Qwen/Qwen3-Embedding-0.6B
LoRA-style vs Qwen/Qwen3-0.6B-Base. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 28 of 28 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/113, MLP 84/84, Attention 112/112
67 derivatives checked
- mlx-community/Qwen3-Embedding-0.6B-4bit-DWQ: Quantization
- Octen/Octen-Embedding-0.6B: Full fine-tune
- pipizhao/SkillRouter-Embedding-0.6B: Partial fine-tune
- mlx-community/Qwen3-Embedding-0.6B-8bit: Quantization
- kwisschen/qwen3-embedding-0.6b-patent-oa-rejections: LoRA-style
- furiosa-ai/Qwen3-Embedding-0.6B: Identical copy
- PeterAM4/Qwen3-Embedding-0.6B-GGUF: Same weights, converted
- futur/Qwen3-Embedding-0.6B-model2vec-onnx: Different architecture
- OpenMed/OpenMed-PII-QwenMed-XLarge-600M-v1: Partial fine-tune
- Hanno-Labs/dinghy-law-0.6b-v1: LoRA-style
- mradermacher/Qwen3-Embedding-0.6B-GGUF: Same weights, converted
- xthor/Qwen3-Embedding-0.6B-GraphQL: Partial fine-tune
- mradermacher/Qwen3-Embedding-0.6B-i1-GGUF: Quantization
- ThakiCloud/SKILLRET-Embedding-0.6B: Partial fine-tune
- drawais/Qwen3-Embedding-0.6B-AWQ-INT4: Quantization
- zhiqian99/Qwen3-Embedding-0.6B-GGUF: Same weights, converted
- Casual-Autopsy/Qwen3-Embedding-0.6B-GGUFs: Same weights, converted
- enacimie/Qwen3-Embedding-0.6B-Q4_K_M-GGUF: Quantization
- HIT-TMG/JevEmbed-Qwen3-Embedding-0.6B: LoRA-style
- mhiveai/Qwen-Insure: Partial fine-tune
- tencent/R3-embedding-0.6b: Partial fine-tune
- cstr/qwen3-embed-0.6b-GGUF: Quantization
- npario/Qwen3-Embedding-0.6B-8bit: Quantization
- farbodtavakkoli/OTel-Embedding-600M: Partial fine-tune
- JusteLeo/Qwen3-0.6B-T5-xxl-GGUF: Partial fine-tune
- ZhishanQ/qwen3-embedding-redundancy-detector-0.6B: Partial fine-tune
- binedge/Qwen3-Embedding-0.6B-FP8: Quantization
- batiai/Qwen3-Embedding-0.6B-GGUF: Quantization
- AQ-MedAI/Diver-Retriever-0.6B: Full fine-tune
- Gideon531/Qwen3-Embedding-0.6B-Q8_0-GGUF: Quantization
38 other repos hold byte-for-byte identical weight files.