Skip to content

Use Flex processing for the OpenAI judge model #152

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.

Sign up for GitHub

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

Merged
JosephMarinier merged 1 commit into from
Jun 18, 2026
+8 −7
Merged

Use Flex processing for the OpenAI judge model #152

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/eva/metrics/accuracy/faithfulness.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,6 +60,7 @@ class FaithfulnessJudgeMetric(ConversationTextJudgeMetric):
)
category = "accuracy"
default_model = "us.anthropic.claude-opus-4-6-v1"
default_params = {"max_tokens": 100000} # Drop the OpenAI-only flex tier inherited from TextJudgeMetric.

@fanny-riols fanny-riols Jun 17, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

are we sure 100k is enough? Did you check the prompt size after 10 min convo?

@JosephMarinier JosephMarinier Jun 17, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I did not check that myself, but it has been 100k since the beginning of EVA (inherited from src/eva/metrics/base.py). Is that OK?

@fanny-riols fanny-riols Jun 17, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ah right, I guess so then.

rating_scale = (1, 3)

def get_prompt_variables(self, context: MetricContext, transcript_text: str) -> dict[str, Any]:
Expand Down
2 changes: 1 addition & 1 deletion src/eva/metrics/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -330,7 +330,7 @@ class TextJudgeMetric(BaseMetric):

# Subclasses can override these
default_model = "gpt-5.2"
default_params: dict[str, Any] = {"max_tokens": 100000}
default_params: dict[str, Any] = {"max_tokens": 100000, "service_tier": "flex"}
rating_scale: tuple[int, int] = (1, 3) # (min, max)

def __init__(self, config: dict[str, Any] | None = None):
Expand Down
12 changes: 6 additions & 6 deletions tests/fixtures/metric_signatures.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@
"ConcisenessJudgeMetric": {
"name": "conciseness",
"prompt_hash": "57caf9a18a3c",
"source_hash": "e39cacd719a2",
"source_hash": "9f183871e45a",
"version": "v0.1"
},
"ConversationCorrectlyFinishedMetric": {
Expand All @@ -20,7 +20,7 @@
"ConversationProgressionJudgeMetric": {
"name": "conversation_progression",
"prompt_hash": "174cdf93b398",
"source_hash": "8d1a32e8a4fe",
"source_hash": "4e5f59c32946",
"version": "v0.1"
},
"ConversationTimeLimitExceededMetric": {
Expand All @@ -38,7 +38,7 @@
"FaithfulnessJudgeMetric": {
"name": "faithfulness",
"prompt_hash": "060e3f1f9e7a",
"source_hash": "7db40c133d8a",
"source_hash": "93a559d0a028",
"version": "v0.2"
},
"ResponseSpeedMetric": {
Expand All @@ -56,7 +56,7 @@
"SpeakabilityJudgeMetric": {
"name": "speakability",
"prompt_hash": "1e4f78cb051e",
"source_hash": "dd4feece62c1",
"source_hash": "c8a781f5795b",
"version": "v0.2"
},
"SpeechFidelityMetric": {
Expand Down Expand Up @@ -86,7 +86,7 @@
"TranscriptionAccuracyKeyEntitiesMetric": {
"name": "transcription_accuracy_key_entities",
"prompt_hash": "aa1a7fdd6df7",
"source_hash": "80a4d892e1f6",
"source_hash": "96271903ded0",
"version": "v0.2"
},
"TurnTakingMetric": {
Expand All @@ -98,7 +98,7 @@
"UserBehavioralFidelityMetric": {
"name": "user_behavioral_fidelity",
"prompt_hash": "21527fa8d79b",
"source_hash": "fded1ad48aa4",
"source_hash": "7f5dcec51568",
"version": "v0.1"
},
"UserSpeechFidelityMetric": {
Expand Down
Loading