-
Notifications
You must be signed in to change notification settings - Fork 5k
chore(*): culmulative AI GW fixes 20251028 #14785
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
Open
fffonion
wants to merge
54
commits into
master
Choose a base branch
from
aigw-393
base: master
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
+6,282
−1,087
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Increase ai prompt message max length from 500 to 100000. AG-287 Signed-off-by: hackerchai <[email protected]>
…est_body_table_inuse for fixing user defiend fields missing - Add should_set_body parameter to control request body setting - Update prompt decorator to use new parameter test(ai-prompt-decorator): add test for preserving model and temperature fields - Add test case for full chat request - Verify model and temperature preservation test(ai-prompt-decorator): add integration test for preserving model and temperature fields - Add test case for openai_full_chat configuration - Verify model, temperature and max_tokens preservation - Check message decoration and context setting doc(changelog): Add fix_ai_prompt_decorator_missing_fields changlog doc(changelog): use correct type of changelog & polish message Signed-off-by: Eason Chai <[email protected]>
e6eb4cd to
95d2fa3
Compare
From #14500 AG-309 Signed-off-by: spacewander <[email protected]> Signed-off-by: Zexuan Luo <[email protected]> Co-authored-by: spacewander <[email protected]> Co-authored-by: Zexuan Luo <[email protected]> Co-authored-by: Jun Ouyang <[email protected]>
Previously, stale SSE events was not dropped, which causes repeated body (like `The answer to 1 + 1 is 2.The answer to 1 + 1 is 2.`) for observability. Signed-off-by: Zexuan Luo <[email protected]>
AG-329 --------- Signed-off-by: Zexuan Luo <[email protected]>
…ent was truncated (#13452) The previous fix (#13315) missed a branch. This is caught by a more careful fuzzing. Signed-off-by: Zexuan Luo <[email protected]> AG-385
…e was incomplete (#13430) FTI-6842 --------- Signed-off-by: Zexuan Luo <[email protected]>
…re events (#13588) AG-401 This affects Gemini streaming chunk parsing and OpenAI's /v1/files route. When using iterator in the `for` loop, the loop is terminated when the first returned value is nil, which causes the missing state update. Show by the code below: ``` local function itertool(x) local i = 0 return function() i = i + 1 if i <= #x then return x[i] end end end local function main() local x = {1, nil, 3, 4, 5} for v in itertool(x) do print(v) end end local function better_main() local x = {1, nil, 3, 4, 5} local iter = itertool(x) local eos = 5 local count = 0 while true do count = count + 1 if count > eos then break end local v = iter() if v ~= nil then print(v) end end end main() print("Fix it") better_main() ``` This PR also 1. Fixes an incorrect delimiter skipping 2. Supports using `\r` as line separator --------- Signed-off-by: Zexuan Luo <[email protected]>
…thropic provider (#13355) AG-391 --------- Signed-off-by: Zexuan Luo <[email protected]>
…om Gemini provider in some situations Signed-off-by: Zexuan Luo <[email protected]>
…ar used as model name
…ing in llm/v1/chat
…gw-only] (#14137) "Floor" is set and then prompts must abide by specific rulesets (e.g. hate, violence) else it will be blocked. Kong was not correctly handling a "bad" or "blocked" response from GCP. This PR makes that work. With this patch, the user no longer gets 500 'an error occured' and instead gets 400:
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
cherry-pick kong-ee
schedule this PR for cherry-picking to kong/kong-ee
core/pdk
plugins/ai-prompt-decorator
plugins/ai-proxy
plugins/ai-request-transformer
plugins/ai-response-transformer
schema-change-noteworthy
size/XXL
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Checklist
changelog/unreleased/kongorskip-changeloglabel added on PR if changelog is unnecessary. README.mdIssue reference
AG-532