TeamIDE-methodology/templates/ResearchStarter/results/TEMPLATE-results.md

Experiment N Results: [Short Title]

Phase [A/B/C...] — [Phase Name]

Setup

• Architecture: [Model/system description]
• Source: experiments/run1-descriptive-name.ext
• Config: [Key hyperparameters or settings]

---

Summary of All Runs

Run	Key Variable	Config	Best Metric	Key Finding
1	baseline	...	...	[One-line takeaway]
2	changed X	...	...	[One-line takeaway]
3	changed Y	...	...	[One-line takeaway]

---

Detailed Findings

Finding 1: [Descriptive Title]

Runs [N-M] ([what was varied])

[Explanation of what happened and why. Include raw output.]

[Raw output from the run — config, per-epoch metrics, final results]

Verdict: [One-line conclusion. Bold the action item.]

Finding 2: [Next Discovery]

...

---

Current Best Result

Run N: [Short description of the winning config]

[Full output of the best run]

---

Conclusions

[What the phase proved or disproved. Structural insights, not just numbers.]

1. [Finding]: [Explanation]
2. [Finding]: [Explanation]
3. [Finding]: [Explanation]

---

Next Steps

Priority order:

1. Run N: [Description] — DONE. [Result summary.]
2. Run N+1: [Description] — Next up. [Why this is the next thing to try.]
3. Run N+2: [Description] — [Brief rationale.]

Decision points:

• >[Target] achieved: Proceed to [next phase]
• Stuck at [range]: [What that means and what to try]
• Below [floor] after remaining runs: [Kill gate — abandon or pivot]

Literature Context

Method	Result	Notes
[Existing work]	[Their result]	[How it compares]
Us (Run N)	[Our result]	[Where we stand]