llama.cpp: Pass server storage paths as parameters

Signed-off-by: Dorin Geman <dorin.geman@docker.com>
This commit is contained in:
Dorin Geman 2025-04-02 17:29:10 +03:00 committed by Dorin-Andrei Geman
parent 184f9c426b
commit 473b9df71e
1 changed files with 16 additions and 30 deletions

View File

@ -20,19 +20,6 @@ const (
Name = "llama.cpp"
)
// VendoredServerStoragePath returns the parent path of the vendored version of
// com.docker.llama-server. It can be overridden during init().
var VendoredServerStoragePath = func() (string, error) {
return ".", nil
}
// UpdatedServerStoragePath returns the parent path of the updated version of
// com.docker.llama-server. It is also where updates will be stored when
// downloaded. It can be overridden during init().
var UpdatedServerStoragePath = func() (string, error) {
return ".", nil
}
// llamaCpp is the llama.cpp-based backend implementation.
type llamaCpp struct {
// log is the associated logger.
@ -42,6 +29,11 @@ type llamaCpp struct {
// serverLog is the logger to use for the llama.cpp server process.
serverLog logging.Logger
updatedLlamaCpp bool
// vendoredServerStoragePath is the parent path of the vendored version of com.docker.llama-server.
vendoredServerStoragePath string
// updatedServerStoragePath is the parent path of the updated version of com.docker.llama-server.
// It is also where updates will be stored when downloaded.
updatedServerStoragePath string
}
// New creates a new llama.cpp-based backend.
@ -49,11 +41,15 @@ func New(
log logging.Logger,
modelManager *models.Manager,
serverLog logging.Logger,
vendoredServerStoragePath string,
updatedServerStoragePath string,
) (inference.Backend, error) {
return &llamaCpp{
log: log,
modelManager: modelManager,
serverLog: serverLog,
log: log,
modelManager: modelManager,
serverLog: serverLog,
vendoredServerStoragePath: vendoredServerStoragePath,
updatedServerStoragePath: updatedServerStoragePath,
}, nil
}
@ -82,11 +78,7 @@ func (l *llamaCpp) Install(ctx context.Context, httpClient *http.Client) error {
// Internet access and an available docker/docker-model-backend-llamacpp:latest-update on Docker Hub are required.
// Even if docker/docker-model-backend-llamacpp:latest-update has been downloaded before, we still require its
// digest to be equal to the one on Docker Hub.
llamaCppStorage, err := UpdatedServerStoragePath()
if err != nil {
return fmt.Errorf("unable to determine llama.cpp path: %w", err)
}
llamaCppPath := filepath.Join(llamaCppStorage, "com.docker.llama-server")
llamaCppPath := filepath.Join(l.updatedServerStoragePath, "com.docker.llama-server")
if err := ensureLatestLlamaCpp(ctx, l.log, httpClient, llamaCppPath); err != nil {
l.log.Infof("failed to ensure latest llama.cpp: %v\n", err)
if errors.Is(err, context.Canceled) {
@ -112,15 +104,9 @@ func (l *llamaCpp) Run(ctx context.Context, socket, model string, mode inference
l.log.Warnln("llama.cpp may not be able to start")
}
binPath, err := UpdatedServerStoragePath()
if err != nil {
return fmt.Errorf("unable to determine llama.cpp path: %w", err)
}
if !l.updatedLlamaCpp {
binPath, err = VendoredServerStoragePath()
if err != nil {
return fmt.Errorf("unable to determine vendored llama.cpp path: %w", err)
}
binPath := l.vendoredServerStoragePath
if l.updatedLlamaCpp {
binPath = l.updatedServerStoragePath
}
llamaCppArgs := []string{"--model", modelPath, "--jinja"}
if mode == inference.BackendModeEmbedding {