| tags: | |
| - Qwen3-MoE | |
| - safetensors | |
| **Qwen3-30B-A3B-W4AFP8** | |
| This model is a mixed-precision quantized Qwen3-30B-A3B, with dense layer using FP8_BLOCK_SCALING, MoE layers uses INT4 weights and FP8 activation. |
| tags: | |
| - Qwen3-MoE | |
| - safetensors | |
| **Qwen3-30B-A3B-W4AFP8** | |
| This model is a mixed-precision quantized Qwen3-30B-A3B, with dense layer using FP8_BLOCK_SCALING, MoE layers uses INT4 weights and FP8 activation. |