Commits

Breadcrumbs

History for

onbatch-processing

User selector

All users

All time

Commit History

Commits on Nov 30, 2023

• Refactor _create_completion

  abetlen

  committed
  40f2293

  Copy full SHA for 40f2293

Commits on Nov 28, 2023

• Add sampling context

  abetlen

  committed
  3a1ba77

  Copy full SHA for 3a1ba77

• Clean up llama.py into seperate modules.

  abetlen

  committed
  3c13436

  Copy full SHA for 3c13436

Commits on Nov 26, 2023

• docs: Update Llama docs

  abetlen

  committed
  6308f21

  Copy full SHA for 6308f21

Commits on Nov 24, 2023

• docs: Update completion and chat_completion parameter docstrings

  abetlen

  committed
  4026166

  Copy full SHA for 4026166

Commits on Nov 23, 2023

• docs: Add Llama class example

  abetlen

  committed
  b6bb7ac

  Copy full SHA for b6bb7ac

Commits on Nov 21, 2023

• Format

  abetlen

  committed
  7a3f878

  Copy full SHA for 7a3f878

• Fix: Add logit_bias to all completion api methods

  abetlen

  committed
  422ebc8

  Copy full SHA for 422ebc8

• Add support for logit_bias outside of server api. Closes #827

  abetlen

  committed
  07e47f5

  Copy full SHA for 07e47f5

• Added support for min_p (#921)

  Show description for b8438f7
  tk-master

  authored
  b8438f7

  Copy full SHA for b8438f7

• Fix #929

  abetlen

  committed
  a34d480

  Copy full SHA for a34d480

Commits on Nov 10, 2023

• Fix sampling bug when logits_all=False

  abetlen

  committed
  6f0b0b1

  Copy full SHA for 6f0b0b1

• Potential bugfix for eval

  abetlen

  committed
  d9b38e3

  Copy full SHA for d9b38e3

• Fix: default max_tokens matches openai api (16 for completion, max length for chat completion)

  abetlen

  committed
  e7962d2

  Copy full SHA for e7962d2

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