Update README.md
Browse files
README.md
CHANGED
|
@@ -41,6 +41,8 @@ tags:
|
|
| 41 |
4. Run the command in your terminal
|
| 42 |
5. Now you have Andy-3.6 installed!
|
| 43 |
|
|
|
|
|
|
|
| 44 |
# How was model trained?
|
| 45 |
|
| 46 |
The model was trained on the [MindCraft dataset](https://huggingface.co/datasets/Sweaterdog/Andy-3.5-MASSIVE) for Andy-3.6, a curated dataset for Q & A, reasoning, and playing, which includes ~22,000 prompts.
|
|
@@ -52,8 +54,7 @@ Andy-3.6 also knows how to build / use !newAction to perform commands, it was tr
|
|
| 52 |
|
| 53 |
# What models can I choose?
|
| 54 |
|
| 55 |
-
There are going to be
|
| 56 |
-
* Large will be a 32B parameter model, tuned from [Deepseek-R1 Distilled](https://huggingface.co/unsloth/DeepSeek-R1-Distill-Qwen-32B-bnb-4bit)
|
| 57 |
* Regular is a 7B parameter model, tuned from [Deepseek-R1 Distilled](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B)
|
| 58 |
* Small is a 3B parameter model, tuned from [Qwen2.5 3B](Qwen/Qwen2.5-3B-Instruct)
|
| 59 |
|
|
@@ -61,8 +62,6 @@ All models will have **case-by-case reasoning** baked **into** the model, meanin
|
|
| 61 |
|
| 62 |
You can also *prompt* Andy-3.6 to reason for better performance
|
| 63 |
|
| 64 |
-
There will be a Andy-3.6-code, a model designed strictly for coding tasks, which can be found [here](https://huggingface.co/Sweaterdog/Andy-3.6-code)
|
| 65 |
-
|
| 66 |
# Safety and FAQ
|
| 67 |
|
| 68 |
Q: Is this model safe to use?
|
|
@@ -83,8 +82,8 @@ A. No, if you are making a post about MindCraft, and using this model, you only
|
|
| 83 |
|
| 84 |
# 🔥UPDATE🔥
|
| 85 |
|
| 86 |
-
## **Andy-3.6 Release!**
|
| 87 |
-
Andy-3.6 is
|
| 88 |
|
| 89 |
> # I want to thank all supporters! [!NOTE]
|
| 90 |
> I would love to thank everyone who supported this project, there is a list of supporters in the files section.
|
|
|
|
| 41 |
4. Run the command in your terminal
|
| 42 |
5. Now you have Andy-3.6 installed!
|
| 43 |
|
| 44 |
+
If you would prefer to use Andy-3.6-small, you can find that model [here](https://huggingface.co/Sweaterdog/Andy-3.6-small)
|
| 45 |
+
|
| 46 |
# How was model trained?
|
| 47 |
|
| 48 |
The model was trained on the [MindCraft dataset](https://huggingface.co/datasets/Sweaterdog/Andy-3.5-MASSIVE) for Andy-3.6, a curated dataset for Q & A, reasoning, and playing, which includes ~22,000 prompts.
|
|
|
|
| 54 |
|
| 55 |
# What models can I choose?
|
| 56 |
|
| 57 |
+
There are going to be 2 model sizes avaliable, Regular, and Small
|
|
|
|
| 58 |
* Regular is a 7B parameter model, tuned from [Deepseek-R1 Distilled](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B)
|
| 59 |
* Small is a 3B parameter model, tuned from [Qwen2.5 3B](Qwen/Qwen2.5-3B-Instruct)
|
| 60 |
|
|
|
|
| 62 |
|
| 63 |
You can also *prompt* Andy-3.6 to reason for better performance
|
| 64 |
|
|
|
|
|
|
|
| 65 |
# Safety and FAQ
|
| 66 |
|
| 67 |
Q: Is this model safe to use?
|
|
|
|
| 82 |
|
| 83 |
# 🔥UPDATE🔥
|
| 84 |
|
| 85 |
+
## **Andy-3.6-small Release!**
|
| 86 |
+
Andy-3.6-small is the latest model, as well as the last model in the Andy-3.6 generation. That model is capable of more reasoning than Andy-3.6 is.
|
| 87 |
|
| 88 |
> # I want to thank all supporters! [!NOTE]
|
| 89 |
> I would love to thank everyone who supported this project, there is a list of supporters in the files section.
|