of cleaned Wikipedia articles. It is suitable for text generation, language modeling, and masked language modeling tasks, with one subset per language and article-level examples. The dataset is extremely large, in the 10M to 100M example range, and spans a v... | huggingface | wikimedia/wikipedia |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: EasyNegative
Description: Negative Embedding. This 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:other, size_categories:n<1K, format:imagefolder, modality:image
... | EasyNegative is an image-focused asset intended for Stable Diffusion workflows rather than a conventional LLM supervised fine-tuning dataset. It appears to be a negative embedding resource used in image generation pipelines, with very small scale and image modality metadata. Because it is not a text instruction dataset... | huggingface | gsdf/EasyNegative |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | 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
Downloads: 2084
Likes: 1151 | RedPajama-Data-1T is a large English text corpus designed as an open-source reproduction of the data philosophy behind LLaMA-style pretraining. It is intended for text generation and large-scale language model pretraining rather than narrow task supervision. The dataset is text-only and falls into a very large scale ca... | huggingface | togethercomputer/RedPajama-Data-1T |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: medical-o1-reasoning-SFT
Description: News: [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. [2025/02/22] We released the distilled dataset from Deepseek-R1 based on medical verifiab... | medical-o1-reasoning-SFT is a bilingual English-Chinese medical reasoning dataset designed specifically for supervised fine-tuning. It supports question answering and text generation, with a focus on verifiable medical problems and distilled reasoning traces. The dataset is in JSON format, text-only, and medium-sized i... | huggingface | FreedomIntelligence/medical-o1-reasoning-SFT |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: HuggingFaceFW/fineweb-edu
Name: fineweb-edu
Description: 1.3 trillion tokens of educational web data filtered from FineWeb using an educational quality classifier.
Tags: task_categories:text-generation, language:en, license:odc-by, size_categories:1B<n<10B, format:parquet, modality:tabul... | fineweb-edu is a large English educational text dataset on Hugging Face containing 1.3 trillion tokens filtered from FineWeb for high educational quality. It is intended for text generation and language model pretraining research, distributed in parquet format with text and tabular modalities under the ODC-By license. ... | huggingface | HuggingFaceFW/fineweb-edu |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: allenai/dolma
Name: dolma
Description: Dolma: 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, arxiv:2301.13688, region:us, language-modeling, casual... | dolma is an English open corpus for language model pretraining research containing three trillion tokens. It is aimed at text generation and language modeling use cases, published under the ODC-By license, and associated with LLM and causal language modeling research. The dataset has 3,026 downloads and 1,021 likes on ... | huggingface | allenai/dolma |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: bigcode/the-stack
Name: the-stack
Description: The Stack is a near-deduplicated 3TB source code dataset with permissive licenses across many programming languages.
Tags: task_categories:text-generation, language_creators:crowdsourced, language_creators:expert-generated, multilinguality:m... | the-stack is a multilingual code dataset designed for text generation and code model training. It contains near-deduplicated source code spanning many programming languages, released in parquet format with text and tabular modalities. The dataset emphasizes permissive licensing coverage and has 11,947 downloads with 98... | huggingface | bigcode/the-stack |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: roneneldan/TinyStories
Name: TinyStories
Description: Dataset containing synthetically generated short stories by GPT-3.5 and GPT-4 using a small vocabulary.
Tags: task_categories:text-generation, language:en, license:cdla-sharing-1.0, size_categories:1M<n<10M, format:parquet, modality:t... | TinyStories is an English text generation dataset of synthetically generated short stories created by GPT-3.5 and GPT-4 using a restricted vocabulary. It is useful for language modeling experiments focused on simple narrative text and is distributed in parquet format under the CDLA-Sharing-1.0 license. The dataset has ... | huggingface | roneneldan/TinyStories |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: databricks/databricks-dolly-15k
Name: databricks-dolly-15k
Description: Open source instruction-following dataset generated by Databricks employees across tasks such as brainstorming, classification, QA, generation, information extraction, and summarization.
