ReddRadar
Agent skill
hf-job-submit
Submit training jobs to Hugging Face cloud GPUs
Stars
163
Forks
31
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/hf-job-submit
SKILL.md
HF Job Submit
Submit a Python training script to run on Hugging Face's cloud GPU infrastructure.
How to Use
bash
$CLAUDE_CONFIG_DIR/skills/hf-job-submit/scripts/hf-job-submit.sh <script_path> <hardware> <packages>
Arguments
| Argument | Description | Example |
|---|---|---|
script_path |
Path to Python training script | /home/user/workspace/train.py |
hardware |
GPU hardware type | t4-small, a10g-large |
packages |
Space-separated pip packages | "trl transformers datasets" |
Hardware Options
| Type | GPU | VRAM | Best For |
|---|---|---|---|
t4-small |
T4 | 16GB | Models < 1B |
t4-medium |
T4 | 16GB | Models 1-3B |
a10g-small |
A10G | 24GB | Models 1-3B |
a10g-large |
A10G | 24GB | Models 3-7B (LoRA) |
a100-large |
A100 | 40GB | Models 3-7B (LoRA) |
Prerequisites
HF_TOKENenvironment variable must be set- Hugging Face Pro or Team account with GPU Jobs access
Output
Returns the Job ID on success, which can be used with hf-job-status and hf-job-logs.
Example
bash
$CLAUDE_CONFIG_DIR/skills/hf-job-submit/scripts/hf-job-submit.sh \
"/home/user/workspace/train.py" \
"t4-small" \
"trl transformers datasets peft accelerate"
Didn't find tool you were looking for?