Qwen/Qwen3-8B
Full fine-tune vs Qwen/Qwen3-8B-Base. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 143/145, MLP 108/108, Attention 144/144, lm_head 1/1
574 derivatives checked
- Qwen/Qwen3-8B-AWQ: Quantization
- Qwen/Qwen3-8B-FP8: Quantization
- Qwen/Qwen3-8B-GGUF: Quantization
- AnkitAI/Parable-Qwen3-8B-Claude-Fable-5-GGUF: LoRA-style
- MaziyarPanahi/Qwen3-8B-GGUF: Same weights, converted
- mlx-community/Qwen3-8B-4bit: Quantization
- nvidia/Qwen3-8B-NVFP4: Quantization
- unsloth/Qwen3-8B-unsloth-bnb-4bit: Quantization
- unsloth/Qwen3-8B-GGUF: Same weights, converted
- unsloth/Qwen3-8B: Same weights, repacked
- lmstudio-community/Qwen3-8B-MLX-4bit: Quantization
- lmstudio-community/Qwen3-8B-MLX-8bit: Quantization
- allenai/Molmo2-8B: Different architecture
- Qwen/Qwen3Guard-Gen-8B: Full fine-tune
- lmstudio-community/Qwen3-8B-GGUF: Quantization
- bartowski/Qwen_Qwen3-8B-GGUF: Same weights, converted
- huihui-ai/Huihui-Qwen3-8B-abliterated-v2: Abliterated
- unsloth/Qwen3-8B-bnb-4bit: Quantization
- luckypoolio/Qwen3-8B-Q8_0-full-v2.gguf: Quantization
- Open-Bee/Bee-8B-RL: Full fine-tune (fewer layers)
- Aldaris/Qwen3-8B-Q4_K_M-GGUF: Quantization
- allenai/MolmoPoint-8B: Different architecture
- DreamFast/qwen3-8b-heretic: Abliterated
- RedHatAI/Qwen3-8B-quantized.w4a16: Quantization
- RefalMachine/RuadaptQwen3-8B-Hybrid: LoRA-style
- inference-optimization/Qwen3-8B-from-Qwen3-8B_regen-speculators.eagle3-qwen3arch-ckpt1: Partial fine-tune
- inference-optimization/Qwen3-8B-speculators.peagle-qwen3arch-ckpt4: Full fine-tune
- t-tech/T-lite-it-2.1: Full fine-tune
- RedHatAI/Qwen3-8B-FP8-dynamic: Quantization
- reaperdoesntknow/Symbiotic-8B: Same weights, converted
92 other repos hold byte-for-byte identical weight files.