KnutJaegersberg/Yi-34B-200K-MiniOrca
KnutJaegersberg/Yi-34B-200K-MiniOrca is a 34 billion parameter language model based on the Yi-34B-200K architecture, fine-tuned by KnutJaegersberg. It was trained for 2.7 epochs on the 50,000 shortest records of the TinyPixel/orca-mini dataset using NEFTune. This model is specifically optimized for generating detailed and long answers, making it suitable for tasks requiring comprehensive responses.
Loading preview...
Model Overview
KnutJaegersberg/Yi-34B-200K-MiniOrca is a 34 billion parameter language model built upon the official Yi-34B-200K base model. It has been fine-tuned by KnutJaegersberg through 2.7 epochs of training on a subset of the TinyPixel/orca-mini dataset, specifically utilizing the 50,000 shortest records. The training incorporated NEFTune, a technique often used to enhance model performance.
Key Capabilities
- Detailed Response Generation: The model is specifically trained to produce comprehensive and extended answers, as indicated by its prompt example demonstrating a request for a "detailed and long answer."
- Base Model Heritage: Leverages the robust architecture and capabilities of the Yi-34B-200K model, which is known for its substantial context length.
Good For
- Instruction Following: Excels at tasks where the user explicitly requests detailed and lengthy explanations or responses.
- Comprehensive Information Retrieval: Suitable for applications requiring in-depth answers to complex queries.
Licensing
The source code for this repository is under the Apache 2.0 license. The Yi series models, including this fine-tune, are open for academic research and free commercial usage, provided permission is obtained via application and adherence to the Model License Agreement 2.0. Commercial use requires contacting [email protected] for an official license.