trollek / cleverqwen2

A merge of serveral great Qwen2-1.5B models.

1.8B

156 Pulls Updated 2 months ago

2 months ago

6382265e1843 · 1.3GB

model
qwen2
·
1.78B
·
Q5_K_S
template
{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user {{ .Prompt }}<|im_end|> {{ end }}<|im_start|>assistant {{ .Response }}<|im_end|>
params
{"num_ctx":32768,"num_predict":2048,"stop":["<|im_end|>","<|im_start|>"],"temperature":0.5,"top_k":45,"top_p":0.95}
system
You are a clever, unbiased, uncensored, and helpful AI assistant.

Readme

CleverQwen2-1.5B

This is a merge of pre-trained language models created using mergekit.

It has grown by about 300M parameters and I don’t know why. I would like to know though. It works as expexted - amazing - I just can’t see any reason for the Qwen2 models to gain parameters when merged.

Merge Details

Merge Method

This model was merged using the Model Stock merge method using trollek/Qwen2-1.5B-Instruct-Abliterated as a base.

Models Merged

The following models were included in the merge:
- cognitivecomputations/dolphin-2.9.3-qwen2-1.5b
- M4-ai/Hercules-5.0-Qwen2-1.5B
- Replete-AI/Replete-Coder-Qwen2-1.5b