aLoRA Support#15327

Merged

gabe-l-hart merged 19 commits intoggml-org:masterfrom

gabe-l-hart:gabe-l-hart/alora-support

Sep 5, 2025

Commits on Aug 22, 2025

feat: Add python-side constants and conversion for adapter.lora.invocation_string
gabe-l-hart
committed
feat: Add c++ side constants for adapter.lora.invocation_string
gabe-l-hart
committed
feat: Parse invocation string for adapters from GGUF
gabe-l-hart
committed
fix(python): Update conversion to alora_invocation_tokens
gabe-l-hart
committed
fix(cpp): Update to alora_invocation_tokens on c++ side
gabe-l-hart
committed
feat: Add C APIs to get alora invocation token array from lora
gabe-l-hart
committed
feat: Initial implementation of alora cache logic in server
gabe-l-hart
committed
feat: Identify alora invocation sequences
gabe-l-hart
committed
feat: Only reuse cache for tokens before the alora invocation start
gabe-l-hart
committed
feat: Handle un-cached tokens that come before the alora activation
gabe-l-hart
committed
fix: Use || instead of 'or'
gabe-l-hart
committed
fix: Fix off-by-one for limiting cached tokens to before alora start
gabe-l-hart
committed
fix: Support backwards-compatibility for "invocation_string" in adapter_config.json
gabe-l-hart
committed

Commits on Aug 28, 2025

Commits on Sep 4, 2025

Merge remote-tracking branch 'origin/master' into gabe-l-hart/alora-support
gabe-l-hart
committed

Commits on Sep 5, 2025