Downloads · 30 days
0
Kevletesteur/chimere-dflash-data
chimere-dflash-data is a machine learning model from Kevletesteur. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Prompt datasets used to train the DFlash block diffusion drafter for speculative decoding on Qwen3.5-35B-A3B.
Downloads · 30 days
0
Access
Public
Updated Mar 29, 2026
Repo size
—
Likes
0
Public
Click a slice to open those files.
.jsonl6.5 MB · 100%
From the Hugging Face model README
Prompt datasets used to train the DFlash block diffusion drafter for speculative decoding on Qwen3.5-35B-A3B.
all_prompts.jsonl — 3,927 diverse prompts (5.1 MB)holdout_v8_500.jsonl — 500 holdout prompts for evaluationeval_holdout_200.jsonl — 200 eval promptseval_prompts.jsonl — 500 eval promptsdiverse_prompts.jsonl — 140 diversity-focused promptsDFlash drafter trained on these prompts achieves τ = 9.4 tokens/step offline (+47% vs the original DFlash paper's τ ≈ 6.4).
See chimere for the full code.
Kevin Remondiere — Independent ML researcher