Package Details: llama.cpp-git b2698-1

Git Clone URL: https://aur.archlinux.org/llama.cpp-git.git (read-only, click to copy)
Package Base: llama.cpp-git
Description: Port of Facebook's LLaMA model in C/C++ (with OPENBlas CPU optimizations)
Upstream URL: https://github.com/ggerganov/llama.cpp
Licenses: MIT
Conflicts: llama.cpp
Provides: llama.cpp
Submitter: robertfoster
Maintainer: robertfoster
Last Packager: robertfoster
Votes: 6
Popularity: 1.33
First Submitted: 2023-03-27 22:24 (UTC)
Last Updated: 2024-04-19 16:24 (UTC)

Latest Comments

1 2 Next › Last »

dreieck commented on 2024-04-17 11:18 (UTC)

ccache should not be an optional dependency.

It is only relevant for building.

And the user can specify the usage of it by setting the ccache-option in e.g. /etc/makepkg.conf.

Please remove it from optdepends.

Regards!

lapsus commented on 2024-04-16 22:20 (UTC)

[ 19%] Built target common
make: *** [Makefile:146: all] Error 2
==> ERROR: A failure occurred in build().
Aborting...
error: failed to build 'llama.cpp-git-b2684-1 (llama.cpp-cublas-git)':
error: packages failed to build: llama.cpp-git-b2684-1 (llama.cpp-cublas-git)

lapsus commented on 2024-04-15 22:49 (UTC)

/home/user/.cache/paru/clone/llama.cpp-git/PKGBUILD: line 137: cd: /home/user/.cache/paru/clone/llama.cpp-git/src/llama.cpp-sycl: No such file or directory
==> ERROR: A failure occurred in build().
Aborting...
error: failed to build 'llama.cpp-git-b2646.r7.f7001ccc5-1 (llama.cpp-cublas-git)':
error: packages failed to build: llama.cpp-git-b2646.r7.f7001ccc5-1 (llama.cpp-cublas-git)

<deleted-account> commented on 2024-04-10 05:23 (UTC)

I believe it's LLAMA_CUDA now instead of LLAMA_CUBLAS.

Just an FYI.

danisztls commented on 2024-04-04 02:11 (UTC) (edited on 2024-04-04 02:11 (UTC) by danisztls)

/bin/ld: ../common/libcommon.a: error adding symbols: archive has no index; run ranlib to add one
icpx: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [tests/CMakeFiles/test-quantize-fns.dir/build.make:117: bin/test-quantize-fns] Error 1
make[1]: *** [CMakeFiles/Makefile2:1728: tests/CMakeFiles/test-quantize-fns.dir/all] Error 2
make: *** [Makefile:146: all] Error 2

dreieck commented on 2024-03-27 09:36 (UTC)

Since several packages use ggml I made an extra package for it:
libggml-git.

Regards!

dreieck commented on 2024-03-26 10:51 (UTC)

llama.cpp and whisper.cpp need to conflict with each other:

error: failed to commit transaction (conflicting files)
llama.cpp-hipblas-git: /usr/include/ggml.h exists in filesystem (owned by whisper.cpp)

(Maybe better to split out /usr/include/ggml.h in a separate package and depend on it.)

Regards and thanks for maintaining!

dreieck commented on 2024-03-26 10:48 (UTC)

According to namcap, you need to add missing dependencies:
llama.cpp-hipblas-git misses rocblas and hipblas:

