Datasets:

---

fka

/

prompts.chat

like 9.72k

Identifies the most frequent action types and their associated prompts, revealing patterns in how different actions are described and potentially uncovering semantic relationships between actions and their textual representations.

Identifies the most frequent game-related prompts in training data, revealing popular gaming themes and topics that could inform content generation strategies and identify gaps in training examples.

Identifies the most frequent conversational actions and reveals patterns in how different actions relate to prompt diversity and length, helping understand the variety and complexity of interactions in the dataset.

Provides meaningful analysis of frontend development prompts and actions, revealing patterns in developer requests and prompt lengths for popular frontend technologies like React, Angular, and Vue.

Identifies the most frequently occurring words in training prompts, revealing common terminology and potential biases in the dataset that could inform model training and bias mitigation strategies.

Identifies the most frequent words in training prompts, revealing common vocabulary patterns that could inform language model training strategies and text preprocessing approaches.

Identifies and displays the longest stock investor-related prompts and actions, providing insights into the types of queries and responses related to stock investment in the dataset.

Provides detailed statistics and a distribution chart for prompt lengths across different acts, revealing patterns in data density and prompt variability.

Displays the top 10 acts by count, along with the average prompt length and a visual representation of the count, revealing patterns in prompt length across different acts.

Provides a summary of total records, unique actions, and average prompt length with a visual bar chart, offering insight into data patterns.

Provides a visual representation of the frequency of actions along with the corresponding prompt lengths, highlighting the most common patterns which can be useful for training an LLM.

Filters and analyzes game-related prompts and actions to show text length distributions, helping identify patterns in game content creation.

Identifies the longest prompts in the training dataset and visualizes their lengths using a bar chart, which helps understand the distribution of prompt sizes and potential impact on model processing.

Filters and retrieves developer-focused prompts related to orchestration and agents, creating a useful subset for building specialized training materials.

Filters prompts containing research-related keywords to identify exploratory and investigative queries, providing useful insights into the types of analytical tasks included in the training data.

Shows the distribution of different types in the training dataset, revealing which categories are most and least represented.

Retrieves samples from the dataset related to crypto and trading, excluding those marked for developers, providing insights into user queries and assistant responses on these topics.

Displays the 20 longest distinct prompts along with their lengths and a visual bar representation of these lengths.

Displays the most common acts along with their associated prompts, ordered by the frequency of acts and the length of prompts.

Identifies infrequently occurring action codes in the 'train' dataset, which could help in understanding rare cases or outliers in the data.

Displays the top 10 longest prompts along with their lengths and a visual bar chart representation of these lengths.

The query provides insights into the distribution of prompt lengths across different acts, helping to understand variability and frequency in the dataset.

Retrieves prompts that do not contain common food-related keywords, potentially highlighting a diverse set of non-culinary topics.

Retrieves records filtered by a boolean flag, providing basic dataset segmentation but offering limited analytical insight beyond simple data retrieval.

Retrieves examples where the 'act' field mentions image-related terms, but only provides basic filtering without deeper analysis of patterns or relationships in the data.