git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit

]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit

overview / pkg / ggml / sources / llama.cpp / commit

author	Oliver Simons <redacted>
	Tue, 4 Aug 2026 18:28:55 +0000 (20:28 +0200)
committer	GitHub <redacted>
	Tue, 4 Aug 2026 18:28:55 +0000 (21:28 +0300)
commit	a6aa6f5450eaad18b3c86631b5c3fff330f5a46e
tree	57ce14eb02b3b6d3aeb399383d1db13a6ac059c9	tree
parent	76c956c1372962842e1134f86b7c5d4517597e7e	commit \| diff

sampler : remove "full-context windows" from history-based samplers (#26524)

* Resolve -1 to 1024 instead of ctx-len for samplers

Because of backend-sampling we initialize samplers before the complete
llama_context is there. Therefore, we cannot infer the resolved context
length yet at the time we construct the samplers.

* Shared default of 64 for history-based samplers, remove context_size

18 files changed:

common/arg.cpp		diff \| blob \| history
common/common.cpp		diff \| blob \| history
common/common.h		diff \| blob \| history
common/sampling.cpp		diff \| blob \| history
common/sampling.h		diff \| blob \| history
include/llama.h		diff \| blob \| history
src/llama-sampler.cpp		diff \| blob \| history
src/llama-sampler.h		diff \| blob \| history
tests/test-arg-parser.cpp		diff \| blob \| history
tests/test-sampling.cpp		diff \| blob \| history
tools/cli/README.md		diff \| blob \| history
tools/completion/README.md		diff \| blob \| history
tools/mtmd/tests/test-deepseek-ocr.py		diff \| blob \| history
tools/server/README.md		diff \| blob \| history
tools/server/server-context.cpp		diff \| blob \| history
tools/server/server-schema.cpp		diff \| blob \| history
tools/server/server-schema.h		diff \| blob \| history
tools/ui/src/lib/services/parameter-sync.service.spec.ts		diff \| blob \| history

Packaging of ggml-org/llama.cpp

RSS Atom