]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
llama : fix llm_build_k_shift to use correct n_rot (#4889)
authorGeorgi Gerganov <redacted>
Fri, 12 Jan 2024 11:01:56 +0000 (13:01 +0200)
committerGitHub <redacted>
Fri, 12 Jan 2024 11:01:56 +0000 (13:01 +0200)
commitf445c0e68cf8e1faca0b2aa8dfb9d48231cec301
treea230e263424ba71153764e21fef7b92f46bd9fd9
parent326b418b59b6d48d854c4461a2303e8ac0a311e6
llama : fix llm_build_k_shift to use correct n_rot (#4889)

* llama : fix llm_build_k_shift to use correct n_rot

ggml-ci

* llama : always use hparams.n_rot for ggml_rope_custom

ggml-ci

* convert : fix persimmon conversion to write correct n_rot
common/common.cpp
convert-hf-to-gguf.py
gguf-py/gguf/tensor_mapping.py
llama.cpp