sometimesanotion/Qwentessential-14B-v3

Notes

This is not intended for end usage, unless you specifically know why you'd want to use a no-frills Qwen2.5 model, merged using highly stable methods from proven ancestors. It's a baseline merge component.

Merge Details

Merge Method

This model was merged using the TIES merge method using merges/Qwentessential-14B-slerp as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

base_model: merges/Qwentessential-14B-slerp
dtype: float32
merge_method: ties
name: Qwentessential-14B-v3
out_dtype: bfloat16
parameters:
  density: 1.0
  int8_mask: true
  normalize: true
  rescale: false
  weight: 1.0
slices:
- sources:
  - layer_range:
    - 0
    - 2
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 2
    - 6
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 6
    - 10
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 10
    - 14
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 14
    - 18
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 18
    - 22
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 22
    - 26
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 26
    - 30
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 30
    - 34
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 34
    - 38
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 38
    - 42
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 42
    - 46
    model: merges/Qwentessential-14B-slerp
- sources:
  - layer_range:
    - 46
    - 48
    model: merges/Qwentessential-14B-slerp
tokenizer_source: base