Qwen/Qwen3.5-35B-A3B
Full fine-tune vs Qwen/Qwen3.5-35B-A3B-Base. Fine-tune: 99% of the weights changed, across 40 of 40 layers.
Tensors changed: MLP 1057/1165, lm_head 1/1, Embeddings 3/5, Attention 258/392, other 1/1, Norms 138/247
193 derivatives checked
- Qwen/Qwen3.5-35B-A3B-FP8: Quantization
- unsloth/Qwen3.5-35B-A3B-GGUF: Same weights, converted
- Qwen/Qwen3.5-35B-A3B-GPTQ-Int4: Quantization
- HauhauCS/Qwen3.5-35B-A3B-Uncensored-HauhauCS-Aggressive: LoRA-style
- bartowski/Qwen_Qwen3.5-35B-A3B-GGUF: Same weights, converted
- AxionML/Qwen3.5-35B-A3B-NVFP4: Quantization
- lmstudio-community/Qwen3.5-35B-A3B-GGUF: Quantization
- mudler/Qwen3.5-35B-A3B-APEX-GGUF: Quantization
- QuantTrio/Qwen3.5-35B-A3B-AWQ: Quantization
- cyankiwi/Qwen3.5-35B-A3B-AWQ-4bit: Quantization
- DrMoriarty0/Qwen3.5-35B-A3B-GGUF: Quantization
- unsloth/Qwen3.5-35B-A3B-MTP-GGUF: Same weights, converted
- apodex/Apodex-1.0-mini: Full fine-tune
- txn545/Qwen3.5-35B-A3B-NVFP4: Quantization
- z-lab/Qwen3.5-35B-A3B-DFlash: Different architecture
- DJLougen/hermes-qwen3.5-35b-a3b-GGUF: Same weights, converted
- apodex/Apodex-1.1-mini: Full fine-tune
- ugreen-ai/Qwen3.5-35B-A3B-GGUF: Quantization
- apolo13x/Qwen3.5-35B-A3B-quantized.w4a16: Quantization
- unsloth/Qwen3.5-35B-A3B: Identical copy
- codecho/Qwen3.5-35B-A3B-text-only: Same weights, repacked (fewer layers)
- huihui-ai/Huihui-Qwen3.5-35B-A3B-abliterated: Abliterated
- byteshape/Qwen3.5-35B-A3B-GGUF: Quantization
- mradermacher/Qwen3.5-35B-A3B-GGUF: Quantization
- apodex/Apodex-1.1-mini-GGUF: Full fine-tune
- ggml-org/Qwen3.5-35B-A3B-GGUF: Quantization
- tomngdev/MTP-Qwen3.5-35B-A3B-GGUF: Not comparable yet
- mudler/Qwen3.5-35B-A3B-APEX-TQ-GGUF: Quantization
- vcruz305/Apodex-1.1-mini-GGUF: Full fine-tune
- mradermacher/Qwen3.5-35B-A3B-i1-GGUF: Quantization
40 other repos hold byte-for-byte identical weight files.