mirror/LocalAI
1
0
Fork 0
mirror of https://github.com/mudler/LocalAI.git synced 2026-07-13 17:54:02 -04:00
Code Issues Packages Projects Releases Wiki Activity

fix(llama.cpp/mmproj): fix loading mmproj in nested sub-dirs different from model path (#7832)

Browse Source 
fix(mmproj): fix loading mmproj in nested sub-dirs

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
This commit is contained in:
Ettore Di Giacinto
2026-01-02 20:17:30 +01:00
committed by GitHub
parent 1639fc6309
commit 5f6c941399
2 changed files with 10 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
backend/cpp/llama-cpp/grpc-server.cpp
View File

@@ -358,9 +358,7 @@ static void params_parse(server_context& /*ctx_server*/, const backend::ModelOpt
params.model.path = request->modelfile();
if (!request->mmproj().empty()) {
// get the directory of modelfile
std::string model_dir = params.model.path.substr(0, params.model.path.find_last_of("/\\"));
params.mmproj.path = model_dir + "/"+ request->mmproj();
params.mmproj.path = request->mmproj();
}
// params.model_alias ??
params.model_alias = request->modelfile();

13
core/backend/options.go
View File

@@ -36,7 +36,7 @@ func ModelOptions(c config.ModelConfig, so *config.ApplicationConfig, opts ...mo
c.Threads = &threads
grpcOpts := grpcModelOpts(c)
grpcOpts := grpcModelOpts(c, so.SystemState.Model.ModelsPath)
defOpts = append(defOpts, model.WithLoadGRPCLoadModelOpts(grpcOpts))
if so.ParallelBackendRequests {
@@ -72,7 +72,7 @@ func getSeed(c config.ModelConfig) int32 {
return seed
}
func grpcModelOpts(c config.ModelConfig) *pb.ModelOptions {
func grpcModelOpts(c config.ModelConfig, modelPath string) *pb.ModelOptions {
b := 512
if c.Batch != 0 {
b = c.Batch
@@ -131,7 +131,7 @@ func grpcModelOpts(c config.ModelConfig) *pb.ModelOptions {
})
}
return &pb.ModelOptions{
opts := &pb.ModelOptions{
CUDA: c.CUDA || c.Diffusers.CUDA,
SchedulerType: c.Diffusers.SchedulerType,
GrammarTriggers: triggers,
@@ -170,7 +170,6 @@ func grpcModelOpts(c config.ModelConfig) *pb.ModelOptions {
LimitImagePerPrompt: int32(c.LimitMMPerPrompt.LimitImagePerPrompt),
LimitVideoPerPrompt: int32(c.LimitMMPerPrompt.LimitVideoPerPrompt),
LimitAudioPerPrompt: int32(c.LimitMMPerPrompt.LimitAudioPerPrompt),
MMProj: c.MMProj,
FlashAttention: flashAttention,
CacheTypeKey: c.CacheTypeK,
CacheTypeValue: c.CacheTypeV,
@@ -198,6 +197,12 @@ func grpcModelOpts(c config.ModelConfig) *pb.ModelOptions {
// RWKV
Tokenizer: c.Tokenizer,
}
if c.MMProj != "" {
opts.MMProj = filepath.Join(modelPath, c.MMProj)
}
return opts
}
func gRPCPredictOpts(c config.ModelConfig, modelPath string) *pb.PredictOptions {