llama.cpp
feat: support StarCoder model architectures
#3187
Merged
Overview
Commits
30
Changes
View On GitHub
Commits
  • add placeholder of starcoder in gguf / llama.cpp
    wsxiaoys committed 2 years ago
  • support convert starcoder weights to gguf
    wsxiaoys committed 2 years ago
  • convert MQA to MHA
    wsxiaoys committed 2 years ago
  • fix ffn_down name
    wsxiaoys committed 2 years ago
  • add LLM_ARCH_STARCODER to llama.cpp
    wsxiaoys committed 2 years ago
  • set head_count_kv = 1
    wsxiaoys committed 2 years ago
  • load starcoder weight
    wsxiaoys committed 2 years ago
  • add max_position_embeddings
    wsxiaoys committed 2 years ago
  • set n_positions to max_positioin_embeddings
    wsxiaoys committed 2 years ago
  • properly load all starcoder params
    wsxiaoys committed 2 years ago
  • fix head count kv
    wsxiaoys committed 2 years ago
  • fix comments
    wsxiaoys committed 2 years ago
  • fix vram calculation for starcoder
    wsxiaoys committed 2 years ago
  • store mqa directly
    wsxiaoys committed 2 years ago
  • add input embeddings handling
    wsxiaoys committed 2 years ago
  • add TBD
    wsxiaoys committed 2 years ago
  • working in cpu, metal buggy
    wsxiaoys committed 2 years ago
  • cleanup useless code
    wsxiaoys committed 2 years ago
  • metal : fix out-of-bounds access in soft_max kernels
    ggerganov committed 2 years ago
  • llama : make starcoder graph build more consistent with others
    ggerganov committed 2 years ago
  • Merge pull request #2 from ggerganov/support-starcoder-fix
    wsxiaoys committed 2 years ago
  • refactor: cleanup comments a bit
    wsxiaoys committed 2 years ago
  • add other starcoder models: 3B, 7B, 15B
    wsxiaoys committed 2 years ago
  • support-mqa-directly
    wsxiaoys committed 2 years ago
  • Merge pull request #3 from TabbyML/support-starcoder-mqa
    wsxiaoys committed 2 years ago
  • fix: remove max_position_embeddings, use n_train_ctx
    wsxiaoys committed 2 years ago
  • Update llama.cpp
    wsxiaoys committed 2 years ago
  • Update llama.cpp
    wsxiaoys committed 2 years ago
  • Apply suggestions from code review
    wsxiaoys committed 2 years ago
  • fix: switch to space from tab
    wsxiaoys committed 2 years ago
FAQ Terms Privacy Refunds Impressum
Loading