llama.cpp
finetune: Add -ngl parameter
#3762
Merged
Overview
Commits
16
Changes
View On GitHub
Commits
  • Add '-ngl' support to finetune.cpp
    AndrewGodfrey committed 2 years ago
  • Add fprintf in ggml_cuda_op_add
    AndrewGodfrey committed 2 years ago
  • Add 'finetune.sh', which currently fails when using GPU
    AndrewGodfrey committed 2 years ago
  • tweak finetune.sh
    AndrewGodfrey committed 2 years ago
  • Suppress some warnings in ggml.c
    AndrewGodfrey committed 2 years ago
  • Add f16 implementation to ggml_compute_forward_add_f16_f32
    AndrewGodfrey committed 2 years ago
  • Add an f16 case to ggml_add_cast_impl and llama_build_lora_finetune_graphs
    AndrewGodfrey committed 2 years ago
  • finetune.sh: Edit comments
    AndrewGodfrey committed 2 years ago
  • Add "add_f16_f32_f32_cuda"
    AndrewGodfrey committed 2 years ago
  • Tweak an error message
    AndrewGodfrey committed 2 years ago
  • finetune.sh: Add an optional LLAMA_MODEL_DIR variable
    AndrewGodfrey committed 2 years ago
  • finetune.sh: Add an optional LLAMA_TRAINING_DIR variable
    AndrewGodfrey committed 2 years ago
  • train : minor
    ggerganov committed 2 years ago
  • tabs to spaces
    AndrewGodfrey committed 2 years ago
  • Merge branch 'master' of https://github.com/AndrewGodfrey/llama.cpp into finetune_enableGpu
    AndrewGodfrey committed 2 years ago
  • Merge branch 'master' of https://github.com/ggerganov/llama.cpp into finetune_enableGpu
    cebtenzzre committed 2 years ago
FAQ Terms Privacy Refunds Impressum
Loading