Tags: task_categories:questio... | databricks-dolly-15k is an English instruction-following dataset containing about 15,000 human-written records created by Databricks employees. It covers tasks such as question answering, summarization, classification, brainstorming, and information extraction, and is distributed in JSON format under the CC-BY-SA-3.0 l... | huggingface | databricks/databricks-dolly-15k |
Summarize the dataset based on its metadata and description. | Name: alpaca
Source: huggingface
ID: tatsu-lab/alpaca
Description: 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 instruction better. The a... | Alpaca is a Hugging Face text dataset for instruction tuning. It contains about 52,000 instruction-response demonstrations generated with text-davinci-003 and adapted from the Self-Instruct pipeline. The dataset is in English, uses CC-BY-NC-4.0 licensing, and is suitable for text generation and instruction-finetuning t... | huggingface | tatsu-lab/alpaca |
Summarize the dataset based on its metadata and description. | Name: bad_prompt
Source: huggingface
ID: Nerfgun3/bad_prompt
Description: Negative Embedding / Textual Inversion. The idea behind this embedding was to somehow train the negative prompt as an embedding, thus unifying the basis of the negative prompt into one word or embedding. Embedding has proven to be very helpful fo... | bad_prompt is a Hugging Face image-related resource focused on negative embedding or textual inversion for Stable Diffusion workflows. It is designed to encode a negative prompt into a reusable embedding, potentially improving generation quality such as hand rendering. The dataset is small, in English, and associated w... | huggingface | Nerfgun3/bad_prompt |
Summarize the dataset and extract its key metadata for SFT training. | Name: falcon-refinedweb
Source: huggingface
Dataset 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 RefinedW... | Falcon RefinedWeb is a large English web text dataset from TII, derived from CommonCrawl using strict filtering and large-scale deduplication. It is intended for text generation and is described as multimodal-friendly due to inclusion of links and alt text. The dataset is distributed in parquet format, supports multipl... | huggingface | tiiuae/falcon-refinedweb |
Summarize the dataset and extract its key metadata for SFT training. | Name: lmsys-chat-1m
Source: huggingface
Dataset 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 on the Vicuna demo and Chatbot Arena website from Apr... | LMSYS-Chat-1M is a real-world conversation dataset containing one million chats involving 25 modern LLMs. The data was collected from Vicuna demo and Chatbot Arena traffic between April and August 2023 across 210K unique IP addresses. Each record includes conversation ID, model name, conversation content in OpenAI API ... | huggingface | lmsys/lmsys-chat-1m |
Summarize the dataset metadata into a concise description for supervised fine-tuning data discovery. | Source: huggingface
Dataset ID: anon8231489123/ShareGPT_Vicuna_unfiltered
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, bu... | ShareGPT_Vicuna_unfiltered is an English Hugging Face dataset intended for conversational model training and cleaning workflows. It contains a further-cleaned ShareGPT/Vicuna-style corpus, with noted variants that either remove or retain responses beginning with "I'm sorry, but". The dataset is licensed under Apache-2.... | huggingface | anon8231489123/ShareGPT_Vicuna_unfiltered |
Summarize the dataset metadata into a concise description for supervised fine-tuning data discovery. | Source: huggingface
Dataset 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_categories:text-... | finepdfs is a large-scale Hugging Face text dataset built exclusively from PDF sources. It is described as the largest publicly available PDF-derived corpus, with about 3 trillion tokens spanning 475 million documents in 1733 languages. The dataset supports text-generation use cases, is distributed in Parquet format, i... | huggingface | HuggingFaceFW/finepdfs |
Summarize the dataset based on the provided metadata. | Name: PhysicalAI-Autonomous-Vehicles
Source: huggingface
ID: nvidia/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 gen... | PhysicalAI-Autonomous-Vehicles is a Hugging Face dataset from NVIDIA focused on autonomous vehicles. It contains a large, geographically diverse collection of multi-sensor driving data designed to support research and development of Physical AI end-to-end driving systems. The data and labels are collected automatically... | huggingface | nvidia/PhysicalAI-Autonomous-Vehicles |
Summarize the dataset based on the provided metadata. | Name: OpenThoughts-114k
Source: huggingface
ID: open-thoughts/OpenThoughts-114k
Description: Open-Thoughts-114k. Open synthetic reasoning dataset with 114k high-quality examples covering math, science, code, and puzzles. Available subsets include a default subset containing ready-to-train data used to finetune the Open... | OpenThoughts-114k is a Hugging Face synthetic reasoning dataset with 114,000 high-quality text examples spanning math, science, coding, and puzzles. It is distributed in Parquet format and supports multiple data libraries including datasets, Dask, Polars, and mlcroissant. The dataset includes a default ready-to-train s... | huggingface | open-thoughts/OpenThoughts-114k |
Summarize the dataset for instruction-tuning use. | Name: OpenHermes-2.5
Source: huggingface
ID: teknium/OpenHermes-2.5
Description: This dataset made OpenHermes 2.5 and Nous Hermes 2 series of models. It is described as an exact compilation and curation of many open source datasets and custom synthetic datasets used to advance SOTA LLMs.
