File size: 1,670 Bytes
8d6c84a a8e551c 8d6c84a a8e551c 8d6c84a 690e1f6 8d6c84a 53288da d27cace 8d6c84a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 |
---
language:
- en
license: llama3
library_name: transformers
tags:
- merge
base_model:
- SillyTilly/Meta-Llama-3.1-70B
- SillyTilly/Meta-Llama-3.1-70B-Instruct
- unsloth/Llama-3.3-70B-Instruct
- SicariusSicariiStuff/Negative_LLAMA_70B
---
#
<img src=https://huggingface.co/altomek/RE-70B-AS3D/resolve/main/RE.png>
<a href="https://www.youtube.com/watch?v=kYje-wdAUsg" title="i_o - Audio Dust" target="_blank">intro music...</a>
## Llama RE-70B-AS3D
I desired a model that would unlock full Llama personality but still could follow instructions.
This is first interesting result from the voyage...
### Ingridients
- [Llama-3.1-70B](https://huggingface.co/SillyTilly/Meta-Llama-3.1-70B)
- [Llama-3.1-70B-Instruct](https://huggingface.co/SillyTilly/Meta-Llama-3.1-70B-Instruct)
- [Llama-3.3-70B-Instruct](https://huggingface.co/unsloth/Llama-3.3-70B-Instruct)
- [Negative_LLAMA_70B](https://huggingface.co/SicariusSicariiStuff/Negative_LLAMA_70B)
### Settings
Use Llama 3 template, model with ALpaca template sometimes halucinates and generates training materials ;P
### Quants
More quants are comming soon... and I need to redo GGUF quants(!) as they do not encode chat template in tokenizer_config... if your inference engine uses chat template from GGUF file you will see halucinations. However GGUFs work fine for me in SillyTawern & text-generation-webui combo...
- [GGUF](https://huggingface.co/altomek/RE-70B-AS3D-GGUF) --> TO BE UPDATED!
- [3.5 BPW](https://huggingface.co/altomek/RE-70B-AS3D-3.5bpw-EXL2)
- [3.75 BPW](https://huggingface.co/altomek/RE-70B-AS3D-3.75bpw-EXL2)
- [4 BPW](https://huggingface.co/altomek/RE-70B-AS3D-4bpw-EXL2)
|