⬆️ Update ggerganov/llama.cpp

[localai-bot]

Bump of ggerganov/llama.cpp version

[netlify]

✅ Deploy Preview for localai canceled.

Name	Link
🔨 Latest commit	7378af0
🔍 Latest deploy log	https://app.netlify.com/sites/localai/deploys/65d8fac2dc23b500083a1423

[olariuromeo]

Building images with this change get error, instead, the old version works perfectly. With the current version it is not possible to prepare the backend or generate images:

make clean
make GO_TAGS=stablediffusion,tts BUILD_TYPE=cuBLAS BUILD_GRPC_FOR_BACKEND_LLAMA=true build
mkdir -p backend-assets
LLAMA_VERSION=201294ae177b308fb3a99dc504dd6d27e8afa907 make -C backend/cpp/llama llama.cpp	
make[1]: Entering directory '/home/user/apps/localai/backend/cpp/llama'
git clone --recurse-submodules https://github.com/ggerganov/llama.cpp llama.cpp
Cloning into 'llama.cpp'...
remote: Enumerating objects: 19344, done.
remote: Counting objects: 100% (3/3), done.
remote: Compressing objects: 100% (3/3), done.
remote: Total 19344 (delta 0), reused 1 (delta 0), pack-reused 19341
Receiving objects: 100% (19344/19344), 22.71 MiB | 11.65 MiB/s, done.
Resolving deltas: 100% (13524/13524), done.
Submodule 'kompute' (https://github.com/nomic-ai/kompute.git) registered for path 'kompute'
Cloning into '/home/user/apps/localai/backend/cpp/llama/llama.cpp/kompute'...
remote: Enumerating objects: 9078, done.        
remote: Counting objects: 100% (213/213), done.        
remote: Compressing objects: 100% (125/125), done.        
remote: Total 9078 (delta 94), reused 167 (delta 78), pack-reused 8865        
Receiving objects: 100% (9078/9078), 17.58 MiB | 10.31 MiB/s, done.
Resolving deltas: 100% (5701/5701), done.
remote: Enumerating objects: 1, done.
remote: Counting objects: 100% (1/1), done.
remote: Total 1 (delta 0), reused 1 (delta 0), pack-reused 0
Unpacking objects: 100% (1/1), 213 bytes | 213.00 KiB/s, done.
From https://github.com/nomic-ai/kompute
 * branch            4565194ed7c32d1d2efa32ceab4d3c6cae006306 -> FETCH_HEAD
Submodule path 'kompute': checked out '4565194ed7c32d1d2efa32ceab4d3c6cae006306'
if [ -z "201294ae177b308fb3a99dc504dd6d27e8afa907" ]; then \
	exit 1; \
fi
cd llama.cpp && git checkout -b build 201294ae177b308fb3a99dc504dd6d27e8afa907 && git submodule update --init --recursive --depth 1
Switched to a new branch 'build'
make[1]: Leaving directory '/home/user/apps/localai/backend/cpp/llama'
git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp sources/go-llama
Cloning into 'sources/go-llama'...
remote: Enumerating objects: 769, done.
remote: Counting objects: 100% (412/412), done.
remote: Compressing objects: 100% (166/166), done.
remote: Total 769 (delta 287), reused 317 (delta 234), pack-reused 357
Receiving objects: 100% (769/769), 237.45 KiB | 2.05 MiB/s, done.
Resolving deltas: 100% (445/445), done.
Submodule 'llama.cpp' (https://github.com/ggerganov/llama.cpp) registered for path 'llama.cpp'
Cloning into '/home/user/apps/localai/sources/go-llama/llama.cpp'...
remote: Enumerating objects: 19344, done.        
remote: Counting objects: 100% (3/3), done.        
remote: Compressing objects: 100% (3/3), done.        
remote: Total 19344 (delta 0), reused 1 (delta 0), pack-reused 19341        
Receiving objects: 100% (19344/19344), 22.71 MiB | 10.49 MiB/s, done.
Resolving deltas: 100% (13524/13524), done.
Submodule path 'llama.cpp': checked out 'ac43576124a75c2de6e333ac31a3444ff9eb9458'
cd sources/go-llama && git checkout -b build aeba71ee842819da681ea537e78846dc75949ac0 && git submodule update --init --recursive --depth 1
Switched to a new branch 'build'
git clone --recurse-submodules https://github.com/go-skynet/go-llama.cpp sources/go-llama-ggml
Cloning into 'sources/go-llama-ggml'...
remote: Enumerating objects: 769, done.
remote: Counting objects: 100% (437/437), done.
remote: Compressing objects: 100% (172/172), done.
remote: Total 769 (delta 311), reused 336 (delta 253), pack-reused 332
Receiving objects: 100% (769/769), 234.34 KiB | 1.71 MiB/s, done.
Resolving deltas: 100% (448/448), done.
Submodule 'llama.cpp' (https://github.com/ggerganov/llama.cpp) registered for path 'llama.cpp'
Cloning into '/home/user/apps/localai/sources/go-llama-ggml/llama.cpp'...
remote: Enumerating objects: 19344, done.        
remote: Counting objects: 100% (3/3), done.        
remote: Compressing objects: 100% (3/3), done.        
remote: Total 19344 (delta 0), reused 1 (delta 0), pack-reused 19341        
Receiving objects: 100% (19344/19344), 22.49 MiB | 11.36 MiB/s, done.
Resolving deltas: 100% (13526/13526), done.
Submodule path 'llama.cpp': checked out 'ac43576124a75c2de6e333ac31a3444ff9eb9458'
cd sources/go-llama-ggml && git checkout -b build fd43d66f46ee3b5345fb8a74a252d86ccd34a409 && git submodule update --init --recursive --depth 1
fatal: reference is not a tree: fd43d66f46ee3b5345fb8a74a252d86ccd34a409
make: *** [Makefile:241: sources/go-llama-ggml] Error 128