Dark Mode
ManaGPT Futures Dataset
Education & Learning Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset contains 4,080 single-sentence texts produced by the ManaGPT-1020 large language model. These texts represent responses to 102 distinct input prompts, with each prompt being used 20 times to generate a response. ManaGPT-1020 is a free, open-source 1.5-billion-parameter large language model, fine-tuned on a specialised English-language corpus of over 509,000 words from the domain of organisational futures studies. The model is designed to generate analysis, predictions, and recommendations regarding the emerging roles of advanced AI, social robotics, ubiquitous computing, virtual reality, neurocybernetic augmentation, and other "posthumanising" technologies within organisational life. The input prompts were created by concatenating 12 different subject phrases with 17 different modal variants in every possible combination.
Columns
- subject_of_prompt: One of 12 distinct noun phrases used as the initial words in an input sequence or prompt.
- modal_variant_of_prompt: One of 17 phrases expressing different types of linguistic modality, used as the ending portion of an input sequence or prompt.
- complete_prompt: The entire prompt supplied to the model, formed by concatenating the subject and modal variant.
- generated_response: The entire response text that was generated by the model.
- generated_response_excluding_prompt: The new portion of the response generated by the model, after the input sequence has been removed from the beginning of the sentence.
Distribution
The dataset consists of 4,080 unique texts, typically provided as a data file in CSV format. Each generated text corresponds to a single record or row. The dataset structure includes distinct columns for prompt components and generated responses.
Usage
This dataset is ideal for a range of applications in natural language processing (NLP), machine learning (ML), and artificial intelligence (AI). It can be used for:
- Fine-tuning and evaluating other large language models.
- Analysing and understanding text generation patterns from advanced AI models.
- Developing applications focused on future organisational structures and technological impacts.
- Educational purposes in courses related to AI, NLP, and machine learning, offering practical examples of text synthesis.
Coverage
The dataset has a global regional coverage. Its scope is primarily focused on the domain of organisational futures studies, specifically addressing the role of advanced AI, social robotics, ubiquitous computing, virtual reality, neurocybernetic augmentation, and other "posthumanising" technologies in organisational life. The content is in English. In a small percentage of cases (approximately 2%), the model did not generate output beyond the input sequence, especially when an empty string was used as part of the prompt.
License
CC-BY-SA
Who Can Use It
- AI and ML Developers: For building and refining text generation models.
- NLP Researchers: For conducting studies on language models and their output characteristics.
- Data Scientists: For exploring and extracting insights from synthetic text data.
- Educators and Students: As a practical resource for learning and experimentation in AI and related fields.
- Organisational Theorists and Futurists: To gain perspectives on AI-generated analyses of future workplace scenarios.
Dataset Name Suggestions
- ManaGPT-1020 Text Generation Data
- AI Generated Organisational Texts
- LLM Prompt Response Archive
- ManaGPT Futures Dataset
Attributes
Original Data Source: ManaGPT: 4,080 NLP prompts and generated texts
QUALITY