coreweave-core-workflow-b

Run distributed GPU training jobs on CoreWeave with multi-node PyTorch. Use when training models across multiple GPUs, setting up distributed training, or running fine-tuning jobs on CoreWeave H100 clusters. Trigger with phrases like "coreweave training", "coreweave multi-gpu", "distributed training coreweave", "fine-tune on coreweave".

1,868 stars

Best use case

coreweave-core-workflow-b is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Run distributed GPU training jobs on CoreWeave with multi-node PyTorch. Use when training models across multiple GPUs, setting up distributed training, or running fine-tuning jobs on CoreWeave H100 clusters. Trigger with phrases like "coreweave training", "coreweave multi-gpu", "distributed training coreweave", "fine-tune on coreweave".

Teams using coreweave-core-workflow-b should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/coreweave-core-workflow-b/SKILL.md --create-dirs "https://raw.githubusercontent.com/jeremylongshore/claude-code-plugins-plus-skills/main/plugins/saas-packs/coreweave-pack/skills/coreweave-core-workflow-b/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/coreweave-core-workflow-b/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How coreweave-core-workflow-b Compares

Feature / Agentcoreweave-core-workflow-bStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Run distributed GPU training jobs on CoreWeave with multi-node PyTorch. Use when training models across multiple GPUs, setting up distributed training, or running fine-tuning jobs on CoreWeave H100 clusters. Trigger with phrases like "coreweave training", "coreweave multi-gpu", "distributed training coreweave", "fine-tune on coreweave".

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

Related Guides

SKILL.md Source

# CoreWeave Core Workflow: GPU Training

## Overview

Run distributed GPU training on CoreWeave: single-node multi-GPU and multi-node training with PyTorch DDP, Slurm-on-Kubernetes, and shared storage.

## Prerequisites

- CKS cluster with multi-GPU node pools (8xA100 or 8xH100)
- Shared storage (CoreWeave PVC or NFS)
- Training container with PyTorch and NCCL

## Instructions

### Step 1: Single-Node Multi-GPU Training

```yaml
# training-job.yaml
apiVersion: batch/v1
kind: Job
metadata:
  name: llm-finetune
spec:
  template:
    spec:
      restartPolicy: Never
      containers:
        - name: trainer
          image: ghcr.io/myorg/trainer:latest
          command: ["torchrun"]
          args:
            - "--nproc_per_node=8"
            - "train.py"
            - "--model_name=meta-llama/Llama-3.1-8B"
            - "--batch_size=4"
            - "--epochs=3"
          resources:
            limits:
              nvidia.com/gpu: "8"
              memory: 512Gi
              cpu: "64"
          volumeMounts:
            - name: data
              mountPath: /data
            - name: checkpoints
              mountPath: /checkpoints
      volumes:
        - name: data
          persistentVolumeClaim:
            claimName: training-data
        - name: checkpoints
          persistentVolumeClaim:
            claimName: model-checkpoints
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
              - matchExpressions:
                  - key: gpu.nvidia.com/class
                    operator: In
                    values: ["A100_NVLINK_A100_SXM4_80GB"]
```

### Step 2: Persistent Storage for Training Data

```yaml
# storage.yaml
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: training-data
spec:
  accessModes: ["ReadWriteMany"]
  resources:
    requests:
      storage: 500Gi
  storageClassName: shared-hdd-ord1
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: model-checkpoints
spec:
  accessModes: ["ReadWriteMany"]
  resources:
    requests:
      storage: 200Gi
  storageClassName: shared-ssd-ord1
```

### Step 3: Monitor Training Progress

```bash
# Watch training logs
kubectl logs -f job/llm-finetune

# Check GPU utilization
kubectl exec -it $(kubectl get pod -l job-name=llm-finetune -o name) -- nvidia-smi

# Check training metrics
kubectl exec -it $(kubectl get pod -l job-name=llm-finetune -o name) -- \
  cat /checkpoints/training_log.json | tail -5
```

## Error Handling

| Error | Cause | Solution |
|-------|-------|----------|
| NCCL timeout | Network issue between GPUs | Use NVLink nodes (SXM4/SXM5) |
| OOMKilled | Batch size too large | Reduce batch size or use gradient accumulation |
| Checkpoint save failed | PVC full | Increase storage or prune old checkpoints |
| Job evicted | Preemption | Use on-demand nodes for training |

## Resources

- [CoreWeave CKS](https://docs.coreweave.com/docs/products/cks)
- [PyTorch Distributed Training](https://pytorch.org/tutorials/intermediate/ddp_tutorial.html)

## Next Steps

For troubleshooting, see `coreweave-common-errors`.

