specific method · filed under inference & serving
DSpark
A speculative-decoding architecture that combines semi-autoregressive draft generation with confidence-scheduled verification.
Also called DSpark speculative decoding, DSpark speculative-decoding module, DSpark speculative decoding architecture.
- sources
- 8
- models
- 5
- labs adopt it
- 2
- strongest
- core
How sources treat it
One count per evidence span, weakest treatment to strongest.
Documented in
Further reading
Picked by hand, not extracted: where to read more, not evidence for anything on this page.
Evidence
10 spans quoted from the sources, strongest treatment first.
We equip DeepSeek-V4.1-Flash with DSpark (Cheng et al., 2026a), a speculative decoding module that combines semi-autoregressive drafting with confidence-scheduled verification.
The repo has the DSpark speculative-decoding module attached
We also introduce the DSpark (Cheng et al., 2026a) speculative decoding architecture to improve decoding efficiency through semi-autoregressive draft generation and confidence-scheduled verification.
We omit the MTP module during backbone pre-training and use DSpark (Cheng et al., 2026a) for speculative decoding.
DSpark speculative decoding is enabled with a single flag
DSpark speculative decoding (semi-autoregressive draft generation with confidence-scheduled verification)
Beyond MTP, two draft models are provided with Nemotron 3.5 Lightning: DSpark, which is recommended for DGX Spark inference workloads and low concurrency data center workloads.
DSpark: A semi-autoregressive speculative-decoding drafter that proposes a whole block of candidate tokens in a single forward pass from a parallel backbone.
Enable DSpark with --speculative-algorithm DSPARK
DSpark speculative decoding is enabled with a single flag
Filed alongside
Other methods under inference & serving :: decoding strategy.