char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

YiChen Lvmetal : per-device tuned (Q, NE) for flash-attn vec (#26570)f280b2698

char/aliceai-t5-moe
brancheschar/aliceai-t5-moeall refs →
1015 B45 linesraw
1# dependencies
2
3find_package(Threads REQUIRED)
4
5# third-party
6
7# ...
8
9# flags
10
11llama_add_compile_flags()
12
13# tools
14
15if (EMSCRIPTEN)
16else()
17    add_subdirectory(batched-bench)
18    add_subdirectory(gguf-split)
19    add_subdirectory(imatrix)
20    add_subdirectory(llama-bench)
21    add_subdirectory(completion)
22    add_subdirectory(perplexity)
23    add_subdirectory(quantize)
24    if (LLAMA_BUILD_SERVER)
25        add_subdirectory(ui)
26        add_subdirectory(cli)
27        add_subdirectory(server)
28    endif()
29    add_subdirectory(tokenize)
30    add_subdirectory(tts)
31    add_subdirectory(mtmd)
32    if (GGML_RPC)
33        add_subdirectory(rpc)
34    endif()
35    if (NOT GGML_BACKEND_DL AND GGML_CPU)
36        # these tools use backends directly (no dynamic loading) and depend on CPU backend symbols
37        add_subdirectory(cvector-generator)
38        add_subdirectory(export-lora)
39    endif()
40    add_subdirectory(fit-params)
41    if (GGML_METAL)
42        add_subdirectory(tuning)
43    endif()
44    add_subdirectory(results)
45endif()