NANBEIGE LYRICAL TRANSLATOR (Rus2Eng) & General Assistant

Experimental Nanbeige4 MoE Merge via SoonMerger

By Aleksey Calvin Tsukanov & SilverAgePoets.com Combines Nanbeige/Nanbeige4-3B-Thinking-2511, arnomatic/Nanbeige4-3B-Thinking-2511-heretic, AlekseyCalvin/Lyrical_ru2en_Nanbeige4-3B-Thinking-Ties_SFT (our lyrical finetune).
Merged using our versatile SoonMerger Toolkit GUI space, which features easy to use implementations of multiple frameworks and toolkits for merging, extracting, and modifying model weights (including advanced MergeKit methods).
One of the merged-in expert checkpoints fine-tuned over our custom bilingual translations dataset.

MERGING CONFIG:

base_model: Nanbeige/Nanbeige4-3B-Thinking-2511
gate_mode: cheap_embed
dtype: bfloat16
experts:
- source_model: Nanbeige/Nanbeige4-3B-Thinking-2511
  positive_prompts:
  - You are a helpful assistant
- source_model: AlekseyCalvin/Lyrical_ru2en_Nanbeige4-3B-Thinking-Ties_SFT
  positive_prompts:
  - Translate the following poem
- source_model: arnomatic/Nanbeige4-3B-Thinking-2511-heretic
  positive_prompts:
  - Write a story for me
tokenizer_source: union
Downloads last month
24
Safetensors
Model size
9B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for AlekseyCalvin/Lyrical_ru2en_NanbeigeMoeTest

Dataset used to train AlekseyCalvin/Lyrical_ru2en_NanbeigeMoeTest