[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7D2W8E5Q4B41EDZP9TF3NNR":3,"skill-same:01M7D2W8E5Q4B41EDZP9TF3NNR":64},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":38,"commitSha":39,"skillDirectory":40,"forksCount":13,"repositoryHtmlUrl":41,"repositoryHomepageUrl":6,"ownerHtmlUrl":42,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":43,"install":44,"resources":57},"01M7D2W8E5Q4B41EDZP9TF3NNR","presenting-conference-talks",null,"Generates conference presentation slides (Beamer LaTeX PDF and editable PPTX) from a compiled paper with speaker notes and talk script. Use when preparing oral talks, spotlight presentations, or invited talks for ML and systems conferences.","beyondai\u002Fsandbox","beyondai","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F2845958?v=4","01M7D24SBNM47BJGJ9J2HB13BP","01M7D24SCDC9DYE005XH6R553S",0,"MIT","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.codex\u002Fskills\u002Fpresenting-conference-talks\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: presenting-conference-talks\ndescription: Generates conference presentation slides (Beamer LaTeX PDF and editable PPTX) from a compiled paper with speaker notes and talk script. Use when preparing oral talks, spotlight presentations, or invited talks for ML and systems conferences.\nversion: 1.0.0\nauthor: Orchestra Research\nlicense: MIT\ntags: [Presenting Conference Talks, Beamer, PPTX, Slides, Speaker Notes, OSDI, SOSP, ASPLOS, NeurIPS, ICML]\ndependencies: [python-pptx>=0.6.21]\n---\n\n# Presenting Conference Talks: From Paper to Slides\n\nGenerate conference presentation slides from a compiled research paper. Produces both **Beamer LaTeX PDF** (for polished typesetting) and **editable PPTX** (for last-minute adjustments), with speaker notes and an optional talk script.\n\n## When to Use This Skill\n\n| Scenario | Use This Skill | Use Other Skills Instead |\n|----------|---------------|--------------------------|\n| Preparing oral\u002Fspotlight\u002Fposter-talk slides | ✅ | |\n| Generating Beamer PDF + PPTX from paper | ✅ | |\n| Speaker notes and talk script | ✅ | |\n| Writing the paper itself | | ml-paper-writing |\n| Structuring a systems paper | | systems-paper-writing |\n| Creating publication-quality plots | | academic-plotting |\n\n**Attribution**: This skill's structure draws inspiration from the ARIS paper-slides skill (570 lines, supporting poster\u002Fspotlight\u002Foral\u002Finvited with Beamer+PPTX). This is an independent implementation for the AI-Research-SKILLs ecosystem.\n\n---\n\n## Talk Types and Slide Counts\n\n| Talk Type | Duration | Slides | Content Depth |\n|-----------|----------|--------|---------------|\n| poster-talk | 3–5 min | 5–8 | Problem + key result only |\n| spotlight | 5–8 min | 8–12 | Problem + approach + key results |\n| oral | 15–20 min | 15–22 | Full story with evaluation highlights |\n| invited | 30–45 min | 25–40 | Deep dive with context and demos |\n\n**Rule of thumb**: ~1 slide per minute for oral, ~1.5 slides per minute for spotlight.\n\n---\n\n## Slide Structure Templates\n\n### Poster-Talk (5–8 slides)\n\n```text\nSlide 1: Title + Authors + Affiliation\nSlide 2: Problem — Why this matters (1 motivating figure)\nSlide 3: Key Insight — One-sentence thesis\nSlide 4: Approach Overview — Architecture diagram\nSlide 5: Main Result — Headline numbers (1 figure)\nSlide 6: Takeaway + QR code to paper\u002Fcode\n```\n\n### Spotlight (8–12 slides)\n\n```text\nSlide 1:  Title + Authors\nSlide 2:  Problem Statement — Concrete, quantified\nSlide 3:  Motivation — Why existing solutions fall short\nSlide 4:  Key Insight — Thesis statement\nSlide 5:  System Overview — Architecture diagram\nSlide 6:  Design Highlight 1 — Core mechanism\nSlide 7:  Design Highlight 2 — Key