Ezekielmsc/onyx-1.1-14b-hf
Onyx 1.1 is a 14 billion parameter language model developed by Ezekielmsc, fine-tuned from Qwen3-14B. This model is specifically optimized for NARA and is provided in safetensors (fp16) format. It is designed to be served efficiently using inference engines like vLLM or TGI, making it suitable for applications requiring a robust and specialized language model.
Loading preview...
Onyx 1.1 - 14B Overview
Onyx 1.1 is a 14 billion parameter language model, developed by Ezekielmsc, that has been fine-tuned from the Qwen3-14B architecture. This model is specifically tailored for applications related to NARA, indicating a specialized training focus.
Key Characteristics
- Base Model: Fine-tuned from Qwen3-14B.
- Parameter Count: 14 billion parameters.
- Format: Provided in
safetensorsformat, utilizingfp16(half-precision floating-point) for efficient storage and inference. - Inference Compatibility: Designed for deployment with high-performance inference engines such as vLLM or Text Generation Inference (TGI), ensuring optimized serving capabilities.
Use Cases
This model is particularly well-suited for scenarios requiring a specialized language model with a focus on NARA-related tasks. Its fp16 format and compatibility with vLLM/TGI make it a strong candidate for efficient, high-throughput inference in production environments where the Qwen3-14B architecture's strengths are beneficial and NARA-specific fine-tuning is advantageous.