Toppy-M-7B-exl2 / README.md
LogicismTV's picture
Upload folder using huggingface_hub
b2d345c verified
---
base_model: Undi95/Toppy-M-7B
inference: false
license: cc-by-nc-4.0
model_creator: Undi
model_name: Toppy M 7B
model_type: mistral
prompt_template: 'Below is an instruction that describes a task. Write a response
that appropriately completes the request.
### Instruction:
{prompt}
### Response:
'
quantized_by: LogicismTV
---
<div style="width: auto; margin-left: auto; margin-right: auto">
<img src="https://i.imgur.com/T1kcNir.jpg" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://logicism.tv/">Vist my Website</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/nStuNeZsWz">Join my Discord</a></p>
</div>
</div>
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
# Toppy M 7B - ExLlama V2
Original model: [Toppy M 7B](https://huggingface.co/Undi95/Toppy-M-7B)
# Description
This is an EXL2 quantization of the Undi95's Toppy M 7B model.
## Prompt template: Alpaca
```
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
```
# Quantizations
| Bits Per Weight | Size |
| --------------- | ---- |
| [main (2.4bpw)](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/main) | 2.29 GB |
| [3bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/3bpw) | 2.78 GB |
| [3.5bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/3.5bpw) | 3.19 GB |
| [4bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/4bpw) | 3.59 GB |
| [4.5bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/4.5bpw) | 4.00 GB |
| [5bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/5bpw) | 4.41 GB |
| [6bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/6bpw) | 5.22 GB |
| [8bpw](https://huggingface.co/LogicismTV/Toppy-M-7B-exl2/tree/8bpw) | 6.84 GB |
# Original model card: Carsten Kragelund's Chronomaid Storytelling 13B
<!-- description start -->
## Description
This repo contains fp16 files of Toppy-M-7B, a merge I have done with the new task_arithmetic merge method from mergekit.
This project was a request from [BlueNipples](https://huggingface.co/BlueNipples) : [link](https://huggingface.co/Undi95/Utopia-13B/discussions/1)
<!-- description end -->
<!-- description start -->
## Models and loras used
- [openchat/openchat_3.5](https://huggingface.co/openchat/openchat_3.5)
- [NousResearch/Nous-Capybara-7B-V1.9](https://huggingface.co/NousResearch/Nous-Capybara-7B-V1.9)
- [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta)
- [lemonilia/AshhLimaRP-Mistral-7B](lemonilia/AshhLimaRP-Mistral-7B)
- [Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b](https://huggingface.co/Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b)
- [Undi95/Mistral-pippa-sharegpt-7b-qlora](Undi95/Mistral-pippa-sharegpt-7b-qlora)
<!-- description end -->
## The sauce
```
openchat/openchat_3.5
lemonilia/AshhLimaRP-Mistral-7B (LoRA) x 0.38
NousResearch/Nous-Capybara-7B-V1.9
Vulkane/120-Days-of-Sodom-LoRA-Mistral-7b x 0.27
HuggingFaceH4/zephyr-7b-beta
Undi95/Mistral-pippa-sharegpt-7b-qlora x 0.38
merge_method: task_arithmetic
base_model: mistralai/Mistral-7B-v0.1
models:
- model: mistralai/Mistral-7B-v0.1
- model: Undi95/zephyr-7b-beta-pippa-sharegpt
parameters:
weight: 0.42
- model: Undi95/Nous-Capybara-7B-V1.9-120-Days
parameters:
weight: 0.29
- model: Undi95/openchat_3.5-LimaRP-13B
parameters:
weight: 0.48
dtype: bfloat16
```
<!-- prompt-template start -->
## Prompt template: Alpaca
```
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
```
If you want to support me, you can [here](https://ko-fi.com/undiai).