innovation\nSlide 8:  Evaluation Setup — Baselines and workloads (brief)\nSlide 9:  Main Results — Headline performance figure\nSlide 10: Ablation \u002F Breakdown — What contributes most\nSlide 11: Summary + Contributions\nSlide 12: Thank You + Links\n```\n\n### Oral (15–22 slides)\n\n```text\nSlide 1:  Title + Authors + Venue\nSlide 2:  Outline (optional — \"roadmap\" slide)\nSlide 3:  Problem Context — Domain importance\nSlide 4:  Problem Statement — Specific challenge\nSlide 5:  Motivation — Gaps in existing systems\nSlide 6:  Key Insight — Thesis\nSlide 7:  System Overview — Architecture diagram\nSlide 8:  Design Component 1 — Detailed walkthrough\nSlide 9:  Design Component 2 — Detailed walkthrough\nSlide 10: Design Component 3 — Detailed walkthrough\nSlide 11: Design Alternatives — Why not other approaches\nSlide 12: Implementation — Key engineering highlights\nSlide 13: Evaluation Setup — Testbed, baselines, metrics\nSlide 14: End-to-End Results — Main performance\nSlide 15: Result Deep Dive — Breakdown or per-workload\nSlide 16: Ablation Study — Component contributions\nSlide 17: Scalability — Scaling behavior\nSlide 18: Demo Slide (systems talks) — Screenshot or recording\nSlide 19: Related Work — Positioning (brief)\nSlide 20: Summary — Contributions restated\nSlide 21: Future Work — Open questions\nSlide 22: Thank You + Paper Link + QR Code\n```\n\n### Invited Talk (25–40 slides)\nExtends the oral structure with:\n- Additional context slides (field overview, historical progression)\n- Multiple demo\u002Fwalkthrough slides\n- Deeper evaluation analysis\n- Broader implications and future directions\n- Q&A preparation slides (hidden, for backup)\n\n---\n\n## Systems Talk Specifics\n\nSystems conference talks have unique requirements compared to ML talks:\n\n### Demo Slide\n- Include a **live demo** or **pre-recorded screencast** of the system in action\n- Always have a **recorded backup** — live demos fail at the worst times\n- Show the system under realistic load, not toy examples\n\n### Architecture Walkthrough\n- Animate the architecture diagram: highlight components as you explain them\n- Use Beamer `\\only\u003CN>` or `\\onslide\u003CN>` for progressive reveal\n- Walk through a **concrete request** end-to-end through the system\n\n### Evaluation Highlights\n- Select 2–3 strongest figures from the paper\n- Annotate figures on slides (arrows, circles highlighting key points)\n- State the takeaway **before** showing the figure (\"Our system is 2x faster — here's the data\")\n\n---\n\n## Speaker Notes Guidelines\n\n### Structure per Slide\n```text\n[Timing: X minutes]\n[Key point to convey]\n[Transition sentence to next slide]\n```\n\n### Mike Dahlin's Layered Approach\nApply \"Say what you're going to say, say it, then say what you said\" at three levels:\n\n1. **Talk level**: Outline slide → body → summary slide\n2. **Section level**: Section heading → content slides → section takeaway\n3. **Slide level**: Headline statement → supporting evidence → transition\n\n### Timing Guidelines\n- Poster-talk: 30–60 sec per slide\n- Spotlight: 30–45 sec per slide\n- Oral: 45–90 sec per slide\n- Invited: 60–120 sec per slide\n\n---\n\n## Output Formats\n\n### Beamer LaTeX → PDF\n\nAdvantages: Professional typesetting, math support, version control friendly.\n\n```latex\n\\documentclass[aspectratio=169]{beamer}\n\\usetheme{metropolis}  % Clean, modern theme\n\\usepackage{appendixnumberbeamer}\n\n\\title{Your Paper Title}\n\\subtitle{Venue Year}\n\\author{Author 1 \\and Author 2}\n\\institute{Institution}\n\\date{}\n\n\\begin{document}\n\\maketitle\n\n\\begin{frame}{Problem}\n  \\begin{itemize}\n    \\item Key problem statement\n    \\item Concrete motivation with numbers\n  \\end{itemize}\n  \\note{Speaker note: Start with the big picture...