Skip to content

Conversation

@ngxson
Copy link
Collaborator

@ngxson ngxson commented Sep 23, 2024

Superseding #9544

Fix #9390

In this PR:

  • Add --no-context-shift to server
  • Added test cases for no context shift
  • Added one test case for embedding when input prompt is too big

@github-actions github-actions bot added examples python python script changes server labels Sep 23, 2024
@ngxson ngxson mentioned this pull request Sep 23, 2024
4 tasks
@ngxson ngxson merged commit 0b3bf96 into master Sep 23, 2024
55 checks passed
dsx1986 pushed a commit to dsx1986/llama.cpp that referenced this pull request Oct 29, 2024
* server : add --no-context-shift option

* small fix

* Update examples/server/tests/features/embeddings.feature

Co-authored-by: Georgi Gerganov <[email protected]>

* tests : minor fix

* revert usage of GGML_ASSERT

* update server documentation

---------

Co-authored-by: Georgi Gerganov <[email protected]>
arthw pushed a commit to arthw/llama.cpp that referenced this pull request Nov 15, 2024
* server : add --no-context-shift option

* small fix

* Update examples/server/tests/features/embeddings.feature

Co-authored-by: Georgi Gerganov <[email protected]>

* tests : minor fix

* revert usage of GGML_ASSERT

* update server documentation

---------

Co-authored-by: Georgi Gerganov <[email protected]>
arthw pushed a commit to arthw/llama.cpp that referenced this pull request Nov 18, 2024
* server : add --no-context-shift option

* small fix

* Update examples/server/tests/features/embeddings.feature

Co-authored-by: Georgi Gerganov <[email protected]>

* tests : minor fix

* revert usage of GGML_ASSERT

* update server documentation

---------

Co-authored-by: Georgi Gerganov <[email protected]>
@ngxson ngxson deleted the xsn/server_disable_ctx_shift branch December 13, 2024 11:27
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

examples python python script changes server

Projects

None yet

Development

Successfully merging this pull request may close these issues.

server : ability to disable context shift

2 participants