A collection of parsers for LLM benchmark datasets
Support by Parquet, CSV, Jsonl, XLS
Generate dataset for machine learning
Browse TheBloke models' history
Explore datasets on a Nomic Atlas map
Explore and manage datasets for machine learning
Build datasets using natural language
Save user inputs to datasets on Hugging Face
Label data for machine learning models
Speech Corpus Creation Tool
Create datasets with FAQs and SFT prompts
Display instructional dataset
Explore, annotate, and manage datasets
The LLMEval Dataset Parser is a collection of parsers designed to work with large language model (LLM) benchmark datasets. It provides a standardized way to browse and parse LLM benchmark datasets, making it easier to work with diverse dataset formats and structures.
• Multiple Dataset Support: Handles various benchmark datasets for LLM evaluation.
• Metadata Extraction: Extracts detailed metadata from datasets, including task descriptions and metrics.
• Data Validation: Ensures data integrity by validating dataset structures and formats.
• Versioning Support: Manages different versions of datasets for reproducibility.
• Cross-Platform Compatibility: Works seamlessly across different operating systems and environments.
• User-Friendly Interface: Provides a simple and intuitive CLI for parsing and managing datasets.
llm-eval parse --dataset [dataset_name] --path [dataset_path]```
from llm_eval import LLM Evaluations
dataset = LLM Evaluations.parse(dataset_name)```
What is the purpose of the LLMEval Dataset Parser?
The LLMEval Dataset Parser simplifies the process of working with LLM benchmark datasets by providing a standardized interface for parsing, validating, and managing datasets.
How do I install the LLMEval Dataset Parser?
You can install it using pip:
pip install llm-eval-parser```
**Can I use the parser with custom or unsupported datasets?**
Yes, the parser supports custom datasets. Contact the developers for guidance on integrating unsupported formats.