}\n\\end{frame}\n\n% ... more frames ...\n\\end{document}\n```\n\n### python-pptx → Editable PPTX\n\nAdvantages: Easy last-minute edits, corporate template compatibility, animations.\n\n```python\nfrom pptx import Presentation\nfrom pptx.util import Inches, Pt\nfrom pptx.enum.text import PP_ALIGN\n\nprs = Presentation()\nprs.slide_width = Inches(13.333)  # 16:9\nprs.slide_height = Inches(7.5)\n\n# Title slide\nslide = prs.slides.add_slide(prs.slide_layouts[0])\nslide.shapes.title.text = \"Your Paper Title\"\nslide.placeholders[1].text = \"Author 1, Author 2\\nVenue Year\"\n\n# Content slide\nslide = prs.slides.add_slide(prs.slide_layouts[1])\nslide.shapes.title.text = \"Problem Statement\"\nbody = slide.placeholders[1]\nbody.text = \"Key point 1\\nKey point 2\"\n\n# Add speaker notes\nnotes_slide = slide.notes_slide\nnotes_slide.notes_text_frame.text = \"Speaker note: explain the motivation...\"\n\nprs.save(\"talk.pptx\")\n```\n\n---\n\n## Color Scheme Suggestions\n\n> These are aesthetic suggestions, not official venue requirements. Adjust freely.\n\n| Venue Type | Primary | Accent | Background |\n|-----------|---------|--------|------------|\n| USENIX (OSDI\u002FNSDI) | Dark Blue (#003366) | Red (#CC0000) | White |\n| ACM (SOSP\u002FASPLOS) | ACM Blue (#0071BC) | Dark Gray (#333333) | White |\n| NeurIPS | Purple (#7B2D8E) | Gold (#F0AD00) | White |\n| ICML | Teal (#008080) | Orange (#FF6600) | White |\n| Generic | Dark Gray (#333333) | Blue (#0066CC) | White |\n\n---\n\n## Workflow\n\n### Step 1: Content Extraction\n```text\n- Read the compiled paper (PDF or LaTeX source)\n- Identify: thesis, contributions, architecture figure, key eval figures\n- Note the talk type and duration\n```\n\n### Step 2: Outline Generation\n```text\n- Select the appropriate slide structure template (above)\n- Map paper sections to slide groups\n- Allocate time per slide group\n```\n\n### Step 3: Slide-by-Slide Generation\n```text\n- Generate Beamer source slide by slide\n- Add speaker notes per slide\n- Include figures from paper (copy to slides\u002F directory)\n- Generate python-pptx script for PPTX version\n```\n\n### Step 4: Review and Polish\n```text\n- Check total slide count matches talk duration\n- Verify all figures are readable at presentation resolution\n- Run Beamer compilation: latexmk -pdf slides.tex\n- Run PPTX generation: python3 generate_slides.py\n- Review speaker notes for timing and transitions\n```\n\n### Quick Checklist\n- [ ] Slide count appropriate for talk type\u002Fduration\n- [ ] Title slide has correct authors, affiliations, venue\n- [ ] Architecture diagram included and clearly labeled\n- [ ] Key eval figures annotated with takeaways\n- [ ] Speaker notes include timing markers\n- [ ] Transitions between sections are smooth\n- [ ] Demo slide has recorded backup\n- [ ] Thank-you slide includes paper link \u002F QR code\n- [ ] Font sizes ≥ 24pt for readability from back of room\n- [ ] Consistent color scheme throughout\n\n---\n\n## Common Issues and Solutions\n\n| Issue | Solution |\n|-------|----------|\n| Too many slides for time limit | Cut details, keep one figure per point |\n| Slides feel like paper paragraphs | Use bullet points (≤ 6 per slide), let figures tell the story |\n| Audience lost during design section | Add architecture walkthrough with progressive reveal |\n| Evaluation slides overwhelming | Show 2–3 strongest figures, put rest in backup slides |\n| Speaker notes too long | Target 3–4 sentences per slide, focus on transitions |\n| Beamer compilation fails | Check figure paths, use `\\graphicspath{{figures\u002F}}` |\n| PPTX looks different from Beamer | Adjust python-pptx font sizes and margins manually |\n\n---\n\n## References\n\n- [references\u002Fslide-templates.md](references\u002Fslide-templates.md) — Complete Beamer template code and python-pptx generation script\n- Mike Dahlin, \"Giving a Conference Talk\" — https:\u002F\u002Fwww.cs.utexas.edu\u002F~dahlin\u002Fprofessional\u002FgoodTalk.pdf\n","# Presenting Conference Talks: From Paper to Slides\n\nGenerate conference presentation slides from a compiled research paper. Produces both **Beamer LaTeX PDF** (for polished typesetting) and **editable PPTX** (for last-minute adjustments), with speaker notes and an optional talk script.