Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on May 9, 2024

• feat: Support multiple chat templates - step 1 (#1396)

  Show description for 5ab40e6
  

  CISC

  and

  abetlen

  authored
  5ab40e6

  Copy full SHA for 5ab40e6

Commits on May 8, 2024

• feat: fill-in-middle support (#1386)

  Show description for 4a7122d
  

  CISC

  and

  abetlen

  authored
  4a7122d

  Copy full SHA for 4a7122d

• fix: chat_format log where auto-detected format prints `None` (#1434)

  balvisio

  authored
  a50d24e

  Copy full SHA for a50d24e

Commits on May 3, 2024

• fix: Use memmove to copy str_value kv_override. Closes #1417

  abetlen

  committed
  9f7a855

  Copy full SHA for 9f7a855

Commits on Apr 30, 2024

• fix: wrong parameter for flash attention in pickle __getstate__

  abetlen

  committed
  29b6e9a

  Copy full SHA for 29b6e9a

• feat: Add option to enable `flash_attn` to Lllama params and ModelSettings

  abetlen

  committed
  22d77ee

  Copy full SHA for 22d77ee

Commits on Apr 28, 2024

• feat: Add support for str type kv_overrides

  abetlen

  committed
  a411612

  Copy full SHA for a411612

Commits on Apr 26, 2024

• feat: Allow for possibly non-pooled embeddings (#1380)

  Show description for f6ed21f
  

  iamlemec

  and

  abetlen

  authored
  f6ed21f

  Copy full SHA for f6ed21f

Commits on Apr 22, 2024

• feat: Use new llama_token_is_eog in create_completions

  abetlen

  committed
  d40a250

  Copy full SHA for d40a250

Commits on Apr 20, 2024

• feat: Add stopping_criteria to ChatFormatter, allow stopping on arbitrary token ids, fixes llama3 instruct

  abetlen

  committed
  cc81afe

  Copy full SHA for cc81afe

Commits on Apr 17, 2024

• feat: Make saved state more compact on-disk (#1296)

  Show description for 4924455
  tc-wolf

  authored
  4924455

  Copy full SHA for 4924455

• feat: Use all available CPUs for batch processing (#1345)

  

  ddh0

  authored and

  abetlen

  committed
  c96b2da

  Copy full SHA for c96b2da

Commits on Apr 10, 2024

• fix: pass correct type to chat handlers for chat completion logprobs

  abetlen

  committed
  bb65b4d

  Copy full SHA for bb65b4d

Commits on Apr 3, 2024

• fix: segfault when logits_all=False. Closes #1319

  abetlen

  committed
  8649d76

  Copy full SHA for 8649d76

Commits on Apr 1, 2024

• feat: add support for KV cache quantization options (#1307)

  Show description for f165048
  

  Limour-dev

  and

  abetlen

  authored
  f165048

  Copy full SHA for f165048

Commits on Mar 31, 2024

• feat: Add logprobs support to chat completions (#1311)

  Show description for aa9f1ae
  

  windspirit95

  and

  abetlen

  authored
  aa9f1ae

  Copy full SHA for aa9f1ae

Commits on Mar 14, 2024

• fix: set default pooling type to unspecified

  abetlen

  committed
  4084aab

  Copy full SHA for 4084aab

• fix: Set default pooling_type to mean, check for null pointer.

  abetlen

  committed
  d318cc8

  Copy full SHA for d318cc8

Commits on Mar 9, 2024

• feat: Switch embed to llama_get_embeddings_seq (#1263)

  Show description for 2811014
  

  iamlemec

  and

  abetlen

  authored
  2811014

  Copy full SHA for 2811014

Commits on Mar 6, 2024

• Update llama.cpp

  abetlen

  committed
  93dc56a

  Copy full SHA for 93dc56a

Commits on Mar 1, 2024

• docs: Add information re: auto chat formats. Closes #1236

  abetlen

  committed
  97aa3a1

  Copy full SHA for 97aa3a1

• feat: Update llama.cpp

  abetlen

  committed
  f062a7f

  Copy full SHA for f062a7f

Commits on Feb 28, 2024

• fix: eos/bos_token set correctly for Jinja2ChatFormatter and automatic chat formatter (#1230)

  Show description for c36ab15
  CISC

  authored
  c36ab15

  Copy full SHA for c36ab15

Commits on Feb 25, 2024

• feat: Update llama.cpp

  abetlen

  committed
  2292af5

  Copy full SHA for 2292af5

Commits on Feb 23, 2024

• fix: LlamaHFTokenizer now receives pre_tokens

  abetlen

  committed
  47bad30

  Copy full SHA for 47bad30

• fix: module 'llama_cpp.llama_cpp' has no attribute 'c_uint8'

  abetlen

  committed
  db776a8

  Copy full SHA for db776a8

Commits on Feb 22, 2024

• fix: Update from_pretrained defaults to match hf_hub_download

  abetlen

  committed
  e6d6260

  Copy full SHA for e6d6260

Commits on Feb 21, 2024

• feat(low-level-api): Improve API static type-safety and performance (#1205)

  abetlen

  authored
  7f51b60

  Copy full SHA for 7f51b60

• feat: Pull models directly from huggingface (#1206)

  Show description for 0f8aa4a
  abetlen

  authored
  0f8aa4a

  Copy full SHA for 0f8aa4a

Commits on Feb 17, 2024

• fix: self.numa missing

  abetlen

  committed
  53f6f5f

  Copy full SHA for 53f6f5f

• feat: Update llama.cpp

  abetlen

  committed
  fdce078

  Copy full SHA for fdce078

Commits on Feb 15, 2024

• fix: create_embedding broken response for input type str

  abetlen

  committed
  0ce66bc

  Copy full SHA for 0ce66bc

• fix: Incorporate embedding pooling layer fixes (#1194)

  Show description for 7bb91f0
  iamlemec

  authored
  7bb91f0

  Copy full SHA for 7bb91f0

Commits on Feb 14, 2024

• feat: Support batch embeddings (#1186)

  Show description for d7a6791
  

  iamlemec

  and

  abetlen

  authored
  d7a6791

  Copy full SHA for d7a6791

Commits on Feb 13, 2024

• fix: sample idx off-by-one error for logit_processors (#1179)

  Show description for d6be533
  

  3 peopleauthored
  d6be533

  Copy full SHA for d6be533