DSpark-family drafter (dspark), block size 7, 5 draft layers, target tap layers [1, 12, 23, 34, 45]. Trained on the open-perfectblend corpus regenerated by Qwen3-30B-A3B (1,339,038 cache-valid rows) with a cosine LR schedule spanning 10 epochs (maxtrainsteps 26150, 2615 steps per epoch). Checkpoints at epoch boundaries live under epochN/step{N2615}/; each sits at that step of the 10-epoch schedule. Any epochN/stepS/config.json + model.safetensors pair loads like the DSpark paper's released drafters. Serve together with the target model Qwen/Qwen3-30B-A3B. 1. Rebuild the activation cache from the dataset repo xf15/ssd-perfectblend-qwen3-30b-a3b-regen with the DSpark cache builder…
Independent publisher
Xinze Feng
xf15
Models in Library1
Datasets in Library0
Models on Hugging Face5
Followers—
Compare the published evidence
This profile connects the 1 model and dataset record shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.