llama.cpp-hipblas-git E: Dependency rocblas detected and not included (libraries ['opt/rocm/lib/librocblas.so.4'] needed in files ['usr/bin/llama.cpp-infill', 'usr/bin/llama.cpp-server', 'usr/bin/llama.cpp-test-tokenizer-1-starcoder', 'usr/bin/llama.cpp-test-tokenizer-1-falcon', 'usr/bin/llama.cpp-train-text-from-scratch', 'usr/bin/llama.cpp-save-load-state', 'usr/bin/llama.cpp-llava-cli', 'usr/bin/llama.cpp-test-tokenizer-1-gpt-neox', 'usr/bin/llama.cpp-quantize-stats', 'usr/bin/llama.cpp-perplexity', 'usr/bin/llama.cpp-test-backend-ops', 'usr/bin/llama.cpp-tokenize', 'usr/bin/llama.cpp-batched', 'usr/bin/llama.cpp-batched-bench', 'usr/bin/llama.cpp', 'usr/bin/llama.cpp-embedding', 'usr/bin/llama.cpp-llama-bench', 'usr/bin/llama.cpp-lookahead', 'usr/bin/llama.cpp-beam-search', 'usr/bin/llama.cpp-test-tokenizer-0-llama', 'usr/bin/llama.cpp-speculative', 'usr/bin/llama.cpp-parallel', 'usr/bin/llama.cpp-test-tokenizer-1-llama', 'usr/bin/llama.cpp-test-tokenizer-0-falcon', 'usr/bin/llama.cpp-test-tokenizer-1-baichuan', 'usr/bin/llama.cpp-test-tokenizer-1-gpt2', 'usr/bin/llama.cpp-retrieval', 'usr/bin/llama.cpp-test-autorelease', 'usr/bin/llama.cpp-test-model-load-cancel', 'usr/bin/llama.cpp-test-tokenizer-1-mpt', 'usr/bin/llama.cpp-gritlm', 'usr/bin/llama.cpp-test-tokenizer-1-refact', 'usr/bin/llama.cpp-lookup', 'usr/bin/llama.cpp-lookup-create', 'usr/bin/llama.cpp-test-tokenizer-1-aquila', 'usr/bin/llama.cpp-test-tokenizer-1-stablelm-3b-4e1t', 'usr/bin/llama.cpp-passkey', 'usr/bin/llama.cpp-lookup-stats', 'usr/bin/llama.cpp-simple', 'usr/bin/llama.cpp-finetune', 'usr/bin/llama.cpp-imatrix'])
llama.cpp-hipblas-git E: Dependency hipblas detected and not included (libraries ['opt/rocm/lib/libhipblas.so.2'] needed in files ['usr/bin/llama.cpp-infill', 'usr/bin/llama.cpp-server', 'usr/bin/llama.cpp-test-tokenizer-1-starcoder', 'usr/bin/llama.cpp-test-tokenizer-1-falcon', 'usr/bin/llama.cpp-train-text-from-scratch', 'usr/bin/llama.cpp-save-load-state', 'usr/bin/llama.cpp-llava-cli', 'usr/bin/llama.cpp-test-tokenizer-1-gpt-neox', 'usr/bin/llama.cpp-quantize-stats', 'usr/bin/llama.cpp-perplexity', 'usr/bin/llama.cpp-test-backend-ops', 'usr/bin/llama.cpp-tokenize', 'usr/bin/llama.cpp-batched', 'usr/bin/llama.cpp-batched-bench', 'usr/bin/llama.cpp', 'usr/bin/llama.cpp-embedding', 'usr/bin/llama.cpp-llama-bench', 'usr/bin/llama.cpp-lookahead', 'usr/bin/llama.cpp-beam-search', 'usr/bin/llama.cpp-test-tokenizer-0-llama', 'usr/bin/llama.cpp-speculative', 'usr/bin/llama.cpp-parallel', 'usr/bin/llama.cpp-test-tokenizer-1-llama', 'usr/bin/llama.cpp-test-tokenizer-0-falcon', 'usr/bin/llama.cpp-test-tokenizer-1-baichuan', 'usr/bin/llama.cpp-test-tokenizer-1-gpt2', 'usr/bin/llama.cpp-retrieval', 'usr/bin/llama.cpp-test-autorelease', 'usr/bin/llama.cpp-test-tokenizer-1-mpt', 'usr/bin/llama.cpp-gritlm', 'usr/bin/llama.cpp-test-tokenizer-1-refact', 'usr/bin/llama.cpp-lookup', 'usr/bin/llama.cpp-lookup-create', 'usr/bin/llama.cpp-test-tokenizer-1-aquila', 'usr/bin/llama.cpp-test-tokenizer-1-stablelm-3b-4e1t', 'usr/bin/llama.cpp-passkey', 'usr/bin/llama.cpp-lookup-stats', 'usr/bin/llama.cpp-simple', 'usr/bin/llama.cpp-finetune', 'usr/bin/llama.cpp-imatrix'])

Regards and thanks for maintaining!

dreieck commented on 2024-03-26 10:46 (UTC)

You need to install the license file into /usr/share/licenses/${pkgname}/:

llama.cpp-hipblas-git E: Uncommon license identifiers such as 'MIT' require license files below /usr/share/licenses/llama.cpp-hipblas-git/ or switching to common license identifiers. Found 0/1 required license files.

Regards and thanks for maintaining!

robertfoster commented on 2024-03-26 09:45 (UTC)

@sr.team in a splitted package, makedepends (not depends), are declared for the only one build() function. on the contrary package() functions declare run-time depends. So, nothing to fix.