13 lines
263 B
YAML
13 lines
263 B
YAML
|
models:
|
||
|
|
- model: Qwen/Qwen2.5-Coder-7B-Instruct
|
||
|
|
parameters:
|
||
|
|
density: 0.50
|
||
|
|
weight: 0.50
|
||
|
|
merge_method: ties
|
||
|
|
base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
|
||
|
|
parameters:
|
||
|
|
normalize: true
|
||
|
|
int8_mask: true
|
||
|
|
dtype: bfloat16
|
||
|
|
tokenizer_source: base
|