为了安全，强烈建议开启2FA双因子认证：User Settings -> Account -> Enable two-factor authentication！！！

Tags

Tags give the ability to mark specific points in history as being important

master-4304bd3

4304bd3c · CUDA: mul_mat_vec_q kernels for k-quants (#2203) · Jul 14, 2023
master-229aab3

229aab35 · make : fix combination of LLAMA_METAL and LLAMA_MPI (#2208) · Jul 14, 2023
master-6979666

69796668 · ggml : sync (ggml_conv_2d, fix mul_mat bug, CUDA GLM rope) · Jul 14, 2023
master-32c5411

32c54116 · Revert "Support using mmap when applying LoRA (#2095)" (#2206) · Jul 13, 2023
master-ff5d58f

ff5d58fa · Fix compile error on Windows CUDA (#2207) · Jul 13, 2023
master-4e7464e

4e7464ef · FP16 is supported in CM=6.0 (#2177) · Jul 12, 2023
master-2b5eb72

2b5eb72e · Fixed __dp4a compute capability: 6.0 -> 6.1 (#2189) · Jul 12, 2023
master-f7d278f

f7d278fa · ggml : revert CUDA broadcast changes from #2183 (#2191) · Jul 12, 2023
master-20d7740

20d7740a · ggml : sync (abort callback, mul / add broadcast, fix alibi) (#2183) · Jul 11, 2023
master-5bf2a27

5bf2a277 · ggml : remove src0 and src1 from ggml_tensor and rename opt to src (#2178) · Jul 11, 2023
master-c9c74b4

c9c74b4e · llama : add classifier-free guidance (#2135) · Jul 11, 2023
master-2347463

23474632 · Support using mmap when applying LoRA (#2095) · Jul 11, 2023
master-bbef282

bbef2821 · Possible solution to allow K-quants on models with n_vocab!=32000 (#2148) · Jul 11, 2023
master-5656d10

5656d105 · mpi : add support for distributed inference via MPI (#2099) · Jul 10, 2023
master-1d16309

1d163099 · llama : remove "first token must be BOS" restriction (#2153) · Jul 09, 2023
master-db4047a

db4047ad · main : escape prompt prefix/suffix (#2151) · Jul 09, 2023
master-3bbc1a1

3bbc1a11 · ggml : fix buidling with Intel MKL but ask for "cblas.h" issue (#2104) (#2115) · Jul 09, 2023
master-6463955

64639555 · Fixed OpenLLaMA 3b CUDA mul_mat_vec_q (#2144) · Jul 08, 2023
ci_cublas_linux-e344540

e3445406 · ci: add linux binaries to release build · Jul 08, 2023
master-061f5f8

061f5f8d · CUDA: add __restrict__ to mul mat vec kernels (#2140) · Jul 08, 2023