Modelpublic

yuhengtu-bytedance/DataDecide-dolma1_7-no-math-code-1B-55000_57500_60000_62500_65000_weightedavg_merge

sourceHugging Faceupdated 12d agoView on Hugging Face

0likes106downloads

Model Card

DataDecide-dolma17-no-math-code-1B-5500057500600006250065000weightedavg_merge

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

Merge Details

Merge Method

This model was merged using the Linear merge method using /opt/tiger/PanSafetyBetterMeasurement/mergescalingckptscache/sourceckpts/dolma17-no-math-code/step65000 as a base.

Models Merged

The following models were included in the merge:

—/opt/tiger/PanSafetyBetterMeasurement/mergescalingckptscache/sourceckpts/dolma17-no-math-code/step57500
—/opt/tiger/PanSafetyBetterMeasurement/mergescalingckptscache/sourceckpts/dolma17-no-math-code/step60000
—/opt/tiger/PanSafetyBetterMeasurement/mergescalingckptscache/sourceckpts/dolma17-no-math-code/step62500
—/opt/tiger/PanSafetyBetterMeasurement/mergescalingckptscache/sourceckpts/dolma17-no-math-code/step55000

Configuration

The following YAML configuration was used to produce this model:

yaml

models:
- model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step55000
  parameters:
    weight: 1
- model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step57500
  parameters:
    weight: 2
- model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step60000
  parameters:
    weight: 3
- model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step62500
  parameters:
    weight: 4
- model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step65000
  parameters:
    weight: 5
base_model: /opt/tiger/Pan_Safety_Better_Measurement/merge_scaling_ckpts_cache/source_ckpts/dolma1_7-no-math-code/step65000
merge_method: linear
parameters:
  normalize: true
dtype: float32
out_dtype: bfloat16