Qwen3.6-35B-A3B-DFlash GGUF

Q8_0 GGUF conversion of z-lab/Qwen3.6-35B-A3B-DFlash for use with Magnitude.

Converted with llama.cpp at revision 8e7f22b67ef4667b4ddd50230771287f328cfb3f.

Artifact

• Quantization: Q8_0
• SHA-256: bf9215cd0aa077261c9cabe74b9ce43e9b5325a746ab65569511fadf02c3cb4f

Draft metadata

• dflash.attention.causal = [true, true, true, true, true, false]: the first five layers use causal attention; the last uses bidirectional attention.
• dflash.sample_from_anchor = false: proposals come from mask rows, starting after the anchor.
• Derived from the checkpoint configuration and Z-Lab’s reference implementation; verified against reference outputs.