Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Apr 29, 2023

• llama_cpp server: slight refactor to init_llama function

  Show description for efe8e6f
  Stonelinks

  committed
  efe8e6f

  Copy full SHA for efe8e6f

• tests: simple test for server module

  Stonelinks

  committed
  6d8db9d

  Copy full SHA for 6d8db9d

• llama_cpp server: app is now importable, still runnable as a module

  Stonelinks

  committed
  468377b

  Copy full SHA for 468377b

• Merge pull request #118 from SagsMug/main

  Show description for 755f9fa
  abetlen

  authored
  755f9fa

  Copy full SHA for 755f9fa

• Remove excessive errors="ignore" and add utf8 test

  

  Mug

  committed
  18a0c10

  Copy full SHA for 18a0c10

Commits on Apr 28, 2023

• Update llama.cpp

  abetlen

  committed
  ea0faab

  Copy full SHA for ea0faab

• Python weirdness

  

  Mug

  committed
  b7d14ef

  Copy full SHA for b7d14ef

• Dont detect off tokens, detect off detokenized utf8

  

  Mug

  committed
  eed6128

  Copy full SHA for eed6128

• One day, i'll fix off by 1 errors permanently too

  

  Mug

  committed
  3a98747

  Copy full SHA for 3a98747

• Detect multi-byte responses and wait

  

  Mug

  committed
  c39547a

  Copy full SHA for c39547a

Commits on Apr 27, 2023

• Update llama.cpp

  abetlen

  committed
  9339929

  Copy full SHA for 9339929

Commits on Apr 26, 2023

• Also ignore errors on input prompts

  

  Mug

  committed
  5f81400

  Copy full SHA for 5f81400

• Merge branch 'main' of https://github.com/abetlen/llama-cpp-python

  

  Mug

  committed
  be2c961

  Copy full SHA for be2c961

• Fix decode errors permanently

  

  Mug

  committed
  c4a8491

  Copy full SHA for c4a8491

Commits on Apr 25, 2023

• Update llama.cpp

  abetlen

  committed
  cbd26fd

  Copy full SHA for cbd26fd

• Update n_batch for server

  abetlen

  committed
  3cab3ef

  Copy full SHA for 3cab3ef

• Add ctx check and re-order __init__. Closes #112

  abetlen

  committed
  cc706fb

  Copy full SHA for cc706fb

• Bugfix: Check cache keys as prefix to prompt tokens

  abetlen

  committed
  d484c56

  Copy full SHA for d484c56

Commits on Apr 24, 2023

• Add cache implementation using llama state

  abetlen

  committed
  cbe95bb

  Copy full SHA for cbe95bb

• Merge branch 'main' of github.com:abetlen/llama_cpp_python into main

  abetlen

  committed
  2c359a2

  Copy full SHA for 2c359a2

• Add save/load state api for Llama class

  abetlen

  committed
  197cf80

  Copy full SHA for 197cf80

• Refactor internal state for Llama class

  abetlen

  committed
  86f8e5a

  Copy full SHA for 86f8e5a

• Merge pull request #108 from eiery/main

  Show description for f374561
  abetlen

  authored
  f374561

  Copy full SHA for f374561

• Update llama.cpp

  abetlen

  committed
  02cf881

  Copy full SHA for 02cf881

• Update llama.py

  Show description for aa12d8a
  

  eiery

  authored
  aa12d8a

  Copy full SHA for aa12d8a

Commits on Apr 23, 2023

• Disable mmap when applying lora weights. Closes #107

  abetlen

  committed
  7230599

  Copy full SHA for 7230599

Commits on Apr 22, 2023

• Update llama.cpp

  abetlen

  committed
  e99caed

  Copy full SHA for e99caed

Commits on Apr 21, 2023

• Update llama.cpp

  abetlen

  committed
  1eb130a

  Copy full SHA for 1eb130a

Commits on Apr 19, 2023

• Add use_mmap flag to server

  abetlen

  committed
  e4647c7

  Copy full SHA for e4647c7

• If lora base is not set avoid re-loading the model by passing NULL

  abetlen

  committed
  0df4d69

  Copy full SHA for 0df4d69

• Update type signature to allow for null pointer to be passed.

  abetlen

  committed
  95c0dc1

  Copy full SHA for 95c0dc1

Commits on Apr 18, 2023

• Add seperate lora_base path for applying LoRA to quantized models using original unquantized model weights.

  abetlen

  committed
  453e517

  Copy full SHA for 453e517

• Add lora_path parameter to Llama model

  abetlen

  committed
  eb7f278

  Copy full SHA for eb7f278

• Add bindings for LoRA adapters. Closes #88

  abetlen

  committed
  35abf89

  Copy full SHA for 35abf89

Commits on Apr 15, 2023

• Bugfix: only eval new tokens

  abetlen

  committed
  89856ef

  Copy full SHA for 89856ef