empero-ai/Qwen3.8-4B-Distill
Full fine-tune vs Qwen/Qwen3.5-4B. Fine-tune: 90% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/5, MLP 96/199, Attention 177/324, Norms 51/209, other 0/1
27 derivatives checked
- empero-ai/Qwen3.8-4B-Distill-GGUF: Same weights, converted
- mradermacher/Qwen3.8-4B-Distill-GGUF: Same weights, converted
- mradermacher/Qwen3.8-4B-Distill-i1-GGUF: Quantization
- Jadson/Qwen3.5-4B-Prime-GGUF: Quantization
- Atomic-Germ/Qwen3.8-4B-Distill-GGUF: Same weights, converted
- caentzminger/Qwen3.8-4B-Distill-Obliterated-GGUF: Quantization
- hakergeniusz/Qwen3.8-4B-Distill-GGUF: Same weights, converted
- CaseD0rsett/Qwen3.8-4B-Distill-Heretic-Abliterated-MLX-4bit: Quantization
- hakergeniusz/Qwen3.8-4B-Heretic-GGUF: Abliterated
- ermiaazarkhalili/Qwen3.8-4B-Function-Calling-xLAM-Unsloth-GGUF: Full fine-tune
- ermiaazarkhalili/Qwen3.8-4B-SFT-Fable5-Glint-GGUF: Full fine-tune
- yachen4ever/Qwen3.8-4B-Distill-Heretic-Abliterated-MLX-4bit: Quantization
- SiddhJagani/Qwen3.8-4B-mlx-4Bit: Quantization
- nvythong/Qwen3.8-4B-Distill-mlx-4Bit: Quantization
- iBotIA/Qwen3.8-4B-Empero-AI-FullStack-GGUF: Quantization
- SiddhJagani/Qwen3.8-4B-mlx-8Bit: Quantization
- Flowsink/Qwen3.8-4B-Distill-GGUF: Same weights, converted
- Vishva007/Qwen3.8-4B-Distill-W4A16-AutoRound-LLM-Compressor: Quantization
- SiddhJagani/Qwen3.8-4B-mlx-2Bit: Quantization
- ermiaazarkhalili/Qwen3.8-4B-SFT-Fable5-Glint: LoRA-style
- SiddhJagani/Qwen3.8-4B-mlx-6Bit: Quantization
- nicolasramos/Qwen3.8-4B-Distill-MLX-4bit-mtp: Quantization
- SiddhJagani/Qwen3.8-4B-mlx-3Bit: Quantization
- SiddhJagani/Qwen3.8-4B-mlx-fp16: Partial fine-tune (fewer layers)
- olluyeancah/margin-pilot1-gguf: Quantization
- iBotIA/Qwen3.8-4B-Empero-AI-FullStack: LoRA-style
- ermiaazarkhalili/Qwen3.8-4B-SFT-Fable5: LoRA-style