How hsmshubham007-code/week3day4proj1 works
This repository is a deterministic customer-message classification evaluation app with Streamlit dashboards. The main workflow is to load golden cases, classify each message, score results and regression status, and record evaluation and observability logs; dashboards present those records. Source excerpts support the classifier call, CSV reads/writes, and dashboard behavior. The README is empty, and the dataset contents and some evaluator details were not sampled, so the graph omits unverified connections.
Main entry points: evaluate.py Evaluation runner · eval_log.csv Evaluation history · observability_log.csv Observability log · ui.py Evaluation dashboard
Components 7
Classification
- Message classifier
app/prompt.py
Evaluation
- Evaluation runner
evals/evaluate.py - Golden dataset
evals/golden_dataset.json - Evaluation history
evals/eval_log.csv - Observability log
evals/observability_log.csv
Dashboards
- Evaluation dashboard
app/ui.py - Observability dashboard
app/observability.py
Connections 7
- Evaluation runner → Golden dataset: loads cases
evals/evaluate.py - Evaluation runner → Message classifier: classifies messages
evals/evaluate.py - Evaluation runner → Evaluation history: reads history
evals/evaluate.py - Evaluation runner → Evaluation history: appends results
evals/evaluate.py - Evaluation runner → Observability log: appends metrics
evals/evaluate.py - Evaluation dashboard → Evaluation history: reads history
app/ui.py - Observability dashboard → Observability log: reads metrics
app/observability.py