waldie/Qwen2.5-32B-EVA-Instruct-Merge-0.1-4bpw-h6-exl2

Qwen2.5-32B-EVA-Instruct-Merge-0.1

This is a merge of EVA 32B 0.1 with Qwen's 32B instruct model, and EVA 0.0, at low weights, using mergekit.

Also see: https://huggingface.co/ParasiticRogue/EVA-Instruct-32B

Merge Details

Merge Method

This model was merged using the della merge method using /home/a/Models/Raw/Qwen_Qwen2.5-32B as a base.

Models Merged

The following models were included in the merge:

• /home/a/Models/Raw/EVA-UNIT-01_EVA-Qwen2.5-32B-v0.1
• /home/a/Models/Raw/Qwen_Qwen2.5-32B-Instruct
• /home/a/Models/Raw/EVA-UNIT-01_EVA-Qwen2.5-32B-v0.0

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: /home/a/Models/Raw/Qwen_Qwen2.5-32B
  # No parameters necessary for base model
  - model: /home/a/Models/Raw/EVA-UNIT-01_EVA-Qwen2.5-32B-v0.1
    parameters:
      weight: 0.7
      density: 0.7
  - model: /home/a/Models/Raw/EVA-UNIT-01_EVA-Qwen2.5-32B-v0.0
    parameters:
      weight: 0.11
      density: 0.3
  - model: /home/a/Models/Raw/Qwen_Qwen2.5-32B-Instruct
    parameters:
      weight: 0.19
      density: 0.3
merge_method: della
#tokenizer_source: base
base_model: /home/a/Models/Raw/Qwen_Qwen2.5-32B
parameters:
  int8_mask: true
  epsilon: 0.15
  lambda: 1
dtype: bfloat16