[Obs AI Assistant] Improve error handling in the evaluation framework #212991
+34
−11
Conversation
This file contains 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
viduni94
added
release_note:skip
|
Pinging @elastic/obs-ai-assistant (Team:Obs AI Assistant) |
viduni94
commented
Mar 3, 2025
| const inferenceMessages: InferenceMessage[] = []; | ||
|
|
||
| messages.forEach((message) => { | ||
| if (message.message.role === MessageRole.Assistant) { | ||
| let parsedArguments; | ||
| if (message.message.function_call?.name) { |
There was a problem hiding this comment.
The idea of this change is to not validate the arguments JSON object here because any errors in the function call will be caught in the function validation step.
For example:
When the LLLM hallucinates the function name/arguments such as:
{
"content": "",
"function_call": {
"name": "elasticsearchelasticsearch",
"arguments": "{\"method\": \"POST\", \"path\": \"/testing_ai_assistant/_doc\", \"body\": {\"type\": \"alert\", \"message\": \"This test is for alerts\"}}{\"method\": \"POST\", \"path\": \"/testing_ai_assistant/_doc\", \"body\": {\"type\": \"esql\", \"message\": \"This test is for esql\"}}",
"trigger": "assistant"
},
"role": "assistant"
}
At present, since the JSON is invalid, this is the error thrown:
ERROR ChatCompletionError: SyntaxError: Unexpected non-whitespace character after JSON at position 56
at Object.next (throw_serialized_chat_completion_errors.ts:29:17)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/tap.ts:189:31
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:85:24
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at Observable._subscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/observable/innerFrom.ts:78:18)
at Observable._trySubscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:244:19)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:234:18
at Object.errorContext (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/util/errorContext.ts:29:5)
at Observable.subscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:220:5)
at doInnerSub (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:71:40)
at outerNext (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:53:58)
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/switchMap.ts:115:42
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at processStream (stream_into_observable.ts:17:20)
at processTicksAndRejections (node:internal/process/task_queues:95:5) {
code: 'internalError',
meta: {}
}
With this change (By not propagating the invalid JSON error), we can see the actual error in the terminal:
ERROR ChatCompletionError: Could not find tool call request for elasticsearchelasticsearch
at Object.next (throw_serialized_chat_completion_errors.ts:29:17)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/tap.ts:189:31
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:85:24
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at Observable._subscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/observable/innerFrom.ts:78:18)
at Observable._trySubscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:244:19)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:234:18
at Object.errorContext (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/util/errorContext.ts:29:5)
at Observable.subscribe (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Observable.ts:220:5)
at doInnerSub (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:71:40)
at outerNext (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/mergeInternals.ts:53:58)
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.Subscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at /Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/switchMap.ts:115:42
at OperatorSubscriber._this._next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/operators/OperatorSubscriber.ts:70:13)
at OperatorSubscriber.next (/Users/viduni/Workspace/Elastic/kibana/node_modules/rxjs/src/internal/Subscriber.ts:75:12)
at processStream (stream_into_observable.ts:17:20)
at processTicksAndRejections (node:internal/process/task_queues:95:5) {
code: 'internalError',
meta: {}
}
💚 Build Succeeded
Metrics [docs]
cc @viduni94 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Closes https://github.com/elastic/obs-ai-assistant-team/issues/196
Summary
This PR implements the follows:
429errorsconvertMessagesForInferenceto surface errors related to function callsChecklist
release_note:*label is applied per the guidelines