KaraKaraWitch/MachiNoDolphin-Qwen2.5-72b

MachiNoDolphin-Qwen2.5-72b by KaraKaraWitch is a 72.7 billion parameter merged language model based on the Qwen2.5-72B architecture, created using the TIES merge method. This model integrates several pre-trained models, including Evathene-v1.0, SteyrCannon-0.2-Qwen2.5-72b, banana-2-b-72b, shuttle-3, and Chronos-Platinum-72B. It is designed to be a versatile, mostly uncensored chat-based model, supporting a wide range of languages including English, Chinese, French, Spanish, and more.

Warm
Public
72.7B
FP8
131072
Hugging Face