Dolphin 2.9.3 Qwen2 1.5B Overview

Dolphin 2.9.3 Qwen2 1.5B is a 1.5 billion parameter language model built upon the Qwen2-1.5b base architecture. Developed by Eric Hartford, Lucas Atkins, Fernando Fernandes, and Cognitive Computations, this model is notable for its extensive 131,072-token context window, with fine-tuning conducted at a 16,000-token sequence length. The development process specifically excluded coding, function calling, and systemchat-multilingual datasets to optimize performance for its intended conversational and instructional capabilities.

Key Characteristics

• Base Model: Qwen2-1.5b with a 128k context length.
• Fine-tuning: Full-weight fine-tuning performed with a 16k sequence length.
• Dataset Filtering: Datasets were filtered to remove alignment and bias, resulting in an "uncensored" model that is highly compliant with user requests, including potentially unethical ones. Users are advised to implement their own alignment layers.
• Licensing: Governed by the Apache-2.0 license, permitting commercial use.
• Training Data: Trained on data generated from various sources, including GPT-4.

Use Cases

This model is well-suited for a variety of instruction-following and conversational AI applications where a highly compliant and unfiltered response is desired. Developers should be aware of its uncensored nature and implement appropriate safeguards for public-facing services.