kshitijthakkar/qwen3.5-moe-2.3B-d2B

set tie_word_embeddings=False for GGUF/ollama compatibility

set tie_word_embeddings=True (both tensors identical, matching source models)

set tie_word_embeddings=False (both tensors present in checkpoint)

fix README: add pipeline_tag and image-text-to-text tag

fix config.json: use nested text_config for HF model size detection

Add Qwen3.5 MoE 2.22B (dense-to-MoE from Qwen/Qwen3.5-2B)

initial commit