|
--- |
|
license: apache-2.0 |
|
language: |
|
- en |
|
tags: |
|
- 32 bit upscale |
|
- full 32 bit precision |
|
- master files |
|
--- |
|
**Exllamav2** quant (**exl2** / **3.75 bpw**) made with ExLlamaV2 v0.1.3 |
|
|
|
Other EXL2 quants: |
|
| **Quant** | **Model Size** | **lm_head** | |
|
| ----- | ---------- | ------- | |
|
|<center>**[2.2](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-2_2bpw_exl2)**</center> | <center>5594 MB</center> | <center>6</center> | |
|
|<center>**[2.5](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-2_5bpw_exl2)**</center> | <center>6297 MB</center> | <center>6</center> | |
|
|<center>**[3.0](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-3_0bpw_exl2)**</center> | <center>7470 MB</center> | <center>6</center> | |
|
|<center>**[3.5](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-3_5bpw_exl2)**</center> | <center>8640 MB</center> | <center>6</center> | |
|
|<center>**[3.75](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-3_75bpw_exl2)**</center> | <center>9228 MB</center> | <center>6</center> | |
|
|<center>**[4.0](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-4_0bpw_exl2)**</center> | <center>9813 MB</center> | <center>6</center> | |
|
|<center>**[4.25](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-4_25bpw_exl2)**</center> | <center>10398 MB</center> | <center>6</center> | |
|
|<center>**[5.0](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-5_0bpw_exl2)**</center> | <center>12155 MB</center> | <center>6</center> | |
|
|<center>**[6.0](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-6_0bpw_exl2)**</center> | <center>14506 MB</center> | <center>8</center> | |
|
|<center>**[6.5](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-6_5bpw_exl2)**</center> | <center>15688 MB</center> | <center>8</center> | |
|
|<center>**[8.0](https://huggingface.co/Zoyd/DavidAU_Psyonic-Cetacean-V1-20B-Ultra-Quality-Float32-8_0bpw_exl2)**</center> | <center>16737 MB</center> | <center>8</center> | |
|
|
|
<h3> Master Files for Ultra High Quality Remasters of "Psyonic-Cetacean" 20B </h3> |
|
|
|
<img src="science-mad.jpg" width=300px height=300px style="float:right; padding:5px;"> |
|
|
|
May "Space Whale" swim in the oceans of the universe forever! |
|
|
|
This repo contains the full precision (32 bit) master files for 32 bit upscales created by "DavidAU" of: |
|
|
|
https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF-imatrix |
|
|
|
And |
|
|
|
https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF |
|
|
|
Please view either repo for details on the remaster's results, and other important infomation. |
|
|
|
<font color="red"><B>IMPORTANT NOTES For Maximum Results: </b></font> |
|
|
|
These are "final" result files of the full precision rebuild (including end result merge(s)) minus |
|
GGUF and Imatrix level upscaling / adjustments which occuring during "GGUFing" processes. |
|
|
|
If you use these to create your own GGUFs, please use "outfile" at F32 for best results. If |
|
you use F16 this will reduce the quality by a factor of 2 or higher. |
|
|
|
Imatrix processes should use a stable dataset(s) of at least 500 "chunks" or more. |
|
If smaller dataset(s) are used this may corrupt or reduce the quality of the Imatrix builds. |
|
|
|
Due to the precision remaster there will be "greater" distance between each quant - both |
|
non imatrix and imatrix. |
|
|
|
IE: The jump in quality, instruction following, "ai brainpower", nuance and output |
|
between Q4 and Q5 and likewise Q5 and Q6 will be larger than normal. |
|
|
|
Same applies to "Imatrix" quants. |
|
|
|
In addition there will also be differences between exact Imatrix and non-imatrix quants |
|
especially in terms of "creative uses" and/or uses where there is no "right answer". |
|
|
|
Finally, in terms of prompts: |
|
|
|
You may find longer prompts are no longer required and/or you may need to reduce the size |
|
of prompts in usage. This is a factor due to the precision upscale. |
|
|
|
Doing this will ensure the quality of the upscale is maximized in the GGUFs. |
|
|
|
/* GPTQers: |
|
|
|
Suggest 4bit-Act32 TRUE for best results. |
|
|
|
/* EXL2ers: |
|
|
|
Suggest Min 4.5 BPW or higher ; 6 BPW and up is especially potent. |
|
Strongly suggest you do not reduce layer bit count, as this will affect depth and nuance. |
|
The more BPW the better. |
|
|
|
Happy GGUFing, EXL2ing, GPTQing, AWQing, HQQing and of course "Merging". |
|
|
|
<b>LONG LIVE OPEN SOURCE!</B> |
|
|
|
<I>DavidAU</I> |
|
|
|
/* Drop me a note when up, and I will link the masters to your repos. |
