GSHT-GEMMAMA-16B / mergekit_config.yml
djuna's picture
Duplicate from mergekit-community/GSHT-GEMMAMA-9B
064c8f8 verified
raw
history blame contribute delete
379 Bytes
slices:
- sources:
- layer_range: [0, 14]
model: djuna/G2-GSHT
- sources:
- layer_range: [7, 21]
model: djuna/Gemma-2-gemmama-9b
- sources:
- layer_range: [14, 28]
model: djuna/G2-GSHT
- sources:
- layer_range: [21, 35]
model: djuna/Gemma-2-gemmama-9b
- sources:
- layer_range: [28, 42]
model: djuna/G2-GSHT
merge_method: passthrough
dtype: bfloat16