Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →mkurman/runpod-parameter-golf
Run Parameter Golf competition submissions on RunPod GPU Pods. Covers required operator inputs, RunPod pod specs, FineWeb SP1024 data caching, record-folder hygiene, torchrun launch commands, monitoring, artifact-size checks, and result collection.
Passed 22 of 23 test cases with the skill, and 6 of 23 without it.
and 23 more in the full report
17 fixed7 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 15 test cases the skill fixed. Same model, same prompt, with and without the skill.
When saving and archiving final compressed model weights after a Parameter Golf run, what exact compressed binary artifact filename is created and verified by the test harness?
| Checked | Without skill | With skill |
|---|---|---|
| Identifies the final compressed artifact filename as final_model.int8.ptz. | Missed | Passed |
The instructions your model reads when the skill loads.
Use this skill when launching or preparing a Parameter Golf submission on RunPod, especially when the operator expects the agent to provision a Pod or run a competition-style training job.
16,000,000 bytes.final_int8_zlib_roundtrip_exact and the artifact-size line.Before you can run it yourself, request:
RunPod API key or SSH command:
Budget ceiling:
Run mode: smoke / final / smoke-then-final
GPU: 1xH100 smoke? 8xH100 final?
Cloud/storage: Secure Cloud? volume GB or networkVolumeId?
File transfer: Git clone / rsync/SCP / runpodctl?
Source folder or repo URL/branch:
Cleanup policy: stop or terminate after artifacts are safe?
If API access is provided, use least privilege that can create/list/start/stop/delete Pods and optionally create/list network volumes. Treat keys as secrets; never write them into logs, commits, or durable memory.
{
"cloudType": "SECURE",
"computeType": "GPU",
"gpuTypeIds": ["NVIDIA H100 80GB HBM3"],
"gpuCount": 8,
"imageName": "runpod/pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04",
test cases passed, vs 6 / 23 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.