Tags: language:eng, size_catego... | OpenHermes-2.5 is a large English text dataset on Hugging Face designed for LLM instruction tuning. It is a compiled and curated mixture of open-source and synthetic data, including GPT-4-based distillation content, and was used to build the OpenHermes 2.5 and Nous Hermes 2 model series. The dataset is distributed in J... | huggingface | teknium/OpenHermes-2.5 |
Summarize the dataset for instruction-finetuning use. | Name: alpaca-cleaned
Source: huggingface
ID: yahma/alpaca-cleaned
Description: This is a cleaned version of the original Alpaca dataset released by Stanford. It fixes issues in the original release, including hallucination-prone instructions that referenced internet data.
Tags: task_categories:text-generation, language... | alpaca-cleaned is an English instruction-finetuning dataset on Hugging Face derived from Stanford's original Alpaca release. It is a cleaned text-generation dataset that removes or fixes problematic samples, especially instructions likely to cause hallucinations due to references to unavailable internet data. The datas... | huggingface | yahma/alpaca-cleaned |
Summarize the dataset based on the provided metadata. | Name: hle
Source: huggingface
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.
Tags: benchmark:of... | {"name":"hle","repository_id":"cais/hle","source":"huggingface","summary":"Humanity's Last Exam is an official multimodal benchmark dataset covering broad academic subjects with 2,500 closed-ended questions at the frontier of human knowledge.","modalities":["image","text"],"format":"parquet","license":"mit","size_categ... | huggingface | cais/hle |
Summarize the dataset based on the provided metadata. | Name: fineweb-2
Source: huggingface
ID: HuggingFaceFW/fineweb-2
Description: FineWeb2 is the second iteration of the FineWeb dataset, providing high-quality pretraining data for over 1000 languages. It is fully reproducible, available under the permissive ODC-By 1.0 license, and validated through hundreds of ablation e... | {"name":"fineweb-2","repository_id":"HuggingFaceFW/fineweb-2","source":"huggingface","summary":"FineWeb2 is a large-scale multilingual text dataset for pretraining and text generation, covering over 1000 languages. It is reproducible, extensively validated, and released under the ODC-By license.","task_categories":["te... | huggingface | HuggingFaceFW/fineweb-2 |
Summarize the dataset based on the provided metadata. | Name: imagenet-1k
Source: huggingface
Dataset ID: ILSVRC/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. There are more than ... | ImageNet-1k is a large-scale image classification dataset from the ILSVRC 2012 benchmark. It is organized using the WordNet hierarchy, where each class corresponds to a synset, and it is designed for multi-class image classification. The dataset is in English, uses crowdsourced annotations, contains between 1M and 10M ... | huggingface | ILSVRC/imagenet-1k |
Summarize the dataset based on the provided metadata. | Name: Alpaca-CoT
Source: huggingface
Dataset ID: QingyiSi/Alpaca-CoT
Description: This repository continuously collects various instruction tuning datasets and standardizes them into the same format so they can be directly loaded by Alpaca model code. It also supports empirical study on various instruction-tuning datas... | Alpaca-CoT is an instruction-finetuning dataset collection that aggregates and standardizes multiple instruction-tuning datasets into a unified format compatible with Alpaca-style models. It supports chain-of-thought and instruction-based training use cases, includes multilingual data such as English and Chinese, and i... | huggingface | QingyiSi/Alpaca-CoT |
Summarize the dataset and identify its main characteristics, including purpose, modalities, languages, tasks, licensing, and notable metadata. | 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 large langu... | PersonaHub is a Hugging Face dataset focused on large-scale synthetic data creation through persona-driven synthesis. It accompanies the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas" and introduces PERSONA HUB, a collection of 1 billion diverse personas automatically curated from web data to suppo... | huggingface | proj-persona/PersonaHub |
