cfontes
/

qwen-dflash2-spark

Text Generation

speculative-decoding

Model card Files Files and versions

qwen-dflash2-spark / docs

Commit History

Add prefill benchmark (~3.4k tok/s peak) + GH canonical link

67da175
verified

cfontes commited on 3 days ago

Qwen3.8-27B DFlash2 on 2x DGX Spark: one-stop stack (135 tok/s C1, lm_head BF16 fix)

5be2256
verified

cfontes commited on 4 days ago