FuseChat: Knowledge Fusion of Chat Models
Paper
•
2408.07990
•
Published
•
14
This is a merge of pre-trained language models created using mergekit.
This model was merged using the SCE merge method using Jiayi-Pan/Tiny-Vicuna-1B as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
merge_method: sce
models:
- model: Jiayi-Pan/Tiny-Vicuna-1B
- model: yihanwang617/tinyllama-sft-vicuna-random-100k
- model: NickyNicky/cognitivecomputations_TinyDolphin-2.8-1.1b
- model: QuixiAI/TinyDolphin-2.8.2-1.1b
base_model: Jiayi-Pan/Tiny-Vicuna-1B
tokenizer:
source: base
parameters:
select_topk: 1.1
dtype: float32
out_dtype: bfloat16
normalize: true