github/gpt4all
1
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2025-09-05 18:40:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

expose n_gpu_layers parameter of llama.cpp (#1890)

Browse Source 
Also dynamically limit the GPU layers and context length fields to the maximum supported by the model.

Signed-off-by: Jared Van Bortel <jared@nomic.ai>
This commit is contained in:
Jared Van Bortel
2024-01-31 14:17:44 -05:00
committed by GitHub
parent f549d5a70a
commit 061d1969f8
31 changed files with 381 additions and 157 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
gpt4all-backend/llmodel_c.h
View File

@@ -111,18 +111,20 @@ void llmodel_model_destroy(llmodel_model model);
* @param model A pointer to the llmodel_model instance.
* @param model_path A string representing the path to the model file.
* @param n_ctx Maximum size of context window
* @param ngl Number of GPU layers to use (Vulkan)
* @return size greater than 0 if the model was parsed successfully, 0 if file could not be parsed.
*/
size_t llmodel_required_mem(llmodel_model model, const char *model_path, int n_ctx);
size_t llmodel_required_mem(llmodel_model model, const char *model_path, int n_ctx, int ngl);
/**
* Load a model from a file.
* @param model A pointer to the llmodel_model instance.
* @param model_path A string representing the path to the model file.
* @param n_ctx Maximum size of context window
* @param ngl Number of GPU layers to use (Vulkan)
* @return true if the model was loaded successfully, false otherwise.
*/
bool llmodel_loadModel(llmodel_model model, const char *model_path, int n_ctx);
bool llmodel_loadModel(llmodel_model model, const char *model_path, int n_ctx, int ngl);
/**
* Check if a model is loaded.