Qwen SWE Drafter (DFlash2, block 8)

DFlash2 speculative-decoding drafter for Qwen/Qwen3.6-35B-A3B-FP8, trained for software-engineering and agentic traffic. The target model is unchanged. The drafter only proposes tokens and the target verifies every one, so correctness is unchanged.

  • Architecture: DFlash2DraftModel, block size 8, 6 layers, 8 target taps [1,6,11,16,22,27,32,37]
  • Draft window: 8192
  • Weights: BF16 (model.safetensors, ~1.05 GB)
  • Warm start: intermediate Qwen3.6 drafter checkpoint step 500
  • Training: one pass over 33,868 unseen turns (SWE code-write, agentic support, Open-SWE), soft-label weight 0.50, LR 1e-4, 1048 steps

Background and measurements: Speculative decoding for Qwen3.6 on CoralBricks.

Serving

Step-by-step instructions for serving this drafter with the target on one H100 β€” stock SGLang, CUDA 13 image plus a CUDA 12.9 fallback, required flags, sizing and troubleshooting β€” are in SERVE.md.

Files

  • model.safetensors β€” BF16 weights
  • config.json β€” DFlash2 config (block 8, taps, mask token 248077)
  • SERVE.md β€” serving playbook
Downloads last month
127
Safetensors
Model size
0.5B params
Tensor type
BF16
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for coral-bricks-ai/qwen-swe-drafter

Finetuned
(13)
this model