ggerganov/llama.cpp

Pull Requests Commits

gguf-py : fail fast on nonsensical special token IDs

cebtenzzre committed 1 year ago

b0547d21

server : add optional API Key Authentication example (#4441)

ShadovvBeast committed 1 year ago

Verified 88ae8952

ggml : group mul_mat_id rows by matrix (cpu only) (#4480)

slaren committed 1 year ago

Verified ee4725a6

ggml : use ggml_row_size where possible (#4472)

slaren committed 1 year ago

Verified 6744dbe9

ggml : remove n_dims from ggml_tensor (#4469)

slaren committed 1 year ago

Verified cafcd4f8

py : add protobuf dependency (#4466)

strutive07 committed 1 year ago

Verified c50e4001

ggml : add ggml_row_size() (fixes llama out of space) (#4461)

LostRuins committed 1 year ago

Verified 20a68a70

ggml : fix OpenCL broadcast requirement for ggml_mul (close #4453)

ggerganov committed 1 year ago

Verified 55e87c37

convert : support loading vocab from fast tokenizer config (#3633)

strutive07 committed 1 year ago

Verified 873637af

readme : update supported model list (#4457)

BarfingLemurs committed 1 year ago

Verified 0353a184

server : fix handling of characters that span multiple tokens when streaming (#4446)

shibe2 committed 1 year ago

Verified 948ff137

sync : ggml (SD ops, tests, kernels) (#4444)

ggerganov committed 1 year ago

Verified 4d98d9a6

build : detect host compiler and cuda compiler separately (#4414)

cebtenzzre committed 1 year ago

Verified 70f806b8

common : add `--version` option to show build info in CLI (#4433)

yusiwen committed 1 year ago

Verified 9fb13f95

readme : update hot topics

ggerganov committed 1 year ago

Verified 113f9942

llama : add Mixtral support (#4406)

slaren committed 1 year ago

Verified 799a1cb1

server : tweak default sampling parameters (#4367)

kalomaze committed 1 year ago

Verified fecac456

english : use `typos` to fix comments and logs (#4354)

richardkiss committed 1 year ago

Verified 9494d7c4

build : target Windows 8 for standard mingw-w64 (#4405)

cebtenzzre committed 1 year ago

Verified 6138963f

llama : document logits_all deprecation (#4418)

crasm committed 1 year ago

Verified 6391817c

server : fix local model name in server (#4420)

epicfilemcnulty committed 1 year ago

Verified d9d4cfef

ggml : increased GGML_MAX_PARAMS to allow finetuning of 70b models (#4424)

Taikono-Himazin committed 1 year ago

Verified 41a11aaf

Update README.md (#4388)

y10ab1 committed 1 year ago

Verified 8a7b2fa5

grammar : revert the replacement of llama_token_to_piece with id_to_token (#4396)

AlienKevin committed 1 year ago

Verified e18f7345

sync : ggml (new ops, tests, backend, etc.) (#4359)

ggerganov committed 1 year ago

Verified fe680e3d

llama : per-layer KV cache + quantum K cache (#4309)

ggerganov committed 1 year ago

Verified bcc0eb45

train : fix #4227 (double free in examples/train-text-from-scratch/train-text-from-scratch.cpp) (#4351)

casavaca committed 1 year ago

Verified 81bc9214

server : recognize cache_prompt parameter in OAI API (#4347)

ggerganov committed 1 year ago

Verified 05cd6e50

common : fix compile warning

ggerganov committed 1 year ago

Verified caa92492

speculative : support `--color` (#4343)

stduhpf committed 1 year ago

Verified da5eaef1

Older