| Property | Value |
|---|---|
| Parameter Count | 46.7B |
| Model Type | Mixtral |
| License | Apache 2.0 |
| Context Length | 32K tokens |
What is dolphin-2.5-mixtral-8x7b-GGUF?
Dolphin 2.5 Mixtral is an advanced language model based on the Mixtral-8x7B architecture, specifically optimized for coding tasks and general-purpose applications. It represents a significant evolution in the Dolphin series, incorporating multiple expert models and extensive training on diverse datasets including coding, instruction-following, and general knowledge.
Implementation Details
The model is available in various GGUF quantization formats, from 2-bit to 8-bit precision, allowing users to balance between model size and performance. The model utilizes a mixture of experts architecture with 8 expert models, trained using qLoRA and Axolotl over 1.5 epochs on 4 A100 GPUs.
- Multiple quantization options (Q2_K to Q8_0) for different hardware capabilities
- ChatML prompt format for consistent interaction
- Extensive training on 8 specialized datasets
- 32K context window support
Core Capabilities
- Advanced coding assistance and generation
- Highly compliant instruction following
- Extended context understanding
- Efficient resource utilization through GGUF format
- Balanced performance across general and specialized tasks
Frequently Asked Questions
Q: What makes this model unique?
The model combines Mixtral's powerful architecture with specialized training on coding and instruction-following datasets, making it particularly effective for development tasks while maintaining strong general-purpose capabilities. The GGUF format enables efficient deployment across different hardware configurations.
Q: What are the recommended use cases?
This model excels in software development tasks, technical writing, and general assistance scenarios. It's particularly well-suited for users who need both coding expertise and general knowledge handling, with flexible deployment options through various quantization levels.