Summarize the dataset and explain why it could be useful for supervised fine-tuning or prompt engineering research. | Dataset name: prompts.chat
Source: huggingface
Dataset ID: fka/prompts.chat
Description: a.k.a. Awesome ChatGPT Prompts. This is a Dataset Repository mirror of prompts.chat — a social platform for AI prompts. prompts.chat is an open-source platform where users can share, discover, and collect AI prompts from the commun... | prompts.chat is a community prompt collection mirrored on Hugging Face. It contains prompt examples oriented around question answering, text generation, chatbot use, and prompt engineering across major LLM ecosystems such as ChatGPT, Claude, Gemini, Llama, and Mistral. Because it is licensed under CC0-1.0 and distribut... | huggingface | fka/prompts.chat |
Provide a concise dataset card style summary, including scale, content type, and whether it is appropriate for direct supervised fine-tuning. | Dataset name: fineweb
Source: huggingface
Dataset ID: HuggingFaceFW/fineweb
Description: FineWeb is 15 trillion tokens of cleaned and deduplicated English web data from CommonCrawl, later expanded to more than 18.5T tokens. The processing pipeline is optimized for LLM performance and ran on the datatrove library.
Tags:... | FineWeb is a massive English web corpus containing over 18.5 trillion cleaned and deduplicated tokens derived from CommonCrawl. It is designed primarily as large-scale pretraining data for language models, not as an instruction-response dataset. Its strengths are scale, broad web coverage, and preprocessing optimized f... | huggingface | HuggingFaceFW/fineweb |
Explain what kind of alignment data this dataset contains and how it should and should not be used in LLM training. | Dataset name: hh-rlhf
Source: huggingface
Dataset ID: Anthropic/hh-rlhf
Description: This repository provides human preference data about helpfulness and harmlessness from Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback. These data are meant to train preference or reward models... | HH-RLHF contains human preference data centered on helpfulness and harmlessness, making it a core alignment dataset for reward modeling and preference learning. Its main use case is training reward models or preference models that can be applied in RLHF pipelines. It should not be treated as a standard supervised chat ... | huggingface | Anthropic/hh-rlhf |
Summarize the dataset in terms of task coverage, training value, and likely SFT use cases. | Dataset name: OpenOrca
Source: huggingface
Dataset ID: Open-Orca/OpenOrca
Description: The OpenOrca dataset is a rich collection of augmented FLAN data aligned as closely as possible with the distributions described in the Orca paper. It has been used to generate strong model checkpoints and is positioned as a valuable... | OpenOrca is a large English instruction-style dataset built from augmented FLAN data and shaped to resemble the distributions discussed in the Orca work. It spans many task types, including question answering, summarization, classification, table QA, feature extraction, and text generation. This broad task coverage mak... | huggingface | Open-Orca/OpenOrca |
Describe the dataset’s structure, multilingual properties, and why it is relevant for chat model fine-tuning and alignment. | Dataset name: oasst1
Source: huggingface
Dataset ID: OpenAssistant/oasst1
Description: OpenAssistant Conversations Dataset (OASST1) is a human-generated, human-annotated assistant-style conversation corpus with 161,443 messages in 35 languages, annotated with 461,292 quality ratings and more than 10,000 fully annotated... | OASST1 is a multilingual assistant conversation dataset containing human-written dialogue messages organized into conversation trees, along with extensive human quality annotations. Its 35-language coverage makes it especially valuable for multilingual chat model development, while the tree structure and ratings suppor... | huggingface | OpenAssistant/oasst1 |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: gsm8k
