Independent Core AI conversion of vllm-sr/d3-mini, fine-tuned from
Qwen/Qwen3.5-4B, for text and image decisions.
Upstream: https://huggingface.co/vllm-sr/d3-mini
Revision: bd0090dd758d32ba59ef5c749910a8e87fc4391f

The upstream checkpoint, upstream code and converted weights are governed by
the Apache License 2.0. A copy of the license accompanies this distribution.
Each bundle's checkpoint folder redistributes the upstream code, tokenizer,
configuration, readout and license files unmodified, and the token embeddings
and vision position table from the upstream weights.

The language backbone runs as FP16 Core AI stages and the vision tower as an
FP16 Core AI asset on the Apple Neural Engine. Tokenization, image
preprocessing, position tables, token-embedding lookup and option readout run
on the host CPU. The model has not been retrained. This project is not
affiliated with or endorsed by vllm-sr or Qwen.