Tarek07's picture
Upload folder using huggingface_hub
5009f45 verified
metadata
base_model:
  - Tarek07/Legion-V2.1-LLaMa-70B
library_name: transformers
tags:
  - mergekit
  - merge

merge

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

Merge Details

Merge Method

This model was merged using the Passthrough merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

slices:
- sources:
  - layer_range: [0, 4]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [2, 4]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [4, 8]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [6, 8]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [8, 12]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [10, 12]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [12, 16]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [14, 16]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [16, 20]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [18, 20]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [20, 24]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [22, 24]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [24, 28]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [26, 28]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [28, 32]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [30, 32]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [32, 36]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [34, 36]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [36, 40]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [38, 40]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [40, 44]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [42, 44]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [44, 48]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [46, 48]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [48, 52]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [50, 52]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [52, 56]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [54, 56]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [56, 60]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [58, 60]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [60, 64]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [62, 64]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [64, 68]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [66, 68]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [68, 72]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [70, 72]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [72, 76]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [74, 76]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
- sources:
  - layer_range: [76, 80]
    model: Tarek07/Legion-V2.1-LLaMa-70B
- sources:
  - layer_range: [78, 80]
    model: Tarek07/Legion-V2.1-LLaMa-70B
    parameters:
      scale:
      - filter: o_proj
        value: 0
      - filter: down_proj
        value: 0
      - value: 1
merge_method: passthrough
dtype: bfloat16