ochafik/llama.cpp

Pull Requests Commits

server : (webui) revert hacky solution from #11626 (#11634)

ngxson committed 1 year ago

Verified b3451785

server : (webui) allow typing and submitting during llm response (#11626)

woof-dog committed 1 year ago

Verified 1d1e6a90

simpler hacky fixes for original broken template (+ fix minja example syntax polyfill)

Olivier Chafik committed 1 year ago

1c302e18

rm unneeded vocab

Olivier Chafik committed 1 year ago

c6214ee9

tool-calls: add deepseek r1 template + accommodate broken official template slightly better

Olivier Chafik committed 1 year ago

7dc271fb

Merge branch 'jinja-chatml' into r1-toolcall

Olivier Chafik committed 1 year ago

0be7f652

Simplify default chatml logic

Olivier Chafik committed 1 year ago

d73448de

tool-calls: accommodate variety of wrong tool call opening tags both Qwen 32B and 7B distills like to spit out

Olivier Chafik committed 1 year ago

569610ee

tweak delta logic

Olivier Chafik committed 1 year ago

c397bd1f

tool-calls: r1: add missing <｜tool▁calls▁end｜> to grammar!

Olivier Chafik committed 1 year ago

df3474e2

Merge branch 'jinja-chatml' into r1-toolcall

Olivier Chafik committed 1 year ago

08271b55

add missing try catch around jinja parsing to default to chatml

Olivier Chafik committed 1 year ago

b2dd4909

Merge branch 'jinja-chatml' into r1-toolcall

Olivier Chafik committed 1 year ago

4cb0e1d8

fix build / rm diff

Olivier Chafik committed 1 year ago

2b3c4829

server : remove CPPHTTPLIB_NO_EXCEPTIONS define (#11622)

danbev committed 1 year ago

Verified 5598f475

Olivier Chafik committed 1 year ago

aa98e590

fix double bos issue (drop bos/eos tokens from jinja template)

Olivier Chafik committed 1 year ago

5d18d76b

Olivier Chafik committed 1 year ago

cf83623a

ggerganov committed 1 year ago

Verified 8ec05832

CUDA: fix Volta FlashAttention logic (#11615)

JohannesGaessler committed 1 year ago

Verified 21c84b5d

ochafik committed 1 year ago

a76073cf

Update test_tool_call.py

ochafik committed 1 year ago

77ae97e7

server : (webui) Fix Shift+Enter handling (#11609)

mashdragon committed 1 year ago

Verified d92cb67e

tool-call: allow `--jinja --chat-template chatml`

ochafik committed 1 year ago

1e9acd2d

add deepseek models to server tool call section in readme

ochafik committed 1 year ago

5e6f2a21

tell DS R1 not to overthink (weather test)

ochafik committed 1 year ago

19bea4ec

tool-calls: add DeepSeek R1 Qwen 7B to server test_hello_world

ochafik committed 1 year ago

ae9d5812

tool-call: fix command-r7b parsing when response is multiline

ochafik committed 1 year ago

04be723b

tool-call: allow `--jinja --chat-template chatml`

ochafik committed 1 year ago

73d08d49

ochafik committed 1 year ago

08716281

Newer Older