ggerganov/llama.cpp

Pull Requests Commits

Control vector loading fixes (#8137)

jukofyork committed 1 year ago

Verified 97877eb1

Delete examples/llama.android/llama/CMakeLists.txt (#8165)

criminact committed 1 year ago

Verified 38795265

Add Qwen2MoE 57B-A14B model identifier (#8158)

CISC committed 1 year ago

Verified 6030c612

CUDA: fix MMQ stream-k for --split-mode row (#8167)

JohannesGaessler committed 1 year ago

Verified 85a267da

Added support for Viking pre-tokenizer (#8135)

kustaaya committed 1 year ago

Verified f675b20a

llama : fix CodeLlama FIM token checks (#8144)

CISC committed 1 year ago

Verified 911e35bb

Fix llama-android.cpp for error - "common/common.h not found" (#8145)

criminact committed 1 year ago

Verified ac146628

clip : suppress unused variable warnings (#8105)

danbev committed 1 year ago

Verified 9b31a40c

scripts : fix filename sync

ggerganov committed 1 year ago

Verified c70d117c

ci : publish new docker images only when the files change (#8142)

slaren committed 1 year ago

Verified ae5d0f4b

ggml : add GGML_CUDA_USE_GRAPHS option, restore GGML_CUDA_FORCE_CUBLAS (cmake) (#8140)

slaren committed 1 year ago

Verified 31ec3993

make : fix missing -O3 (#8143)

slaren committed 1 year ago

Verified c7ab7b61

ggerganov committed 1 year ago

Verified f2d48fff

authors : regen

ggerganov committed 1 year ago

Verified 4713bf30

devops : remove clblast + LLAMA_CUDA -> GGML_CUDA (#8139)

ggerganov committed 1 year ago

Verified 0e814dfc

readme : update API notes

ggerganov committed 1 year ago

Verified a95631ee

llama : reorganize source code + improve CMake (#8006)

ggerganov committed 1 year ago

Verified f3f65429

Clarify default MMQ for CUDA and LLAMA_CUDA_FORCE_MMQ flag (#8115)

isaac-mcfadyen committed 1 year ago

Verified 88540445

CUDA: fix misaligned shared memory read (#8123)

JohannesGaessler committed 1 year ago

Verified c8771ab5

llama : extend llm_build_ffn() to support _scale tensors (#8103)

Eddie-Wang1120 committed 1 year ago

Verified 494165f3

`json`: better support for "type" unions (e.g. nullable arrays w/ typed items) (#7863)

ochafik committed 1 year ago

Verified 9b2f16f8

`json`: fix additionalProperties, allow space after enum/const (#7840)

ochafik committed 1 year ago

Verified 6777c544

fixes #7999 (adds control vectors to all `build_XXX()` functions in `llama.cpp` [needs testing] (#8060)

jukofyork committed 1 year ago

Verified 163d50ad

llama : implement Unigram tokenizer needed by T5 and FLAN-T5 model families (#5763)

fairydreaming committed 1 year ago

Verified 6fcbf682

llama : return nullptr from llama_grammar_init (#8093)

danbev committed 1 year ago

Verified e6bf0077

`json`: support integer minimum, maximum, exclusiveMinimum, exclusiveMaximum (#7797)

ochafik committed 1 year ago

Verified 84631fe1

disable docker CI on pull requests (#8110)

slaren committed 1 year ago

Verified dd047b47

Add healthchecks to llama-server containers (#8081)

codearranger committed 1 year ago

Verified 925c3095

Gguf dump start data offset via --data-offset and some extra refactor (#8054)

mofosyne committed 1 year ago

Verified c8ad3595

cvector: better prompt handling, add "mean vector" method (#8069)

ngxson committed 1 year ago

Verified 49c03c79

Newer Older