Challenge Creation Cron #19
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Challenge Creation Cron | |
| on: | |
| schedule: | |
| - cron: "0 2 * * 2,4,6" # Monday, Wednesday, Friday at 6 PM PST | |
| workflow_dispatch: {} # Allow manual trigger | |
| jobs: | |
| create-challenge: | |
| runs-on: ubuntu-latest | |
| timeout-minutes: 30 | |
| permissions: | |
| contents: write | |
| pull-requests: write | |
| issues: write | |
| id-token: write | |
| actions: read | |
| steps: | |
| - name: Checkout repository | |
| uses: actions/checkout@v4 | |
| with: | |
| fetch-depth: 0 | |
| - name: Set up Python | |
| uses: actions/setup-python@v5 | |
| with: | |
| python-version: "3.12" | |
| - name: Install dependencies | |
| run: | | |
| pip install pre-commit requests websocket-client | |
| pre-commit install | |
| - name: Fetch open PRs | |
| id: open-prs | |
| env: | |
| GH_TOKEN: ${{ github.token }} | |
| run: | | |
| { | |
| echo "open_prs<<OPEN_PRS_EOF" | |
| gh pr list --state open --json number,title,headRefName --template '{{range .}}PR #{{.number}}: {{.title}} (branch: {{.headRefName}}){{"\n"}}{{end}}' | |
| echo "OPEN_PRS_EOF" | |
| } >> "$GITHUB_OUTPUT" | |
| - name: Run Claude Code | |
| id: claude | |
| uses: anthropics/claude-code-action@v1 | |
| env: | |
| SERVICE_URL: ${{ secrets.LEETGPU_SERVICE_URL }} | |
| LEETGPU_API_KEY: ${{ secrets.LEETGPU_API_KEY }} | |
| with: | |
| claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }} | |
| claude_args: "--model claude-sonnet-4-6" | |
| settings: | | |
| { | |
| "permissions": { | |
| "allow": ["Bash", "Read", "Write", "Edit", "Glob", "Grep", "WebFetch", "WebSearch", "Task"] | |
| } | |
| } | |
| prompt: | | |
| Read CLAUDE.md and follow the Creation Workflow to create and validate (with `run_challenge.py`) a new LeetGPU coding challenge, then open a PR. | |
| Before writing any code, think hard about what would make a genuinely interesting GPU programming challenge. List existing challenges (ls challenges/*/), study what's already covered, and brainstorm problems that would teach people something new about GPU programming. | |
| IMPORTANT — CHECK FOR CONFLICTS WITH OPEN PRs: | |
| The following open PRs already exist. You MUST avoid picking a challenge number or topic that conflicts with any of these: | |
| ${{ steps.open-prs.outputs.open_prs }} | |
| Pick the next available challenge number that is NOT already taken by a merged challenge or an open PR. Also avoid creating a challenge on the same topic as any pending PR, even if the number differs. | |
| THEME — REAL-WORLD INFERENCE KERNELS: | |
| Focus on challenges inspired by real-world ML inference workloads. Think about the building blocks of modern neural networks (transformers, diffusion models, LLMs, vision models) and the GPU kernels that make them fast. Good examples: | |
| - Fused inference kernels: fused SwiGLU/GeGLU MLP blocks, flash attention, paged attention, speculative decoding verification, quantized matmul (INT8/INT4), fused QKV projection, KV-cache updates | |
| - Sequence/token operations: top-k/top-p sampling, beam search step, KV-cache rotation, causal masking | |
| - Model architecture blocks: full transformer decoder block (like the existing GPT-2 challenge), mixture-of-experts routing, LoRA forward pass | |
| - Online/streaming algorithms: online softmax, streaming attention (process new queries without storing entire rows), continuous batching, prefix caching | |
| Look at `challenges/medium/74_gpt2_block/` as the gold standard for this style of challenge. The solver should implement a meaningful, self-contained inference building block — not a toy operation. | |
| HARD RULES: | |
| - Do NOT create trivial element-wise challenges. We have way too many (sigmoid, relu, silu, clipping, etc). If your idea is just "apply f(x) to every element", pick something else. | |
| - Do NOT duplicate existing challenges — check both the merged challenges in the repo AND the open PRs listed above. | |
| - Do NOT use a challenge number that is already claimed by an open PR. | |
| - Prefer medium or hard difficulty. Only create easy if the topic involves a non-trivial GPU concept (not just a map operation). | |
| - The challenge should require the solver to think about memory access patterns, synchronization, work distribution, or algorithm design — not just write a one-line kernel. | |
| Before opening the PR, verify every item in the "Checklist" section of CLAUDE.md. Run `pre-commit run --all-files` to lint. |