llama.cpp
Unified delta net handling for Qwen3Next and Kimi Linear models
#18792
Open

Unified delta net handling for Qwen3Next and Kimi Linear models #18792

pwilkin wants to merge 5 commits into ggml-org:master from pwilkin:delta_net
pwilkin
pwilkin Unified delta net handling
b3f55ead
pwilkin pwilkin requested a review from CISC CISC 19 hours ago
pwilkin pwilkin requested a review from ggerganov ggerganov 19 hours ago
pwilkin pwilkin removed review request from CISC CISC 19 hours ago
pwilkin pwilkin removed review request from ggerganov ggerganov 19 hours ago
pwilkin pwilkin requested a review from ngxson ngxson 19 hours ago
pwilkin pwilkin requested a review from CISC CISC 19 hours ago
pwilkin pwilkin requested a review from ggerganov ggerganov 19 hours ago
pwilkin Remove old methods.
34e1ed90
ngxson
ngxson commented on 2026-01-12
github-actions github-actions added model
ymcki
pwilkin Refactor and optimize
f98f2856
pwilkin Adapt autoregressive version from @ymcki
08b1ed86
pwilkin
ymcki
pwilkin Change to decay mask approach
e9ad1849
pwilkin

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone