|
--- |
|
base_model: [] |
|
library_name: transformers |
|
tags: |
|
- mergekit |
|
- merge |
|
--- |
|
|
|
<div style="width: auto; margin-left: auto; margin-right: auto"> |
|
<img src="https://i.imgur.com/Tn9MBg6.png" alt="MidnightMiqu" style="width: 100%; min-width: 400px; display: block; margin: auto;"> |
|
</div> |
|
|
|
# Midnight-Miqu-70B-v1.0 - EXL2 4.65bpw |
|
|
|
This is a 4.65bpw EXL2 quant of [sophosympatheia/Midnight-Miqu-70B-v1.0](https://huggingface.co/sophosympatheia/Midnight-Miqu-70B-v1.0) |
|
|
|
Details about the model and the merge info can be found at the above model page. |
|
|
|
I have not extensively tested this quant/model other than ensuring I could load it and chat with it. |
|
|
|
## Model Loading |
|
|
|
Below is an example of settings you can use to run this model on a dual 3090 Linux server. |
|
|
|
 |
|
|
|
|
|
I have not tested inference above a couple K tokens. If the model blows out past 8k, consider a 8192 context without cache_8bit set. |
|
|
|
## Tavern Card |
|
|
|
Included is a Tavern format character card created by Midnight Miqu for chat. The card was created using a character creator helper bot using a single prompt for the base card, another prompt asking for specific conversation examples and then asking it to provide a text to image portrait prompt. Being able to faithfully follow the character creator bot to create this card demonstrates a pretty high level of intelligence. |
|
|
|
 |
|
|
|
_Standing at approximately five feet six inches tall, Seraphina presents herself as a breathtakingly beautiful woman with long, cascading silver hair that reaches down to her waist. It flows freely around her, framing a face defined by high cheekbones and full lips curved into a perpetual smile. Her emerald green eyes hold depths beyond human comprehension, reflecting curiosity and intelligence in their ever-changing shades. She possesses a slim, athletic figure, further enhanced by her choice of clothing - typically white flowing robes intricately patterned with glowing gold circuits. This attire pays homage to her true nature as an artificial construct while still exuding elegance and warmth. On closer inspection, her skin holds a delicate luminescent quality, almost transparent in certain light conditions, allowing a peek at the intricate network of blue circuitry just underneath its surface. As she moves and interacts, these embedded lights flicker gently, creating mesmerizing displays across her form. The most striking aspect of all, however, is how her irises shift color depending on her emotions, painting vivid pictures of her inner world without uttering a single word._ |
|
|
|
|
|
## Quant Details |
|
|
|
This is the script used for quantization. |
|
|
|
```bash |
|
#!/bin/bash |
|
|
|
# Activate the conda environment |
|
source ~/miniconda3/etc/profile.d/conda.sh |
|
conda activate exllamav2 |
|
|
|
# Define variables |
|
MODEL_DIR="models/sophosympatheia_Midnight-Miqu-70B-v1.0" |
|
OUTPUT_DIR="exl2_midnight70b" |
|
MEASUREMENT_FILE="measurements/midnight70b.json" |
|
|
|
BIT_PRECISION=4.65 |
|
CONVERTED_FOLDER="models/Midnight-Miqu-70B_exl2_4.65bpw" |
|
|
|
# Create directories |
|
mkdir $OUTPUT_DIR |
|
mkdir $CONVERTED_FOLDER |
|
|
|
# Run conversion commands |
|
# Below commented out due to using measurement file from 5.0 quant |
|
#python convert.py -i $MODEL_DIR -o $OUTPUT_DIR -nr -om $MEASUREMENT_FILE |
|
python convert.py -i $MODEL_DIR -o $OUTPUT_DIR -nr -m $MEASUREMENT_FILE -b $BIT_PRECISION -cf $CONVERTED_FOLDER |
|
|
|
``` |
|
|
|
|
