chore: switch back llama.cpp to b3027 to avoid deepseekcoder 6.7 regr… · TabbyML/tabby@76d27cd · GitHub

Commit

chore: switch back llama.cpp to b3027 to avoid deepseekcoder 6.7 regr…

Browse files

…ession (#2323)

Loading branch information

wsxiaoys committed Jun 3, 2024

1 parent 63cfd74 commit 76d27cd

crates/llama-cpp-server/llama.cpp

Submodule llama.cpp updated 21 files

+0 −4		CMakeLists.txt
+5 −5		README-sycl.md
+1 −2		README.md
+2 −26		examples/llama-bench/llama-bench.cpp
+1 −3		ggml-cuda.cu
+22 −88		ggml-cuda/concat.cu
+0 −6		ggml-cuda/norm.cu
+10 −8		ggml-cuda/rope.cu
+1 −0		ggml-kompute.cpp
+1 −4		ggml-metal.m
+10 −6		ggml-metal.metal
+68 −56		ggml-sycl.cpp
+48 −69		ggml.c
+1 −14		ggml.h
+3 −1		ggml_vk_generate_shaders.py
+174 −100		llama.cpp
+4 −0		scripts/sync-ggml-am.sh
+1 −1		scripts/sync-ggml.last
+2 −0		scripts/sync-ggml.sh
+32 −78		tests/test-backend-ops.cpp
+7 −13		tests/test-tokenizer-random.py

0 comments on commit `76d27cd`

Please sign in to comment.