ggerganov/llama.cpp

Pull Requests Commits

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

cebtenzzre committed 2 years ago

Verified 6138963f

llama : document logits_all deprecation (#4418)

crasm committed 2 years ago

Verified 6391817c

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

epicfilemcnulty committed 2 years ago

Verified d9d4cfef

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

Taikono-Himazin committed 2 years ago

Verified 41a11aaf

Update README.md (#4388)

y10ab1 committed 2 years ago

Verified 8a7b2fa5

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

AlienKevin committed 2 years ago

Verified e18f7345

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

ggerganov committed 2 years ago

Verified fe680e3d

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

ggerganov committed 2 years ago

Verified bcc0eb45

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

casavaca committed 2 years ago

Verified 81bc9214

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

ggerganov committed 2 years ago

Verified 05cd6e50

common : fix compile warning

ggerganov committed 2 years ago

Verified caa92492

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

stduhpf committed 2 years ago

Verified da5eaef1

grammar : pre-computed pieces + reserve mem + less string copies (#4330)

MarcusDunn committed 2 years ago

Verified 5f6e0c0d

llama : allow overriding GGUF metadata when loading model (#4092)

KerfuffleV2 committed 2 years ago

Verified 5aa365d8

sampling : custom samplers order (#4285)

MaggotHATE committed 2 years ago

Verified 52c8bc3c

swift : revert compiler checks for swift package (#4332)

kchro3 committed 2 years ago

Verified e4b76bbe

simple : update error message for KV cache check (#4324)

danbev committed 2 years ago

Verified 23b5e12e

swift : fix concatenation method to avoid invalid UTF8 stringfication (#4325)

ensan-hcl committed 2 years ago

Verified d208995c

swift : fix prompt tokenization logic (#4321)

ensan-hcl committed 2 years ago

Verified 5c9f90cb

grammar-parser : fix typo (#4318)

eltociear committed 2 years ago

Verified 4fa44e84

ggml : reuse ggml_get_n_tasks() in ggml_graph_plan() (#4308)

ggerganov committed 2 years ago

Verified fbbc4282

ggml : fix soft max out-of-bounds access (#4307)

ggerganov committed 2 years ago

Verified adf3de4f

server : fix OpenAI API `stop` field to be optional (#4299)

Mardak committed 2 years ago

Verified 33e171d1

py : add grammar to oai like api (#4294)

neph1 committed 2 years ago

Verified 6949b50d

llama : pad KV cache size (#4280)

ggerganov committed 2 years ago

Verified d7b800b8

llama : avoid using "optional" keyword (#4283)

ggerganov committed 2 years ago

Verified 5a7d3125

llama : support optional tensors (#4283)

ggerganov committed 2 years ago

Verified d5a1cbde

swift : fix token_to_piece implementation (#4278)

ensan-hcl committed 2 years ago

Verified b220222a

build : enable libstdc++ assertions for debug builds (#4275)

cebtenzzre committed 2 years ago

Verified 511f52c3

llama : support attention bias on LLaMA architecture (#4283)

RealJosephus committed 2 years ago

Verified 03562f3a

Newer Older