|
--- |
|
base_model: unsloth/llama-3-8b-Instruct-bnb-4bit |
|
language: |
|
- en |
|
license: apache-2.0 |
|
tags: |
|
- text-generation-inference |
|
- transformers |
|
- unsloth |
|
- llama |
|
- trl |
|
- sft |
|
datasets: |
|
- OnlyCheeini/greesychat |
|
--- |
|
|
|
 |
|
|
|
# GreesyChat-Turbo AI Model |
|
|
|
## Overview |
|
|
|
GreesyChat-Turbo is an advanced AI model designed for robust text generation using the LLaMA 3 architecture. This model excels in providing high-quality responses for general conversation, mathematical queries, and more. It’s perfect for powering chatbots, virtual assistants, and any application requiring intelligent dialogue capabilities. |
|
|
|
## Benchmark Results |
|
|
|
| Metric | Value | |
|
|--------------------|------------| |
|
| **Perplexity** | 22.5 | |
|
| **Generation Speed** | 75 ms per token | |
|
| **Accuracy** | 70% | |
|
| **Response Time** | 200 ms | |
|
|
|
| Metric | GreesyChat-Turbo | Mixtral-8x7b | GPT-4 | |
|
|---------------|------------------|---------------|-------------| |
|
| **Code** | 79.2 | 75.6 | 83.6 | |
|
| **MMLU** | 74.5 | 79.9 | 85.1 | |
|
| **Gms8k** | 89.2 (5) | 88.7 | 94.2 | |
|
|
|
## Contact |
|
|
|
For support or inquiries, please contact: [mail@nicatdcw.dev](mailto:mail@nicatdcw.dev) |
|
|
