Merge remote-tracking branch 'origin/fedora-headless-WIP' into fedora-headless-WIP

.github/workflows/test.yml

@@ -71,7 +71,7 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install build-essential ccache upx-ucl curl ffmpeg
-          sudo apt-get install -y libgmock-dev
+          sudo apt-get install -y libgmock-dev clang
           curl https://repo.anaconda.com/pkgs/misc/gpgkeys/anaconda.asc | gpg --dearmor > conda.gpg && \
              sudo install -o root -g root -m 644 conda.gpg /usr/share/keyrings/conda-archive-keyring.gpg && \
              gpg --keyring /usr/share/keyrings/conda-archive-keyring.gpg --no-default-keyring --fingerprint 34161F5BF5EB1D4BFBBB8F0A8AEB4F8B29D82806 && \

Makefile

@@ -6,7 +6,7 @@ BINARY_NAME=local-ai
 DETECT_LIBS?=true
 
 # llama.cpp versions
-CPPLLAMA_VERSION?=ecda2ec4b347031a9b8a89ee2efc664ce63f599c
+CPPLLAMA_VERSION?=295354ea6848a77bdee204ee1c971d9b92ffcca9
 
 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggerganov/whisper.cpp

backend/python/bark/requirements.txt

@@ -1,4 +1,4 @@
 bark==0.1.5
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi

backend/python/common/template/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 grpcio-tools

backend/python/coqui/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi
 packaging==24.1

backend/python/diffusers/requirements.txt

@@ -1,5 +1,5 @@
 setuptools
-grpcio==1.71.0
+grpcio==1.72.0
 pillow
 protobuf
 certifi

backend/python/exllama2/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi
 wheel

backend/python/faster-whisper/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 grpcio-tools

backend/python/kokoro/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 phonemizer
 scipy

backend/python/rerankers/requirements.txt

@@ -1,3 +1,3 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi

backend/python/transformers/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi
 setuptools

backend/python/vllm/requirements.txt

@@ -1,4 +1,4 @@
-grpcio==1.71.0
+grpcio==1.72.0
 protobuf
 certifi
 setuptools

core/config/guesser.go

@@ -4,6 +4,7 @@ import (
 	"os"
 	"path/filepath"
 
+	"github.com/mudler/LocalAI/pkg/xsysinfo"
 	"github.com/rs/zerolog/log"
 	gguf "github.com/thxcode/gguf-parser-go"
 )
@@ -35,4 +36,10 @@ func guessDefaultsFromFile(cfg *BackendConfig, modelPath string, defaultCtx int)
 		}
 		cfg.ContextSize = &defaultCtx
 	}
+
+	if cfg.Options == nil {
+		if xsysinfo.HasGPU("nvidia") || xsysinfo.HasGPU("amd") {
+			cfg.Options = []string{"gpu"}
+		}
+	}
 }

gallery/index.yaml

@@ -556,6 +556,21 @@
     - filename: amoral-gemma3-1B-v2.Q4_K_M.gguf
       sha256: 7f2167d91409cabaf0a42e41e833a6ca055c841a37d8d829e11db81fdaed5e4c
       uri: huggingface://mradermacher/amoral-gemma3-1B-v2-GGUF/amoral-gemma3-1B-v2.Q4_K_M.gguf
+- !!merge <<: *gemma3
+  name: "soob3123_veritas-12b"
+  icon: https://cdn-uploads.huggingface.co/production/uploads/62f93f9477b722f1866398c2/IuhCq-5PcEbDBqXD5xnup.png
+  urls:
+    - https://huggingface.co/soob3123/Veritas-12B
+    - https://huggingface.co/bartowski/soob3123_Veritas-12B-GGUF
+  description: |
+    Veritas-12B emerges as a model forged in the pursuit of intellectual clarity and logical rigor. This 12B parameter model possesses superior philosophical reasoning capabilities and analytical depth, ideal for exploring complex ethical dilemmas, deconstructing arguments, and engaging in structured philosophical dialogue. Veritas-12B excels at articulating nuanced positions, identifying logical fallacies, and constructing coherent arguments grounded in reason. Expect discussions characterized by intellectual honesty, critical analysis, and a commitment to exploring ideas with precision.
+  overrides:
+    parameters:
+      model: soob3123_Veritas-12B-Q4_K_M.gguf
+  files:
+    - filename: soob3123_Veritas-12B-Q4_K_M.gguf
+      sha256: 41821d6b0dd2b81a5bddd843a5534fd64d95e75b8e9dc952340868af320d49a7
+      uri: huggingface://bartowski/soob3123_Veritas-12B-GGUF/soob3123_Veritas-12B-Q4_K_M.gguf
 - &llama4
   url: "github:mudler/LocalAI/gallery/llama3.1-instruct.yaml@master"
   icon: https://avatars.githubusercontent.com/u/153379578