\n\n## When to Use This Skill\n\n| Scenario | Use This Skill | Use Other Skills Instead |\n|----------|---------------|--------------------------|\n| Preparing oral\u002Fspotlight\u002Fposter-talk slides | ✅ | |\n| Generating Beamer PDF + PPTX from paper | ✅ | |\n| Speaker notes and talk script | ✅ | |\n| Writing the paper itself | | ml-paper-writing |\n| Structuring a systems paper | | systems-paper-writing |\n| Creating publication-quality plots | | academic-plotting |\n\n**Attribution**: This skill's structure draws inspiration from the ARIS paper-slides skill (570 lines, supporting poster\u002Fspotlight\u002Foral\u002Finvited with Beamer+PPTX). This is an independent implementation for the AI-Research-SKILLs ecosystem.\n\n---\n\n## Talk Types and Slide Counts\n\n| Talk Type | Duration | Slides | Content Depth |\n|-----------|----------|--------|---------------|\n| poster-talk | 3–5 min | 5–8 | Problem + key result only |\n| spotlight | 5–8 min | 8–12 | Problem + approach + key results |\n| oral | 15–20 min | 15–22 | Full story with evaluation highlights |\n| invited | 30–45 min | 25–40 | Deep dive with context and demos |\n\n**Rule of thumb**: ~1 slide per minute for oral, ~1.5 slides per minute for spotlight.\n\n---\n\n## Slide Structure Templates\n\n### Poster-Talk (5–8 slides)\n\n```text\nSlide 1: Title + Authors + Affiliation\nSlide 2: Problem — Why this matters (1 motivating figure)\nSlide 3: Key Insight — One-sentence thesis\nSlide 4: Approach Overview — Architecture diagram\nSlide 5: Main Result — Headline numbers (1 figure)\nSlide 6: Takeaway + QR code to paper\u002Fcode\n```\n\n### Spotlight (8–12 slides)\n\n```text\nSlide 1:  Title + Authors\nSlide 2:  Problem Statement — Concrete, quantified\nSlide 3:  Motivation — Why existing solutions fall short\nSlide 4:  Key Insight — Thesis statement\nSlide 5:  System Overview — Architecture diagram\nSlide 6:  Design Highlight 1 — Core mechanism\nSlide 7:  Design Highlight 2 — Key innovation\nSlide 8:  Evaluation Setup — Baselines and workloads (brief)\nSlide 9:  Main Results — Headline performance figure\nSlide 10: Ablation \u002F Breakdown — What contributes most\nSlide 11: Summary + Contributions\nSlide 12: Thank You + Links\n```\n\n### Oral (15–22 slides)\n\n```text\nSlide 1:  Title + Authors + Venue\nSlide 2:  Outline (optional — \"roadmap\" slide)\nSlide 3:  Problem Context — Domain importance\nSlide 4:  Problem Statement — Specific challenge\nSlide 5:  Motivation — Gaps in existing systems\nSlide 6:  Key Insight — Thesis\nSlide 7:  System Overview — Architecture diagram\nSlide 8:  Design Component 1 — Detailed walkthrough\nSlide 9:  Design Component 2 — Detailed walkthrough\nSlide 10: Design Component 3 — Detailed walkthrough\nSlide 11: Design Alternatives — Why not other approaches\nSlide 12: Implementation — Key engineering highlights\nSlide 13: Evaluation Setup — Testbed, baselines, metrics\nSlide 14: End-to-End Results — Main performance\nSlide 15: Result Deep Dive — Breakdown or per-workload\nSlide 16: Ablation Study — Component contributions\nSlide 17: Scalability — Scaling behavior\nSlide 18: Demo Slide (systems talks) — Screenshot or recording\nSlide 19: Related Work — Positioning (brief)\nSlide 20: Summary — Contributions