ExllamaV2 version of the model created by Sao10K!

Original Model https://huggingface.co/Sao10K/WinterGoddess-1.4x-70B-L2

Requires ExllamaV2, which is being developed by turboderp https://github.com/turboderp/exllamav2 under an MIT license.


Winter Goddess - A 70B L2 Model for General use, or for Roleplay.

I wanted a Smart Model that is Capable of following Instructions, while being able to (e)RP effectively. Sort of like 1.3, but better.

I merged some models as a base, and had tuned on top of it afterwards.

I personally think this mogs Euryale 1.3, but ymmv.


For Transparency's Sake:

Models Used:
Platypus2-70B-instruct
Lila-70B
SunsetBoulevard (at roughly 0.1 weight, boosting coherency)
Private De-alignment LoRA on top.

why does it show mergekit in the safetensors.index metadata? -> I used DARE method to merge the 3 models. Then Axolotl qLoRA. then used lora-merge, copying the files of the base merged model because they didn't save to the new one, only the .safetensor files got saved.


Prompt Format - Alpaca

### Instruction:
<Prompt>

### Response:

OR

### Instruction:
<Prompt>

### Input:
<Insert Context Here>

### Response:


42. A 25-year-old female has been struck in the right eye with a pipe. She has a ruptured right globe, an orbital fracture and no other obvious injury. You should bandage:


A) The right eye tightly
B) Both eyes loosely
C) The right eye loosely
D) Both eyes tightly

Downloads last month
14
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.