Queue pilot_step → Loading…
Step status
Readiness checks on the pilot roster. P1–P3 are offline (no LLM). P4 applies model inclusion from OPS-supplied outcomes; P5 checks OpenRouter routes.
| Step | What | Latest | Logged at | Summary |
|---|
OPS jobs
Queued Track 1 / OpenRouter work outside P1–P5.
| Job | What | Latest | When | Summary |
|---|
Next OPS step
Loading…
Delete pilot AI runs
Wipe exploratory run_model artefacts for the pilot roster so you can
re-enqueue. Keeps Track-1 metrics, oracle_modules, and
pilot_log. Admin only.
Pilot analysis
Exploratory Track-1 skim on the roster (not full-corpus variance_check).
Loading…
First indication
Completeness
| Roster | Measured | Setup failed | Pending | Missing from DB | Complete |
|---|
Short variance & correlation
| Metric | n | mean | std | CV |
|---|
| A | B | ρ | |ρ| |
|---|
AI response / behaviour
| Runs | Instances | Models | Resolved | Mean F2P | Tokens in/out |
|---|
| Model | n | Resolved | Rate | Mean F2P | Flagged | Tokens |
|---|
| Failure flag | n | Resolved |
|---|
Benchmark runs on the pilot roster (compact). Use the Details menu to open response, applied diff, runtime meta, or tokens/prompt under the row.
| id | instance_id | model_id | rep | resolved | f2p | failure_flag | created_at | Details |
|---|
Raw metrics (Track 1)
| Instance | Repo | Lang | mc | mcoh | ih | sc | cc | afs | afl |
|---|
Pilot roster
| Instance | Language | Dataset | Rationale |
|---|
Models (P4 flags)
| Model | Pinned version | Route | Resolved rate | Truncation OK | Included |
|---|
pilot_log
| When | Step | Instance | Used for | Result |
|---|