ochafik/llama.cpp

Pull Requests Commits

grammar: reuse decoded tokens and pieces

Olivier Chafik committed 1 year ago

8ee2a68d

grammars: reuse new_stack vector in llama_grammar_accept

Olivier Chafik committed 1 year ago

15fe172c

grammars: new_stacks reuse

Olivier Chafik committed 1 year ago

959d9d34

grammars: reserve rejects & next candidates

Olivier Chafik committed 1 year ago

a2747858

grammar: cache fully decoded tokens

Olivier Chafik committed 1 year ago

6c5150a2

grammar: cosmetic reserves

Olivier Chafik committed 1 year ago

10226922

json: fix server/README (json_schema in /completion vs. result_format in /v1/chat/completions)

Olivier Chafik committed 1 year ago

9c33ee99

json: fix type error w/ python 3.8

ochafik committed 1 year ago

67a5184f

ochafik committed 1 year ago

3c81e944

server+json: update server/README w/ result_format

ochafik committed 1 year ago

6c885dce

json: support string minLength/maxLength

ochafik committed 1 year ago

de4e60ea

json: unify all repetition code (w/ or w/o sep)

ochafik committed 1 year ago

181f984d

json: cap length of numbers to 15 digits before/after decimal point

ochafik committed 1 year ago

dcf5d328

grammars: add troubleshooting section to readme

ochafik committed 1 year ago

07163fb6

json: optimize repetitions for minItems/maxItems and regexps: `a{,3}` goes from `"a"? "a"? "a"?` (explosive combos) to `(a (a (a)?)?)?`

ochafik committed 1 year ago

a59e9431

json: deps management for primitive rules (+ allow null values)

ochafik committed 1 year ago

159b883b

server: skip null json_schema / grammar fields

ochafik committed 1 year ago

f771a8f1

json: rename python schema converter to make import easier

ochafik committed 1 year ago

2148f244

quantize : fix precedence of cli args (#6541)

ggerganov committed 1 year ago

Verified b73e564b

llama : support negative ith in llama_get_ API (#6519)

TheFlipbook committed 1 year ago

Verified e3c337d8

llama : save and restore kv cache for single seq id (#6341)

kaetemi committed 1 year ago

Verified beea6e1b

remove row=1 cond (#6532)

abhilash1910 committed 1 year ago

Verified 87fb5b42

Adding KodiBot to UI list (#6535)

firatkiral committed 1 year ago

Verified d752327c

Change Windows AMD example to release build to make inference much faster. (#6525)

thebaron88 committed 2 years ago

Verified 855f5440

flake.lock: Update (#6517)

ggerganov committed 2 years ago

Verified b909236c

Add GritLM as supported models. (#6513)

dranger003 committed 2 years ago

Verified e0717e75

ggerganov committed 2 years ago

Verified c3724779

ggml: bypass code incompatible with CUDA < 11.1 (whisper/2020)

primenko-v committed 2 years ago

Verified f77261a7

scripts : sync ggml-cuda folder

ggerganov committed 2 years ago

Verified 43e8995e

Run make to build the project (#6457)

limitedAtonement committed 2 years ago

Verified 9472bce3

Older