Skip to content

搜索技能.../

Agent Skill Search Engine

搜索

搜索
分类
职业

关于

关于
隐私政策
服务条款

© 2026 Skills Pool. 保留所有权利。

Hf Job Ops Playbook | Skills Pool

技能档案

Hf Job Ops Playbook

Operate Hugging Face Jobs reliably for training and evaluation in this repository, with reproducible submission records, monitoring, and retry flow. Use when submitting, monitoring, or triaging HF Jobs.

Haruk1y0 星标2026年3月1日

职业
分类: 实验室工具

技能内容

Use this skill for end-to-end Hugging Face Jobs operations in Mistral_Hackathon.

When To Use

User asks to submit training or eval jobs on Hugging Face.
User asks for job status, logs, failure triage, or rerun.
You need reproducible submission ledgers and job ids.

Preflight

Authenticate CLI:

hf auth whoami

Confirm required secrets are set:

HF_TOKEN
WANDB_API_KEY (if W&B logging is enabled)
MISTRAL_API_KEY (if Mistral API fallback is used by eval)

Run dry-run first:

SFT: node scripts/hf/submit_sft_job.mjs
Eval: node scripts/hf/submit_eval_job.mjs

Submit Workflow

相关技能

快速安装

Hf Job Ops Playbook

npx skillvault add Haruk1y/haruk1y-mistral-hackathon-skills-public-hf-job-ops-playbook-skill-md

下载 Skill 打开源码仓库

作者: Haruk1y
星标: 0
更新时间: 2026年3月1日
职业

本页内容

01When To Use 02

Submit SFT job.

HF_JOB_SUBMIT=true node scripts/hf/submit_sft_job.mjs
Record is appended to artifacts/hf_jobs/submissions.jsonl.

Submit eval job.

HF_EVAL_JOB_SUBMIT=true node scripts/hf/submit_eval_job.mjs
Record is appended to artifacts/hf_jobs/eval_submissions.jsonl.

Capture job id from CLI and ledger.

If id parse fails, keep stdout_tail and stderr_tail for manual lookup.

Monitoring Workflow

Use HF CLI directly:

Running jobs: hf jobs ps --namespace Haruk1y
All jobs: hf jobs ps -a --namespace Haruk1y
Inspect: hf jobs inspect <job_id>
Logs: hf jobs logs -f <job_id>
Usage stats: hf jobs stats <job_id>

Failure Triage

Submission failure:

Check hf auth whoami.
Verify required secrets and env vars.
Re-run dry-run command and inspect built command string.

Runtime failure:

Inspect logs and classify as data, dependency, OOM, or auth issue.
Apply smallest config change needed and re-submit.

Post-run artifact missing:

Confirm push_to_hub and output model id settings.
Confirm local metrics files exist in output dir.

Repository Mapping

SFT submission: scripts/hf/submit_sft_job.mjs
Eval submission: scripts/hf/submit_eval_job.mjs
Training script: scripts/hf/train_sft_request_to_hidden_lm.py
Eval runner: scripts/wandb/weave_eval_runner.py
Submission ledgers: artifacts/hf_jobs/submissions.jsonl, artifacts/hf_jobs/eval_submissions.jsonl

References

references/job-ops-checklist.md

Preflight

03Submit Workflow

04Monitoring Workflow

05Failure Triage

06Repository Mapping

数据科学家

实验室工具

Automation Audit Ops

Evidence-first automation inventory and overlap audit workflow for ECC. Use when the user wants to know which jobs, hooks, connectors, MCP servers, or wrappers are live, broken, redundant, or missing before fixing anything.

实验室工具

Github Qa Labels

Label GitHub issues and PRs found during QA testing. Use when organizing QA findings with proper labels.

storybookjs89.7k

实验室工具

Jupyter Notebook

Gamitin kapag humihiling ang gumagamit na gumawa, mag-scaffold, o mag-edit ng mga Jupyter notebook (`.ipynb`) para sa mga eksperimento, eksplorasyon, o tutorial; mas piliin ang mga naka-bundle na template at patakbuhin ang helper script na `new_notebook.py` upang makabuo ng malinis na panimulang notebook.

实验室工具

Tidb Integrationtest Recorder

Use when recording TiDB integration tests under tests/integrationtest and verifying regenerated result files stay minimal and correct.

实验室工具

Quality Nonconformance

Codified expertise for quality control, non-conformance investigation, root cause analysis, corrective action, and supplier quality management in regulated manufacturing.

实验室工具

Hugging Face Trackio

Track ML experiments with Trackio using Python logging, alerts, and CLI metric retrieval.

数据科学家