Description: Dataset Card for GSM8K. GSM8K (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 mathematical problems that require multi-step reasoning. These problems take ... | GSM8K is an English text dataset for grade-school math word problem solving. It is widely used as a benchmark for question answering and text generation tasks that require multi-step numerical reasoning. The dataset contains roughly 8.5K examples, placing it in the 10K to 100K size category. Its main strength for LLM t... | huggingface | openai/gsm8k |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: wikipedia
Description: Dataset Card for Wikimedia Wikipedia. Wikipedia 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. Each example contains the content of one full Wikipedia article with cleani... | Wikimedia Wikipedia is a large-scale multilingual text corpus of cleaned Wikipedia articles. It is suitable for text generation, language modeling, and masked language modeling tasks, with one subset per language and article-level examples. The dataset is extremely large, in the 10M to 100M example range, and spans a v... | huggingface | wikimedia/wikipedia |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: EasyNegative
Description: Negative Embedding. This 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:other, size_categories:n<1K, format:imagefolder, modality:image
... | EasyNegative is an image-focused asset intended for Stable Diffusion workflows rather than a conventional LLM supervised fine-tuning dataset. It appears to be a negative embedding resource used in image generation pipelines, with very small scale and image modality metadata. Because it is not a text instruction dataset... | huggingface | gsdf/EasyNegative |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | 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
Downloads: 2084
Likes: 1151 | RedPajama-Data-1T is a large English text corpus designed as an open-source reproduction of the data philosophy behind LLaMA-style pretraining. It is intended for text generation and large-scale language model pretraining rather than narrow task supervision. The dataset is text-only and falls into a very large scale ca... | huggingface | togethercomputer/RedPajama-Data-1T |
Summarize the following dataset for LLM training relevance, including domain, task type, language, scale, and notable properties. | Name: medical-o1-reasoning-SFT
Description: News: [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. [2025/02/22] We released the distilled dataset from Deepseek-R1 based on medical verifiab... | medical-o1-reasoning-SFT is a bilingual English-Chinese medical reasoning dataset designed specifically for supervised fine-tuning. It supports question answering and text generation, with a focus on verifiable medical problems and distilled reasoning traces. The dataset is in JSON format, text-only, and medium-sized i... | huggingface | FreedomIntelligence/medical-o1-reasoning-SFT |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: HuggingFaceFW/fineweb-edu
Name: fineweb-edu
Description: 1.3 trillion tokens of educational web data filtered from FineWeb using an educational quality classifier.
Tags: task_categories:text-generation, language:en, license:odc-by, size_categories:1B<n<10B, format:parquet, modality:tabul... | fineweb-edu is a large English educational text dataset on Hugging Face containing 1.3 trillion tokens filtered from FineWeb for high educational quality. It is intended for text generation and language model pretraining research, distributed in parquet format with text and tabular modalities under the ODC-By license. ... | huggingface | HuggingFaceFW/fineweb-edu |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: allenai/dolma
Name: dolma
Description: Dolma: 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, arxiv:2301.13688, region:us, language-modeling, casual... | dolma is an English open corpus for language model pretraining research containing three trillion tokens. It is aimed at text generation and language modeling use cases, published under the ODC-By license, and associated with LLM and causal language modeling research. The dataset has 3,026 downloads and 1,021 likes on ... | huggingface | allenai/dolma |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: bigcode/the-stack
Name: the-stack
Description: The Stack is a near-deduplicated 3TB source code dataset with permissive licenses across many programming languages.
