empero-ai/Qwen3.8-2B-Distill
Partial fine-tune vs Qwen/Qwen3.5-2B. Partial change: 83% of the weights changed (attention, embeddings, mlp, norms), in 24 of 24 layers.
Tensors changed: Attention 132/268, Norms 43/183, Embeddings 1/5, MLP 72/175, other 0/1
20 derivatives checked
- empero-ai/Qwen3.8-2B-Distill-GGUF: Same weights, converted
- Atomic-Germ/Qwen3.8-Distilled-2B-NPU2: Quantization
- piskle/Qwen3.8-2B-Distill-GGUF: Same weights, converted
- Atomic-Germ/Qwen3.8-2B-Distill-GGUF: Same weights, converted
- dalatexcoder/Qwen3.8-2B-MLX-4.83B: Quantization
- ermiaazarkhalili/Qwen3.8-2B-SFT-Fable5-GGUF: Full fine-tune
- ermiaazarkhalili/Qwen3.8-2B-SFT-Fable5-Glint-GGUF: Full fine-tune
- ermiaazarkhalili/Qwen3.8-2B-Function-Calling-xLAM-Unsloth-GGUF: Full fine-tune
- keXjos/Qwen3.8-2B-Distill-mlx-5Bit: Quantization
- SiddhJagani/Qwen3.8-2B-mlx-4Bit: Quantization
- nvythong/Qwen3.8-2B-Distill-mlx-4Bit: Quantization
- SiddhJagani/Qwen3.8-2B-mlx-8Bit: Quantization
- vr3n/Qwen3.8-2B-Distill-GGUF: Same weights, converted
- JeXTCy1/Qwen3.8-2B-Distill-GGUF: Same weights, converted
- SiddhJagani/Qwen3.8-2B-mlx-fp16: Partial fine-tune (fewer layers)
- SiddhJagani/Qwen3.8-2B-mlx-3Bit: Quantization
- dalatexcoder/Qwen3.8-2B-MLX-4.00B: Quantization
- MihaiPopa-1/Qwen3.8-2B-Heretic-Max: Abliterated
- ermiaazarkhalili/Qwen3.8-2B-Function-Calling-xLAM-Unsloth: LoRA-style
- MihaiPopa-1/Qwen3.8-2B-Heretic-Balanced: Abliterated