poolside
/

Laguna-S-2.1-DFlash

speculative-decoding

Model card Files Files and versions

Resources

View closed (6)

DFlash models have rope_theta=500000 but main model uses default 10000 — causes positional encoding mismatch at >32K context

#8 opened 6 days ago by

llama.cpp fork: --spec-draft-device aborts for DFlash drafts that borrow target tensors

#5 opened 10 days ago by