READ BELOW:
This model is HEAVILY in beta, And Needs to be finetuned on a RP Dataset to improve replies, But even in this "Raw" State it out proformes any LLama2 13B RP model in my opion.
I would have finetuned this model to avoid Bad replies, BUT I am currently not skilled enough to and cannot access the resources to. SO If you wanna take a gamble at it, Then Be my guest. Just remeber to link back to this and Undi's original model.
This Model was inspired by Cover3-17B
This Model is based on Mistral 7B but using some frankenMerging it was made into a 12.9B model which is about 13~
merged
This is a merge of pre-trained language models created using mergekit.
Merge Details
This model is meant to be basically a smaller version of Clover3-17B for people who cannot run 17B models.
Merge Method
This model was merged using the passthrough merge method.
Models Merged
The following models were included in the merge:
Configuration
The following YAML configuration was used to produce this model:
dtype: float16
merge_method: passthrough
slices:
- sources:
- layer_range: [0, 8]
model: Masterjp123/NeuralMaid-7b
- sources:
- layer_range: [0, 12]
model: lemonilia/AshhLimaRP-Mistral-7B
- sources:
- layer_range: [9, 18]
model: Masterjp123/NeuralMaid-7b
- sources:
- layer_range: [13, 22]
model: lemonilia/AshhLimaRP-Mistral-7B
- sources:
- layer_range: [19, 31]
model: Masterjp123/NeuralMaid-7b
- sources:
- layer_range: [23, 31]
model: lemonilia/AshhLimaRP-Mistral-7B