🔬 Dashboard — Experiment Status

Placeholder Data

Results marked TBD will be populated after PEC2 experiments complete. Replace each TBD with actual CodeCarbon + statistical output.

---

All Experiments

TABLE experiment-id AS "ID", title AS "Experiment", status AS "Status", methodology AS "Method", tool AS "Tool", date-started AS "Started", date-completed AS "Done"
FROM #experiment
SORT date-started DESC

---

By Status

TABLE WITHOUT ID status AS "Status", length(rows) AS "Count", rows.file.link AS "Experiments"
FROM #experiment
GROUP BY status

---

Pending Experiments

TABLE experiment-id AS "ID", title AS "Experiment", methodology AS "Method"
FROM #experiment
WHERE status = "planned"
SORT file.ctime ASC

---

Stage 1: Triage Results Placeholder

Model	Strategy	F1-macro	Precision	Recall	p-value	r	gCO₂eq
Heuristic baseline	—	TBD	TBD	TBD	—	—	0
TF-IDF + SVM	—	TBD	TBD	TBD	—	—	0
Llama 3.2 8B	Zero-Shot	TBD	TBD	TBD	TBD	TBD	TBD
Llama 3.2 8B	Few-Shot-3	TBD	TBD	TBD	TBD	TBD	TBD
Llama 3.2 8B	Few-Shot-6	TBD	TBD	TBD	TBD	TBD	TBD
Llama 3.2 8B	CoT	TBD	TBD	TBD	TBD	TBD	TBD
Mistral 7B	Zero-Shot	TBD	TBD	TBD	TBD	TBD	TBD
Mistral 7B	Few-Shot-3	TBD	TBD	TBD	TBD	TBD	TBD
Mistral 7B	Few-Shot-6	TBD	TBD	TBD	TBD	TBD	TBD
Mistral 7B	CoT	TBD	TBD	TBD	TBD	TBD	TBD

Update when PEC2 experiments complete. Replace TBD with actual values.

---

Stage 2: Estimation Results Placeholder

Model	Strategy	MAE (SP)	RMSE	vs. Mean Hist.	vs. Deep-SE	vs. CoGEE
Mean historical	—	TBD	TBD	baseline	—	—
Deep-SE	—	~3.2	—	—	baseline	—
CoGEE (GPT-4)	—	~1.9	—	—	—	baseline
Llama 3.2 8B	Zero-Shot	TBD	TBD	TBD	TBD	TBD
Llama 3.2 8B	Few-Shot-3	TBD	TBD	TBD	TBD	TBD
Mistral 7B	CoT	TBD	TBD	TBD	TBD	TBD

---

Carbon Summary

Phase	Conditions	Total gCO₂eq	Per condition	Per issue
Stage 1	8 LLM + 2 baseline	TBD	TBD	TBD
Stage 2	TBD	TBD	TBD	TBD

Fill with CodeCarbon output after experiments.