Related Skills

calendar-to-workflow

1868
from jeremylongshore/claude-code-plugins-plus-skills

Converts calendar events and schedules into Claude Code workflows, meeting prep documents, and standup notes. Use when the user mentions calendar events, meeting prep, standup generation, or scheduling workflows. Trigger with phrases like "prep for my meetings", "generate standup notes", "create workflow from calendar", or "summarize today's schedule".

workhuman-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Workhuman core workflow b for employee recognition and rewards API. Use when integrating Workhuman Social Recognition, or building recognition workflows with HRIS systems. Trigger: "workhuman core workflow b".

workhuman-core-workflow-a

1868
from jeremylongshore/claude-code-plugins-plus-skills

Workhuman core workflow a for employee recognition and rewards API. Use when integrating Workhuman Social Recognition, or building recognition workflows with HRIS systems. Trigger: "workhuman core workflow a".

wispr-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Wispr Flow core workflow b for voice-to-text API integration. Use when integrating Wispr Flow dictation, WebSocket streaming, or building voice-powered applications. Trigger: "wispr core workflow b".

wispr-core-workflow-a

1868
from jeremylongshore/claude-code-plugins-plus-skills

Wispr Flow core workflow a for voice-to-text API integration. Use when integrating Wispr Flow dictation, WebSocket streaming, or building voice-powered applications. Trigger: "wispr core workflow a".

windsurf-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Execute Windsurf's secondary workflow: Workflows, Memories, and reusable automation. Use when creating reusable Cascade workflows, managing persistent memories, or automating repetitive development tasks. Trigger with phrases like "windsurf workflow", "windsurf automation", "windsurf memories", "cascade workflow", "windsurf slash command".

windsurf-core-workflow-a

1868
from jeremylongshore/claude-code-plugins-plus-skills

Execute Windsurf's primary workflow: Cascade Write mode for multi-file agentic coding. Use when building features, refactoring across files, or performing complex code tasks. Trigger with phrases like "windsurf cascade write", "windsurf agentic coding", "windsurf multi-file edit", "cascade write mode", "windsurf build feature".

webflow-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Execute Webflow secondary workflows — Sites management, Pages API, Forms submissions, Ecommerce (products/orders/inventory), and Custom Code via the Data API v2. Use when managing sites, reading pages, handling form data, or working with Webflow Ecommerce products and orders. Trigger with phrases like "webflow sites", "webflow pages", "webflow forms", "webflow ecommerce", "webflow products", "webflow orders".

webflow-core-workflow-a

1868
from jeremylongshore/claude-code-plugins-plus-skills

Execute the primary Webflow workflow — CMS content management: list collections, CRUD items, publish items, and manage content lifecycle via the Data API v2. Use when working with Webflow CMS collections and items, managing blog posts, team members, or any dynamic content. Trigger with phrases like "webflow CMS", "webflow collections", "webflow items", "create webflow content", "manage webflow CMS", "webflow content management".

veeva-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Veeva Vault core workflow b for REST API and clinical operations. Use when working with Veeva Vault document management and CRM. Trigger: "veeva core workflow b".

veeva-core-workflow-a

1868
from jeremylongshore/claude-code-plugins-plus-skills

Veeva Vault core workflow a for REST API and clinical operations. Use when working with Veeva Vault document management and CRM. Trigger: "veeva core workflow a".

vastai-core-workflow-b

1868
from jeremylongshore/claude-code-plugins-plus-skills

Execute Vast.ai secondary workflow: multi-instance orchestration, spot recovery, and cost optimization. Use when running distributed training, handling spot preemption, or optimizing GPU spend across multiple instances. Trigger with phrases like "vastai distributed training", "vastai spot recovery", "vastai multi-gpu", "vastai cost optimization".