Qwen3 Embedding 0.6B — CoreML
CoreML conversion of Qwen3-Embedding-0.6B for on-device inference on Apple platforms (macOS / iOS).
Contents
| File | Description |
|---|---|
encoder.mlmodelc/ |
Compiled CoreML model (~1.1 GB) |
config.json |
Model configuration (Qwen3, 28 layers, hidden_size=1024) |
tokenizer.json |
HuggingFace tokenizer |
tokenizer_config.json |
Tokenizer configuration |
vocab.json |
Vocabulary |
merges.txt |
BPE merges |
Model Details
- Architecture: Qwen3ForCausalLM (28 layers, 16 attention heads, 8 KV heads, GQA)
- Hidden size: 1024
- Intermediate size: 3072
- Max sequence length: 32768
- Parameters: ~0.6B
License
Apache 2.0 — following the upstream Qwen3-Embedding-0.6B license.
Source
Converted from Qwen/Qwen3-Embedding-0.6B using coremltools.
- Downloads last month
- 18