llama.cpp
Support for Phi-2
#4490
Merged
Overview
Commits
17
Changes
View On GitHub
Commits
  • phi2 implementation
    eabraham-1 committed 2 years ago
  • fix breaking change
    eabraham-1 committed 2 years ago
  • phi-2 : various fixes
    ggerganov committed 2 years ago
  • phi-2 : use layer norm eps
    ggerganov committed 2 years ago
  • py : whitespaces
    ggerganov committed 2 years ago
  • llama : fix meta KV override bug
    ggerganov committed 2 years ago
  • convert : phi don't add BOS token
    ggerganov committed 2 years ago
  • convert : revert "added_tokens_decoder" change
    ggerganov committed 2 years ago
  • phi-2 : scale Q instead of KQ for better precision
    ggerganov committed 2 years ago
  • ggml : fix NeoX rope to rotate just first n_dims
    ggerganov committed 2 years ago
  • cuda : less diff in the rope_neox kernel
    ggerganov committed 2 years ago
  • Merge branch 'master' into HEAD
    ggerganov committed 2 years ago
  • ggml : add ggml_mul_mat_set_prec
    ggerganov committed 2 years ago
  • Update ggml-cuda.cu
    ggerganov committed 2 years ago
  • Update ggml-cuda.cu
    ggerganov committed 2 years ago
  • cuda : ggml_cuda_op_mul_mat_cublas support F32 precision
    ggerganov committed 2 years ago
  • cuda : remove oboslete comment
    ggerganov committed 2 years ago
FAQ Terms Privacy Refunds Impressum
Loading