Captain_BMO-12B
| Property | Value |
|---|---|
| Parameter Count | 12.2B |
| Model Type | Mistral-based Language Model |
| Tensor Type | BF16 |
| License | Other |
| Language | English |
What is Captain_BMO-12B?
Captain_BMO-12B is a sophisticated language model built on the Nemo 12B instruct base, specifically engineered for enhanced text completion capabilities. This model represents a specialized fine-tuning experiment that combines robust instruction-following abilities with optimized performance.
Implementation Details
The model underwent a carefully crafted training process utilizing a 200,000 sample randomized subset of GU_instruct-Remastered-1.1, enhanced with 25,000 samples from the hathor/poppy dataset. The training was conducted over three epochs, resulting in a balanced and refined model output.
- Built on Nemo 12B instruct foundation
- Utilizes Mistral formatting for text completion
- Available in multiple quantization formats including GGUF and Exl2
- Supports both 4bpw and 6bpw Exl2 variants
Core Capabilities
- Advanced text completion with Mistral formatting
- Instruction-following capabilities
- Optimized for English language tasks
- Flexible deployment options through various quantization formats
Frequently Asked Questions
Q: What makes this model unique?
The model's unique blend of GU_instruct-Remastered-1.1 and hathor/poppy datasets, combined with its three-epoch training approach, creates a distinctive balance between instruction-following and general text completion capabilities.
Q: What are the recommended use cases?
This model is primarily designed for text completion tasks and internal testing purposes. While it demonstrates good performance, it's important to note that extended support is limited, and results may vary depending on specific use cases and context size requirements.