Tags: task_categories:text-generation, language_creators:crowdsourced, language_creators:expert-generated, multilinguality:m... | the-stack is a multilingual code dataset designed for text generation and code model training. It contains near-deduplicated source code spanning many programming languages, released in parquet format with text and tabular modalities. The dataset emphasizes permissive licensing coverage and has 11,947 downloads with 98... | huggingface | bigcode/the-stack |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: roneneldan/TinyStories
Name: TinyStories
Description: Dataset containing synthetically generated short stories by GPT-3.5 and GPT-4 using a small vocabulary.
Tags: task_categories:text-generation, language:en, license:cdla-sharing-1.0, size_categories:1M<n<10M, format:parquet, modality:t... | TinyStories is an English text generation dataset of synthetically generated short stories created by GPT-3.5 and GPT-4 using a restricted vocabulary. It is useful for language modeling experiments focused on simple narrative text and is distributed in parquet format under the CDLA-Sharing-1.0 license. The dataset has ... | huggingface | roneneldan/TinyStories |
Summarize the dataset metadata into a concise SFT training example. | Source: huggingface
Dataset ID: databricks/databricks-dolly-15k
Name: databricks-dolly-15k
Description: Open source instruction-following dataset generated by Databricks employees across tasks such as brainstorming, classification, QA, generation, information extraction, and summarization.
Tags: task_categories:questio... | databricks-dolly-15k is an English instruction-following dataset containing about 15,000 human-written records created by Databricks employees. It covers tasks such as question answering, summarization, classification, brainstorming, and information extraction, and is distributed in JSON format under the CC-BY-SA-3.0 l... | huggingface | databricks/databricks-dolly-15k |
Summarize the dataset based on its metadata and description. | Name: alpaca
Source: huggingface
ID: tatsu-lab/alpaca
Description: 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 instruction better. The a... | Alpaca is a Hugging Face text dataset for instruction tuning. It contains about 52,000 instruction-response demonstrations generated with text-davinci-003 and adapted from the Self-Instruct pipeline. The dataset is in English, uses CC-BY-NC-4.0 licensing, and is suitable for text generation and instruction-finetuning t... | huggingface | tatsu-lab/alpaca |
Summarize the dataset based on its metadata and description. | Name: bad_prompt
Source: huggingface
ID: Nerfgun3/bad_prompt
Description: Negative Embedding / Textual Inversion. The idea behind this embedding was to somehow train the negative prompt as an embedding, thus unifying the basis of the negative prompt into one word or embedding. Embedding has proven to be very helpful fo... | bad_prompt is a Hugging Face image-related resource focused on negative embedding or textual inversion for Stable Diffusion workflows. It is designed to encode a negative prompt into a reusable embedding, potentially improving generation quality such as hand rendering. The dataset is small, in English, and associated w... | huggingface | Nerfgun3/bad_prompt |
Summarize the dataset and extract its key metadata for SFT training. | Name: falcon-refinedweb
Source: huggingface
Dataset 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 RefinedW... | Falcon RefinedWeb is a large English web text dataset from TII, derived from CommonCrawl using strict filtering and large-scale deduplication. It is intended for text generation and is described as multimodal-friendly due to inclusion of links and alt text. The dataset is distributed in parquet format, supports multipl... | huggingface | tiiuae/falcon-refinedweb |
Summarize the dataset and extract its key metadata for SFT training. | Name: lmsys-chat-1m
Source: huggingface
Dataset 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 on the Vicuna demo and Chatbot Arena website from Apr... | LMSYS-Chat-1M is a real-world conversation dataset containing one million chats involving 25 modern LLMs. The data was collected from Vicuna demo and Chatbot Arena traffic between April and August 2023 across 210K unique IP addresses. Each record includes conversation ID, model name, conversation content in OpenAI API ... | huggingface | lmsys/lmsys-chat-1m |
