Qwen/Qwen3-30B-A3B
Full fine-tune vs Qwen/Qwen3-30B-A3B-Base. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 193/193, MLP 2613/2613, Attention 192/192, lm_head 1/1
80 derivatives checked
- MaziyarPanahi/Qwen3-30B-A3B-GGUF: Quantization
- Qwen/Qwen3-30B-A3B-FP8: Quantization
- Qwen/Qwen3-30B-A3B-GGUF: Quantization
- nytopop/Qwen3-30B-A3B.w8a8: Quantization
- Qwen/Qwen3-30B-A3B-GPTQ-Int4: Quantization
- unsloth/Qwen3-30B-A3B-GGUF: Same weights, converted
- nvidia/Qwen3-30B-A3B-NVFP4: Quantization
- bartowski/Qwen_Qwen3-30B-A3B-GGUF: Same weights, converted
- QuixiAI/Qwen3-30B-A3B-AWQ: Quantization
- inference-optimization/Qwen3-1.6B-A0.9B: Full fine-tune (fewer layers)
- Lucy-in-the-Sky/Qwen3-30B-A3B-Q4_K_M-GGUF: Quantization
- lmstudio-community/Qwen3-30B-A3B-MLX-4bit: Quantization
- lmstudio-community/Qwen3-30B-A3B-MLX-8bit: Quantization
- DavidAU/Qwen3-128k-30B-A3B-NEO-MAX-Imatrix-gguf: Quantization
- BCCard/Qwen3-30B-A3B-FP8-Dynamic: Quantization
- lmstudio-community/Qwen3-30B-A3B-GGUF: Quantization
- RedHatAI/Qwen3-30B-A3B-FP8-block: Quantization
- unsloth/Qwen3-30B-A3B-bnb-4bit: Quantization
- mlx-community/Qwen3-30B-A3B-4bit: Quantization
- unsloth/Qwen3-30B-A3B-128K-GGUF: Same weights, converted
- RedHatAI/Qwen3-30B-A3B-NVFP4: Quantization
- ggml-org/Qwen3-30B-A3B-GGUF: Same weights, converted
- INCModel/Qwen3-30B-A3B-12L-MXFP8-test: Quantization
- groxaxo/Qwen3.6-35B-A3B-Abliterated-Heretic-GGUF: Not comparable yet
- RedHatAI/Qwen3-30B-A3B-quantized.w4a16: Quantization
- mlx-community/Qwen3-30B-A3B-8bit: Quantization
- ELVISIO/Qwen3-30B-A3B-AWQ: Quantization
- unsloth/Qwen3-30B-A3B: Same weights, repacked
- mlabonne/Qwen3-30B-A3B-abliterated: LoRA-style
- RedHatAI/Qwen3-30B-A3B-FP8-dynamic: Quantization
11 other repos hold byte-for-byte identical weight files.