História revízii

Autor SHA1 Správa Dátum
  Georgi Gerganov b7f2aa9e51 metal : restore 363f0bf and fix reduce in F16_F32 kernels (#2986) 2 rokov pred
  Georgi Gerganov d9151e6f57 metal : revert 6af0bab until we fix it 2 rokov pred
  Kawrakow ca82cf7bac metal : more optimizations (#2959) 2 rokov pred
  Georgi Gerganov 13268c5331 metal : slight speed-up for add and mul kernels (#2917) 2 rokov pred
  Kawrakow e8d9158925 metal: somewhat faster f16 x f32 matrix multiply kernel (#2951) 2 rokov pred
  Georgi Gerganov d67777c202 metal : add Q8_0 support (#2763) 2 rokov pred
  Georgi Gerganov cf658adc83 llm : add Falcon support (#2717) 2 rokov pred
  Shouzheng Liu 14b1d7e6f7 metal : add missing barriers for mul-mat (#2699) 2 rokov pred
  Shouzheng Liu dadbed99e6 metal : fix synchronization in new matrix multiplication kernel (#2686) 2 rokov pred
  Shouzheng Liu bf83bff674 metal : matrix-matrix multiplication kernel (#2615) 2 rokov pred
  Matteo Boschini 1873ff586b metal : add gqa8 kernel to allow llama-2-70B on metal (#2459) 2 rokov pred
  Kawrakow 9a08eaf3c4 Another speed gain for Q4_0 and Q4_1 on Metal (#2375) 2 rokov pred
  Jiahao Li 83a00ce69b metal : support bcast add & dup & cont op (#2323) 2 rokov pred
  Kawrakow 4d76a5f49b Faster Q3_K implementation on Metal (#2307) 2 rokov pred
  Kawrakow e68c96f7fe Faster Q2_K on Metal (#2297) 2 rokov pred
  Kawrakow e782c9e735 Faster Q5_K and Q6_K on Metal (#2294) 2 rokov pred
  Kawrakow 785829dfe8 Faster Q4_K on Metal (#2290) 2 rokov pred
  Shouzheng Liu 417a85a001 metal: minor q4 optimization and reduce code size (#2248) 2 rokov pred
  Xiao-Yong Jin 6e7cca4047 llama : add custom RoPE (#2054) 2 rokov pred
  Kawrakow 27ad57a69b Metal: faster Q4_0 and Q4_1 matrix x vector kernels (#2212) 2 rokov pred
  Shouzheng Liu 1cbf561466 metal : new q4_0 matrix-vector kernel (#2188) 2 rokov pred
  Kawrakow 6769e944c7 k-quants : support for super-block size of 64 (#2001) 2 rokov pred
  Aaron Miller 0711a5f6dc metal : add norm, cpy f16->f16, alibi kernels (#1823) 2 rokov pred
  Kawrakow 74a6d922f1 Metal implementation for all k_quants (#1807) 2 rokov pred
  Kawrakow e9b66ee982 metal : add Q4_1 implementation (#1785) 2 rokov pred
  Georgi Gerganov b33dee282f metal : fix build "tanhf" -> "tanh" 2 rokov pred
  AT 92f44ff7f7 metal : add GELU implementation (#1770) 2 rokov pred
  Kawrakow 245fc3c37d metal : faster q4_0 (#1775) 2 rokov pred
  Kawrakow 72ff5282bf metal : add Q2_K implementation (#1762) 2 rokov pred
  Kawrakow 0f291e1f65 metal : Q6_K implementation (#1752) 2 rokov pred