New: [AEA-6581] - llm evaluation poc#565
Open
bencegadanyi1-nhs wants to merge 4 commits intomainfrom
Open
Conversation
There was a problem hiding this comment.
Pull request overview
Adds a proof-of-concept LLM/RAG evaluation suite (DeepEval-based) for the EPS chatbot and wires a smoke run into the PR deployment workflow to provide early quality signals.
Changes:
- Introduces
packages/ragasEvaluationwith DeepEval tests, fixtures, and a Bedrock-backed judge model plus a Lambda/KB client. - Adds a new Poetry dependency group (
ragasEvaluation) and updates the lockfile accordingly. - Adds
make eval-smoke/make eval-fulltargets and runs smoke evaluation in the PR release workflow.
Reviewed changes
Copilot reviewed 11 out of 14 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
pyproject.toml |
Adds ragasEvaluation dependency group; bumps urllib3 patch. |
poetry.lock |
Regenerated lockfile to include DeepEval + transitive deps and new group membership. |
packages/ragasEvaluation/tests/test_chatbot_eval.py |
Smoke/full DeepEval test suite driving live chatbot calls and metric evaluation. |
packages/ragasEvaluation/tests/conftest.py |
Session bootstrap + Bedrock judge fixture with xdist-safe caching. |
packages/ragasEvaluation/test_cases.json |
Defines evaluation prompts and ground-truth expectations (incl. smoke subset). |
packages/ragasEvaluation/pytest.ini |
Pytest defaults for the evaluation suite (incl. xdist settings). |
packages/ragasEvaluation/evaluation/chatbot.py |
Direct Lambda invocation + KB retrieve contexts for metric inputs. |
packages/ragasEvaluation/evaluation/bedrock_judge.py |
Custom DeepEval judge model using Bedrock converse. |
packages/ragasEvaluation/.deepeval/.deepeval_telemetry.txt |
Adds DeepEval telemetry artefact to repo. |
Makefile |
Adds eval-smoke and eval-full targets. |
.grype.yaml |
Adds vulnerability ignore list. |
.github/workflows/release_all_stacks.yml |
Adds PR-only “Chatbot RAG Evaluation” job running smoke evaluation. |
| python_files = test_*.py | ||
| python_functions = test_* | ||
| pythonpath = . | ||
| addopts = -v --tb=short -n 4 |
Comment on lines
+28
to
+45
| def pytest_sessionstart(session: pytest.Session) -> None: | ||
| """Resolve Lambda name and KB ID once before any tests run. | ||
|
|
||
| Uses a file lock so only the first xdist worker calls CloudFormation; | ||
| the rest read cached values from a shared JSON file. | ||
| """ | ||
| # Use a stable path so all workers (separate processes) share it. | ||
| cache_file = Path(tempfile.gettempdir()) / "eval_bootstrap_cache.json" | ||
| lock_file = Path(tempfile.gettempdir()) / "eval_bootstrap_cache.json.lock" | ||
|
|
||
| with FileLock(str(lock_file)): | ||
| if cache_file.is_file(): | ||
| data = json.loads(cache_file.read_text()) | ||
| os.environ["_EVAL_LAMBDA_NAME"] = data["lambda_name"] | ||
| os.environ["_EVAL_KB_ID"] = data["kb_id"] | ||
| else: | ||
| bootstrap() | ||
| cache_file.write_text( |
Comment on lines
+2
to
+8
| - vulnerability: GHSA-38jv-5279-wg99 | ||
| - vulnerability: GHSA-vfmq-68hx-4jfw | ||
| - vulnerability: GHSA-p423-j2cm-9vmq | ||
| - vulnerability: GHSA-58qw-9mgm-455v | ||
| - vulnerability: GHSA-r6ph-v2qm-q3c2 | ||
| - vulnerability: GHSA-6w46-j5rx-g56g | ||
| - vulnerability: GHSA-gc5v-m9x4-r6x2 |
Comment on lines
+213
to
+254
| chatbot_evaluation: | ||
| name: Chatbot RAG Evaluation | ||
| runs-on: ubuntu-22.04 | ||
| container: | ||
| image: ${{ inputs.pinned_image }} | ||
| options: --user 1001:1001 --group-add 128 | ||
| defaults: | ||
| run: | ||
| shell: bash | ||
| if: ${{ always() && !failure() && !cancelled() && inputs.IS_PULL_REQUEST == true | ||
| }} | ||
| needs: [ release_all_code ] | ||
| permissions: | ||
| id-token: write | ||
| contents: read | ||
| steps: | ||
| - name: copy .tool-versions | ||
| run: | | ||
| cp /home/vscode/.tool-versions "$HOME/.tool-versions" | ||
|
|
||
| - name: Checkout code | ||
| uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd | ||
| with: | ||
| persist-credentials: false | ||
|
|
||
| - name: Configure AWS Credentials | ||
| uses: aws-actions/configure-aws-credentials@ec61189d14ec14c8efccab744f656cffd0e33f37 | ||
| with: | ||
| aws-region: eu-west-2 | ||
| role-to-assume: ${{ secrets.DEV_CLOUD_FORMATION_EXECUTE_LAMBDA_ROLE }} | ||
| role-session-name: eps-assist-me-evaluation | ||
|
|
||
| - name: Install dependencies | ||
| run: | | ||
| make install-python | ||
|
|
||
| - name: Run smoke evaluation | ||
| env: | ||
| CHATBOT_STACK_NAME: ${{ inputs.STACK_NAME }} | ||
| AWS_REGION: eu-west-2 | ||
| run: | | ||
| make eval-smoke |
Comment on lines
+1
to
+4
| DEEPEVAL_ID=d0395a29-36cf-4018-afbd-c2834073dfcc | ||
| DEEPEVAL_STATUS=old | ||
| DEEPEVAL_LAST_FEATURE=evaluation | ||
| DEEPEVAL_EVALUATION_STATUS=old |
bencegadanyi1-nhs
temporarily deployed
to
dev-pr
GitHub Actions
Inactive
bencegadanyi1-nhs
temporarily deployed
to
dev-pr
GitHub Actions
Inactive
|
bencegadanyi1-nhs
temporarily deployed
to
dev-pr
GitHub Actions
Inactive
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
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.
Summary
Details