Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
tags:
|
| 6 |
+
- about
|
| 7 |
+
- NotLLM
|
| 8 |
+
- fake
|
| 9 |
+
---
|
| 10 |
+
# About Me - Akicou
|
| 11 |
+
|
| 12 |
+
Hi, I'm **Akicou**, and I specialize in sharing **quantized Large Language Models (LLMs)** in **GGUF format**. These models are processed using my own service, **[GGUFORGE](https://gguforge.com)**, which focuses on converting raw models into GGUF format for easier and more efficient usage.
|
| 13 |
+
|
| 14 |
+
## What I Do
|
| 15 |
+
|
| 16 |
+
- I post models and their **GGUF-converted versions** via [gguforge.com](https://gguforge.com), a platform I created for quantizing LLMs.
|
| 17 |
+
- I **test and verify** model outputs only when I explicitly state that a model **works**. If I haven't personally verified it, I can't guarantee the model's performance — especially when using raw safetensors.
|
| 18 |
+
- To keep my repositories clean and efficient, I **delete models within 5 days** of upload unless they prove to be valuable and worth keeping.
|
| 19 |
+
- My ongoing mission is to **reduce the size of large SOTA (State-of-the-Art) models** while maintaining their functionality and ensuring they **don’t produce garbage output**. As this is a complex and evolving process, I'm constantly experimenting with new methods and techniques.
|
| 20 |
+
|
| 21 |
+
## About GGUFORGE
|
| 22 |
+
|
| 23 |
+
**GGUFORGE** is a self-hosted platform where users can log in via **Hugging Face OAuth** and request conversions of models into GGUF format.
|
| 24 |
+
|
| 25 |
+
### Key Info:
|
| 26 |
+
- **No Guarantees on Speed**: Since the service is self-hosted, processing speed may vary. I’ll do my best to fulfill requests as quickly as possible.
|
| 27 |
+
- **Model Quality**: If a converted GGUF model produces poor results, it's likely due to the **original model's training or fine-tuning**, not the GGUF conversion process itself.
|
| 28 |
+
- **Request Policy**: I may deny conversion requests if there are already **many GGUF versions** of the same model available.
|
| 29 |
+
- **Self-Host Option**: If you prefer to run the conversion yourself, feel free to use the open-source tool I created: [automaticConversion GitHub repo](https://github.com/Akicuo/automaticConversion)
|
| 30 |
+
|
| 31 |
+
---
|
| 32 |
+
|
| 33 |
+
Thanks for stopping by, and happy quantizing!
|
| 34 |
+
— **Akicou**
|