Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

common : disable repeat penalties by default #6127

Merged
merged 1 commit into from Mar 19, 2024

Conversation

ggerganov
Copy link
Owner

@ExtReMLapin
Copy link
Contributor

Isn't that a little dangerous to change it for all others models ?

@ggerganov ggerganov merged commit b80cf3b into master Mar 19, 2024
56 of 64 checks passed
@ggerganov ggerganov deleted the gg/disable-repeat-penalties branch March 19, 2024 08:21
phymbert added a commit that referenced this pull request Mar 20, 2024
@phymbert
Copy link
Collaborator

phymbert commented Mar 20, 2024

FYI it breaks the server tests as new tokens sequences are now generated. Strange it is only on windows. Fixed in:

phymbert added a commit that referenced this pull request Mar 20, 2024
phymbert added a commit that referenced this pull request Mar 20, 2024
phymbert added a commit that referenced this pull request Mar 20, 2024
…6146)

* server tests : remove seemingly redundant newlines in print()

* server tests : use built-in subprocess features, not os.kill and psutil

* server tests : do not catch e.g. SystemExit; use print_exc

* server tests: handle TimeoutExpired exception

* server tests: fix connect on dual-stack systems

* server: tests: add new tokens regex on windows generated following new repeat penalties default changed in (#6127)

* server: tests: remove the hack on windows since now we get the good socket family

* server: tests: add new tokens regex following new repeat penalties default changed in (#6127)

* server: tests: add new tokens regex following new repeat penalties default changed in (#6127)

---------

Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
@ggerganov
Copy link
Owner Author

FYI it breaks the server tests as new tokens sequences are now generated. Strange it is only on windows. Fixed in:

Thanks and apologies for breaking that

@phymbert
Copy link
Collaborator

Thanks and apologies for breaking that

No worries at all, I think this is good that tests fail for this kind of changes

hodlen pushed a commit to hodlen/llama.cpp that referenced this pull request Apr 1, 2024
hodlen pushed a commit to hodlen/llama.cpp that referenced this pull request Apr 1, 2024
…gerganov#6146)

* server tests : remove seemingly redundant newlines in print()

* server tests : use built-in subprocess features, not os.kill and psutil

* server tests : do not catch e.g. SystemExit; use print_exc

* server tests: handle TimeoutExpired exception

* server tests: fix connect on dual-stack systems

* server: tests: add new tokens regex on windows generated following new repeat penalties default changed in (ggerganov#6127)

* server: tests: remove the hack on windows since now we get the good socket family

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

---------

Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
hodlen pushed a commit to hodlen/llama.cpp that referenced this pull request Apr 3, 2024
hodlen pushed a commit to hodlen/llama.cpp that referenced this pull request Apr 3, 2024
…gerganov#6146)

* server tests : remove seemingly redundant newlines in print()

* server tests : use built-in subprocess features, not os.kill and psutil

* server tests : do not catch e.g. SystemExit; use print_exc

* server tests: handle TimeoutExpired exception

* server tests: fix connect on dual-stack systems

* server: tests: add new tokens regex on windows generated following new repeat penalties default changed in (ggerganov#6127)

* server: tests: remove the hack on windows since now we get the good socket family

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

---------

Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
tybalex pushed a commit to tybalex/function.cpp that referenced this pull request Apr 17, 2024
…gerganov#6146)

* server tests : remove seemingly redundant newlines in print()

* server tests : use built-in subprocess features, not os.kill and psutil

* server tests : do not catch e.g. SystemExit; use print_exc

* server tests: handle TimeoutExpired exception

* server tests: fix connect on dual-stack systems

* server: tests: add new tokens regex on windows generated following new repeat penalties default changed in (ggerganov#6127)

* server: tests: remove the hack on windows since now we get the good socket family

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

* server: tests: add new tokens regex following new repeat penalties default changed in (ggerganov#6127)

---------

Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants