datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
flash-moe-weights
Flash-MoE Weights: Qwen3.5-397B-A17B (4-bit, Metal)
Pre-packed weights for flash-moe — a pure C/Metal inference engine that runs the 397B-parameter Qwen3.5 MoE model on a single MacBook.
Source Model
mlx-community/Qwen3.5-397B-A17B-4bit
File Structure
File
Size
Description
model_weights.bin
5.5 GB
Non-expert weights (embeddings, attention, norms, shared expert, routing gates)
model_weights.json
371 KB
Tensor manifest (offsets, shapes, dtypes)… See the full description on the dataset page: https://huggingface.co/datasets/prykin/flash-moe-weights.FlashMoE-routing-historyFlashMoE-data