Summarize the dataset metadata into a concise description for supervised fine-tuning data discovery. | Source: huggingface
Dataset ID: anon8231489123/ShareGPT_Vicuna_unfiltered
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, bu... | ShareGPT_Vicuna_unfiltered is an English Hugging Face dataset intended for conversational model training and cleaning workflows. It contains a further-cleaned ShareGPT/Vicuna-style corpus, with noted variants that either remove or retain responses beginning with "I'm sorry, but". The dataset is licensed under Apache-2.... | huggingface | anon8231489123/ShareGPT_Vicuna_unfiltered |
Summarize the dataset metadata into a concise description for supervised fine-tuning data discovery. | Source: huggingface
Dataset 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_categories:text-... | finepdfs is a large-scale Hugging Face text dataset built exclusively from PDF sources. It is described as the largest publicly available PDF-derived corpus, with about 3 trillion tokens spanning 475 million documents in 1733 languages. The dataset supports text-generation use cases, is distributed in Parquet format, i... | huggingface | HuggingFaceFW/finepdfs |
Summarize the dataset based on the provided metadata. | Name: PhysicalAI-Autonomous-Vehicles
Source: huggingface
ID: nvidia/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 gen... | PhysicalAI-Autonomous-Vehicles is a Hugging Face dataset from NVIDIA focused on autonomous vehicles. It contains a large, geographically diverse collection of multi-sensor driving data designed to support research and development of Physical AI end-to-end driving systems. The data and labels are collected automatically... | huggingface | nvidia/PhysicalAI-Autonomous-Vehicles |
Summarize the dataset based on the provided metadata. | Name: OpenThoughts-114k
Source: huggingface
ID: open-thoughts/OpenThoughts-114k
Description: Open-Thoughts-114k. Open synthetic reasoning dataset with 114k high-quality examples covering math, science, code, and puzzles. Available subsets include a default subset containing ready-to-train data used to finetune the Open... | OpenThoughts-114k is a Hugging Face synthetic reasoning dataset with 114,000 high-quality text examples spanning math, science, coding, and puzzles. It is distributed in Parquet format and supports multiple data libraries including datasets, Dask, Polars, and mlcroissant. The dataset includes a default ready-to-train s... | huggingface | open-thoughts/OpenThoughts-114k |
Summarize the dataset for instruction-tuning use. | Name: OpenHermes-2.5
Source: huggingface
ID: teknium/OpenHermes-2.5
Description: This dataset made OpenHermes 2.5 and Nous Hermes 2 series of models. It is described as an exact compilation and curation of many open source datasets and custom synthetic datasets used to advance SOTA LLMs.
Tags: language:eng, size_catego... | OpenHermes-2.5 is a large English text dataset on Hugging Face designed for LLM instruction tuning. It is a compiled and curated mixture of open-source and synthetic data, including GPT-4-based distillation content, and was used to build the OpenHermes 2.5 and Nous Hermes 2 model series. The dataset is distributed in J... | huggingface | teknium/OpenHermes-2.5 |
Summarize the dataset for instruction-finetuning use. | Name: alpaca-cleaned
Source: huggingface
ID: yahma/alpaca-cleaned
Description: This is a cleaned version of the original Alpaca dataset released by Stanford. It fixes issues in the original release, including hallucination-prone instructions that referenced internet data.
Tags: task_categories:text-generation, language... | alpaca-cleaned is an English instruction-finetuning dataset on Hugging Face derived from Stanford's original Alpaca release. It is a cleaned text-generation dataset that removes or fixes problematic samples, especially instructions likely to cause hallucinations due to references to unavailable internet data. The datas... | huggingface | yahma/alpaca-cleaned |
Summarize the dataset based on the provided metadata. | Name: hle
Source: huggingface
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.
