Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 05:48:47 +01:00
Code Issues Packages Projects Releases Wiki Activity

llama : fix Gemma rope type (#5691)

Browse Source
This commit is contained in:
Georgi Gerganov 2024-02-26 08:30:17 +02:00
parent c393733988
commit 269de86ba0
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llama.cpp
View File

@ -12089,7 +12089,6 @@ enum llama_rope_type llama_rope_type(const struct llama_model * model) {
case LLM_ARCH_ORION: case LLM_ARCH_ORION:
case LLM_ARCH_INTERNLM2: case LLM_ARCH_INTERNLM2:
case LLM_ARCH_MINICPM: case LLM_ARCH_MINICPM:
case LLM_ARCH_GEMMA:
return LLAMA_ROPE_TYPE_NORM; return LLAMA_ROPE_TYPE_NORM;
// the pairs of head values are offset by n_rot/2 // the pairs of head values are offset by n_rot/2
@ -12101,6 +12100,7 @@ enum llama_rope_type llama_rope_type(const struct llama_model * model) {
case LLM_ARCH_QWEN: case LLM_ARCH_QWEN:
case LLM_ARCH_QWEN2: case LLM_ARCH_QWEN2:
case LLM_ARCH_PHI2: case LLM_ARCH_PHI2:
case LLM_ARCH_GEMMA:
return LLAMA_ROPE_TYPE_NEOX; return LLAMA_ROPE_TYPE_NEOX;
// all model arches should be listed explicitly here // all model arches should be listed explicitly here