Repository navigation

Hp/split/load model (test CI) - #4

Open

ngxson wants to merge 11 commits into

hp/split/load-model

Open

Hp/split/load model (test CI)#4
ngxson wants to merge 11 commits into
masterfrom
hp/split/load-model

Commits on Mar 20, 2024
- split: support in llama_model_loader
  phymbert
  committed
  7c64fef
Commits on Mar 21, 2024
- Avoir copying the entire vector
  
  phymbert
  and
  slaren
  authored
  b8feff4
- split: move llama_tensor_offset to llama_model_loader
  phymbert
  committed
  18ff6ca
- Merge branch 'master' into hp/split/load-model
  phymbert
  committed
  60a87ae
- llama_model_loader: PR feedbacks:
  phymbert
  committed
  1892ae7
- avoid copying the entire vector
  phymbert
  committed
  00381b0
- Simplify this by making these optional, switch some layer creation tensor optional
  
  phymbert
  and
  ggerganov
  authored
  c34a5de
- Handle optional tensors
  
  phymbert
  and
  ggerganov
  authored
  1c931f3
- llama_model_loader: fail if backend cannot allocate buffer
  phymbert
  committed
  d8b567d
- fix mmap buffer management
  slaren
  committed
  02020b0
- test new llama_split_prefix
  ngxson
  committed
  9306936