# 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]** |