MachiNoDolphin-Qwen2.5-72b

MachiNoDolphin-Qwen2.5-72b

<div> <a href="https://www.youtube.com/watch?v=qMQ-y9dHE2k"><img src="https://cdn-uploads.huggingface.co/production/uploads/633e85093a17ab61de8d9073/SXuoacvkkgY1tdagpsdlr.png" style="margin-left:auto;margin-right:auto"></a> </div>

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

Pulling in shuttle and Evathene for funnsies. Please ignore model.

30/11/24: Heard that people like this model. So I guess, don't ignore it and give it a try? Also, added Featherless link and GGUF for this. Prompt format is ChatML like most Qwen 2.5 chat based models.

Merge Details

Prompting

Chat format is ChatML. It is mostly uncensored. For 99% of the time, you shouldn't run into any issues. For that 1%, just change your system prompt.

Settings I use in general:

Temp: 1.3-1.2
MinP: 0.05
TopA: 0.2
RepPen: 1.05

Rest is disabled/not used.

Merge Method

This model was merged using the TIES merge method using Qwen/Qwen2.5-72B as a base.

Models Merged

The following models were included in the merge:

• sophosympatheia/Evathene-v1.0
• KaraKaraWitch/SteyrCannon-0.2-Qwen2.5-72b
• m8than/banana-2-b-72b
• shuttleai/shuttle-3
• ZeusLabs/Chronos-Platinum-72B

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: KaraKaraWitch/SteyrCannon-0.2-Qwen2.5-72b
    parameters:
      density: 0.25
      weight: 0.5
  - model: ZeusLabs/Chronos-Platinum-72B
    parameters:
      density: 0.5
      weight: 0.75
  - model: m8than/banana-2-b-72b
    parameters:
      density: 0.65
      weight: 0.40
  - model: shuttleai/shuttle-3
    parameters:
      density: 0.65
      weight: 0.40
  - model: sophosympatheia/Evathene-v1.0
    parameters:
      density: 0.65
      weight: 0.40

merge_method: ties
base_model: Qwen/Qwen2.5-72B
parameters:
  normalize: true
dtype: bfloat16