Tags: benchmark:of... | {"name":"hle","repository_id":"cais/hle","source":"huggingface","summary":"Humanity's Last Exam is an official multimodal benchmark dataset covering broad academic subjects with 2,500 closed-ended questions at the frontier of human knowledge.","modalities":["image","text"],"format":"parquet","license":"mit","size_categ... | huggingface | cais/hle |
Summarize the dataset based on the provided metadata. | Name: fineweb-2
Source: huggingface
ID: HuggingFaceFW/fineweb-2
Description: FineWeb2 is the second iteration of the FineWeb dataset, providing high-quality pretraining data for over 1000 languages. It is fully reproducible, available under the permissive ODC-By 1.0 license, and validated through hundreds of ablation e... | {"name":"fineweb-2","repository_id":"HuggingFaceFW/fineweb-2","source":"huggingface","summary":"FineWeb2 is a large-scale multilingual text dataset for pretraining and text generation, covering over 1000 languages. It is reproducible, extensively validated, and released under the ODC-By license.","task_categories":["te... | huggingface | HuggingFaceFW/fineweb-2 |
Summarize the dataset based on the provided metadata. | Name: imagenet-1k
Source: huggingface
Dataset ID: ILSVRC/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. There are more than ... | ImageNet-1k is a large-scale image classification dataset from the ILSVRC 2012 benchmark. It is organized using the WordNet hierarchy, where each class corresponds to a synset, and it is designed for multi-class image classification. The dataset is in English, uses crowdsourced annotations, contains between 1M and 10M ... | huggingface | ILSVRC/imagenet-1k |
Summarize the dataset based on the provided metadata. | Name: Alpaca-CoT
Source: huggingface
Dataset ID: QingyiSi/Alpaca-CoT
Description: This repository continuously collects various instruction tuning datasets and standardizes them into the same format so they can be directly loaded by Alpaca model code. It also supports empirical study on various instruction-tuning datas... | Alpaca-CoT is an instruction-finetuning dataset collection that aggregates and standardizes multiple instruction-tuning datasets into a unified format compatible with Alpaca-style models. It supports chain-of-thought and instruction-based training use cases, includes multilingual data such as English and Chinese, and i... | huggingface | QingyiSi/Alpaca-CoT |
Summarize the dataset and identify its main characteristics, including purpose, modalities, languages, tasks, licensing, and notable metadata. | 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 large langu... | PersonaHub is a Hugging Face dataset focused on large-scale synthetic data creation through persona-driven synthesis. It accompanies the paper "Scaling Synthetic Data Creation with 1,000,000,000 Personas" and introduces PERSONA HUB, a collection of 1 billion diverse personas automatically curated from web data to suppo... | huggingface | proj-persona/PersonaHub |
SoloAI SFT Dataset: 20260425-0537
📊 数据集概览
| 属性 | 值 |
|---|---|
| 数据条数 | 30 条 |
| 数据源 | huggingface |
| 生成时间 | 2026-04-25 05:37 |
| 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': 'Summarize the dataset and explain why it could be useful for supervised fine-tuning or prompt engineering research.', 'input': 'Dataset name: prompts.chat\nSource: huggingface\nDataset ID: fka/prompts.chat\nDescription: a.k.a. Awesome ChatGPT Prompts. This is a Dataset Repository mirror of prompts.chat — a social platform for AI prompts. prompts.chat is an open-source platform where users can share, discover, and collect AI prompts from the community.\nTags: task_categories:question-answering, task_categories:text-generation, license:cc0-1.0, size_categories:1K<n<10K, format:csv, modality:text, library:datasets, library:pandas, library:polars, library:mlcroissant, region:us, ChatGPT, prompts, AI, GPT, Claude, Gemini, Llama, Mistral, LLM, prompt-engineering, conversational-ai, text-generation, chatbot, awesome-list\nDownloads: 44569\nLikes: 9680', 'output': 'prompts.chat is a community prompt collection mirrored on Hugging Face. It contains prompt examples oriented around question answering, text generation, chatbot use, and prompt engineering across major LLM ecosystems such as ChatGPT, Claude, Gemini, Llama, and Mistral. Because it is licensed under CC0-1.0 and distributed in CSV format, it is easy to reuse for building prompt libraries, instruction templates, synthetic data generation workflows, or studies of prompt patterns and prompt quality. For SFT, it is most useful as a source of prompt formulations and task framing rather than as a direct conversational response dataset.', '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 | 2026-04-25 05:37 | 初始发布，30 条数据 |
- Downloads last month
- 9