Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Nov 8, 2023

• Add set_seed to Llama class

  abetlen

  committed
  fd41ed3

  Copy full SHA for fd41ed3

• Fix destructor NoneType is not callable error

  abetlen

  committed
  ca4cb88

  Copy full SHA for ca4cb88

• Add JSON mode support. Closes #881

  abetlen

  committed
  b30b9c3

  Copy full SHA for b30b9c3

• Add seed parameter support for completion and chat_completion requests. Closes #884

  abetlen

  committed
  86aeb9f

  Copy full SHA for 86aeb9f

• Multimodal Support (Llava 1.5) (#821)

  Show description for aab74f0
  

  damian0815

  and

  abetlen

  authored
  aab74f0

  Copy full SHA for aab74f0

Commits on Nov 6, 2023

• Fix type bug

  abetlen

  committed
  be0add1

  Copy full SHA for be0add1

• Refactor Llama class internals

  abetlen

  committed
  e214a58

  Copy full SHA for e214a58

Commits on Nov 3, 2023

• Clean up stdout / stderr suppression

  abetlen

  committed
  2ec043a

  Copy full SHA for 2ec043a

• Rename internal only module utils to _utils

  abetlen

  committed
  4ea7027

  Copy full SHA for 4ea7027

• Update llama.cpp

  abetlen

  committed
  df9362e

  Copy full SHA for df9362e

• Add functionary support (#784)

  Show description for 3af7b21
  abetlen

  authored
  3af7b21

  Copy full SHA for 3af7b21

• Migrate inference to llama_batch and llama_decode api (#795)

  Show description for ab028cb
  

  abetlen

  and

  antoine-lizee

  authored
  ab028cb

  Copy full SHA for ab028cb

Commits on Nov 2, 2023

• Update llama.cpp

  Show description for fa83cc5
  abetlen

  committed
  fa83cc5

  Copy full SHA for fa83cc5

• fix: tokenization of special characters: (#850)

  Show description for 4d4e0f1
  

  antoine-lizee

  authored and

  abetlen

  committed
  4d4e0f1

  Copy full SHA for 4d4e0f1

Commits on Nov 1, 2023

• llama: fix exception in Llama.__del__ (#846)

  cebtenzzre

  authored
  eefd76f

  Copy full SHA for eefd76f

• fixed Llama._create_completion suffix check, it can be either None or str instance (#854)

  mtasic85

  authored
  9c8f4dc

  Copy full SHA for 9c8f4dc

Commits on Oct 24, 2023

• Update llama.cpp

  abetlen

  committed
  53861c9

  Copy full SHA for 53861c9

Commits on Oct 19, 2023

• Fix streaming doesn't return finish reason (#798)

  Show description for 09a8406
  gmcgoldr

  authored
  09a8406

  Copy full SHA for 09a8406

• Update llama.cpp

  abetlen

  committed
  ff58003

  Copy full SHA for ff58003

Commits on Oct 15, 2023

• Make use of suppress_stdout_stderr when freeing model (#803)

  paschembri

  authored
  10304d7

  Copy full SHA for 10304d7

• Add validation for tensor_split size exceeding LLAMA_MAX_DEVICES (#820)

  Show description for b501665
  eric1932

  authored
  b501665

  Copy full SHA for b501665

Commits on Sep 30, 2023

• Fix logits_all bug

  abetlen

  committed
  d696251

  Copy full SHA for d696251

• Fix bug in embedding

  abetlen

  committed
  42bb721

  Copy full SHA for 42bb721

Commits on Sep 29, 2023

• Configurable Chat Formats (#711)

  Show description for 3bca770
  abetlen

  authored
  3bca770

  Copy full SHA for 3bca770

• Fix rope scaling defaults (#767)

  Show description for a945404
  Josh-XT

  authored
  a945404

  Copy full SHA for a945404

• Update llama.cpp

  abetlen

  committed
  1a1c3dc

  Copy full SHA for 1a1c3dc

Commits on Sep 18, 2023

• Update llama.cpp

  abetlen

  committed
  38e34c9

  Copy full SHA for 38e34c9

Commits on Sep 14, 2023

• Add numa support, low level api users must now explicitly call llama_backend_init at the start of their programs.

  abetlen

  committed
  f4090a0

  Copy full SHA for f4090a0

• Reorder init params to match llama.cpp order

  abetlen

  committed
  6a20293

  Copy full SHA for 6a20293

• Explicitly make all init params other than model_path into keyword only params

  abetlen

  committed
  c8f9b8a

  Copy full SHA for c8f9b8a

• Add kwargs to init to catch extra params

  abetlen

  committed
  a68f9e2

  Copy full SHA for a68f9e2

• remove print

  abetlen

  committed
  9e345a4

  Copy full SHA for 9e345a4

• Convert missed llama.cpp constants into standard python types

  abetlen

  committed
  517f9ed

  Copy full SHA for 517f9ed

• Fix tensor_split cli option

  abetlen

  committed
  c4c440b

  Copy full SHA for c4c440b

Commits on Sep 12, 2023

• Merge branch 'main' into v0.2-wip

  abetlen

  committed
  1910793

  Copy full SHA for 1910793