pkg/model/initializers.go

@@ -181,10 +181,6 @@ func orderBackends(backends map[string][]string) ([]string, error) {
 // selectGRPCProcessByHostCapabilities selects the GRPC process to start based on system capabilities
 // Note: this is now relevant only for llama.cpp
 func selectGRPCProcessByHostCapabilities(backend, assetDir string, f16 bool) string {
-	foundCUDA := false
-	foundAMDGPU := false
-	foundIntelGPU := false
-	var grpcProcess string
 
 	// Select backend now just for llama.cpp
 	if backend != LLamaCPP {
@@ -198,48 +194,24 @@ func selectGRPCProcessByHostCapabilities(backend, assetDir string, f16 bool) str
 	}
 
 	// Check for GPU-binaries that are shipped with single binary releases
-	gpus, err := xsysinfo.GPUs()
+	gpuBinaries := map[string]string{
-	if err == nil {
+		"nvidia": LLamaCPPCUDA,
-		for _, gpu := range gpus {
+		"amd":    LLamaCPPHipblas,
-			if strings.Contains(gpu.String(), "nvidia") {
+		"intel":  LLamaCPPSycl16,
-				p := backendPath(assetDir, LLamaCPPCUDA)
-				if _, err := os.Stat(p); err == nil {
-					log.Info().Msgf("[%s] attempting to load with CUDA variant", backend)
-					grpcProcess = p
-					foundCUDA = true
-				} else {
-					log.Debug().Msgf("Nvidia GPU device found, no embedded CUDA variant found. You can ignore this message if you are using container with CUDA support")
-				}
-			}
-			if strings.Contains(gpu.String(), "amd") {
-				p := backendPath(assetDir, LLamaCPPHipblas)
-				if _, err := os.Stat(p); err == nil {
-					log.Info().Msgf("[%s] attempting to load with HIPBLAS variant", backend)
-					grpcProcess = p
-					foundAMDGPU = true
-				} else {
-					log.Debug().Msgf("AMD GPU device found, no embedded HIPBLAS variant found. You can ignore this message if you are using container with HIPBLAS support")
-				}
-			}
-			if strings.Contains(gpu.String(), "intel") {
-				backend := LLamaCPPSycl16
-				if !f16 {
-					backend = LLamaCPPSycl32
-				}
-				p := backendPath(assetDir, backend)
-				if _, err := os.Stat(p); err == nil {
-					log.Info().Msgf("[%s] attempting to load with Intel variant", backend)
-					grpcProcess = p
-					foundIntelGPU = true
-				} else {
-					log.Debug().Msgf("Intel GPU device found, no embedded SYCL variant found. You can ignore this message if you are using container with SYCL support")
-				}
-			}
-		}
 	}
 
-	if foundCUDA || foundAMDGPU || foundIntelGPU {
+	if !f16 {
-		return grpcProcess
+		gpuBinaries["intel"] = LLamaCPPSycl32
+	}
+
+	for vendor, binary := range gpuBinaries {
+		if xsysinfo.HasGPU(vendor) {
+			p := backendPath(assetDir, binary)
+			if _, err := os.Stat(p); err == nil {
+				log.Info().Msgf("[%s] attempting to load with %s variant (vendor: %s)", backend, binary, vendor)
+				return p
+			}
+		}
 	}
 
 	// No GPU found or no specific binaries found, try to load the CPU variant(s)

pkg/xsysinfo/gpu.go

@@ -1,6 +1,8 @@
 package xsysinfo
 
 import (
+	"strings"
+
 	"github.com/jaypipes/ghw"
 	"github.com/jaypipes/ghw/pkg/gpu"
 )
@@ -13,3 +15,19 @@ func GPUs() ([]*gpu.GraphicsCard, error) {
 
 	return gpu.GraphicsCards, nil
 }
+
+func HasGPU(vendor string) bool {
+	gpus, err := GPUs()
+	if err != nil {
+		return false
+	}
+	if vendor == "" {
+		return len(gpus) > 0
+	}
+	for _, gpu := range gpus {
+		if strings.Contains(gpu.String(), vendor) {
+			return true
+		}
+	}
+	return false
+}