restated\nSlide 21: Future Work — Open questions\nSlide 22: Thank You + Paper Link + QR Code\n```\n\n### Invited Talk (25–40 slides)\nExtends the oral structure with:\n- Additional context slides (field overview, historical progression)\n- Multiple demo\u002Fwalkthrough slides\n- Deeper evaluation analysis\n- Broader implications and future directions\n- Q&A preparation slides (hidden, for backup)\n\n---\n\n## Systems Talk Specifics\n\nSystems conference talks have unique requirements compared to ML talks:\n\n### Demo Slide\n- Include a **live demo** or **pre-recorded screencast** of the system in action\n- Always have a **recorded backup** — live demos fail at the worst times\n- Show the system under realistic load, not toy examples\n\n### Architecture Walkthrough\n- Animate the architecture diagram: highlight components as you explain them\n- Use Beamer `\\only\u003CN>` or `\\onslide\u003CN>` for progressive reveal\n- Walk through a **concrete request** end-to-end through the system\n\n### Evaluation Highlights\n- Select 2–3 strongest figures from the paper\n- Annotate figures on slides (arrows, circles highlighting key points)\n- State the takeaway **before** showing the figure (\"Our system is 2x faster — here's the data\")\n\n---\n\n## Speaker Notes Guidelines\n\n### Structure per Slide\n```text\n[Timing: X minutes]\n[Key point to convey]\n[Transition sentence to next slide]\n```\n\n### Mike Dahlin's Layered Approach\nApply \"Say what you're going to say, say it, then say what you said\" at three levels:\n\n1. **Talk level**: Outline slide → body → summary slide\n2. **Section level**: Section heading → content slides → section takeaway\n3. **Slide level**: Headline statement → supporting evidence → transition\n\n### Timing Guidelines\n- Poster-talk: 30–60 sec per slide\n- Spotlight: 30–45 sec per slide\n- Oral: 45–90 sec per slide\n- Invited: 60–120 sec per slide\n\n---\n\n## Output Formats\n\n### Beamer LaTeX → PDF\n\nAdvantages: Professional typesetting, math support, version control friendly.\n\n```latex\n\\documentclass[aspectratio=169]{beamer}\n\\usetheme{metropolis}  % Clean, modern theme\n\\usepackage{appendixnumberbeamer}\n\n\\title{Your Paper Title}\n\\subtitle{Venue Year}\n\\author{Author 1 \\and Author 2}\n\\institute{Institution}\n\\date{}\n\n\\begin{document}\n\\maketitle\n\n\\begin{frame}{Problem}\n  \\begin{itemize}\n    \\item Key problem statement\n    \\item Concrete motivation with numbers\n  \\end{itemize}\n  \\note{Speaker note: Start with the big picture...}\n\\end{frame}\n\n% ... more frames ...\n\\end{document}\n```\n\n### python-pptx → Editable PPTX\n\nAdvantages: Easy last-minute edits, corporate template compatibility, animations.\n\n```python\nfrom pptx import Presentation\nfrom pptx.util import Inches, Pt\nfrom pptx.enum.text import PP_ALIGN\n\nprs = Presentation()\nprs.slide_width = Inches(13.333)  # 16:9\nprs.slide_height = Inches(7.5)\n\n# Title slide\nslide = prs.slides.add_slide(prs.slide_layouts[0])\nslide.shapes.title.text = \"Your Paper Title\"\nslide.placeholders[1].text = \"Author 1, Author 2\\nVenue Year\"\n\n# Content slide\nslide = prs.slides.add_slide(prs.slide_layouts[1])\nslide.shapes.title.text = \"Problem Statement\"\nbody = slide.placeholders[1]\nbody.text = \"Key point 1\\nKey point 2\"\n\n# Add speaker notes\nnotes_slide = slide.notes_slide\nnotes_slide.notes_text_frame.text = \"Speaker note: explain the motivation...\"\n\nprs.save(\"talk.pptx\")\n```\n\n---\n\n## Color Scheme Suggestions\n\n> These are aesthetic suggestions, not official venue requirements. Adjust freely.\n\n| Venue Type | Primary | Accent | Background |\n|-----------|---------|--------|------------|\n| USENIX (OSDI\u002FNSDI) | Dark Blue (#003366) | Red (#CC0000) | White |\n| ACM (SOSP\u002FASPLOS) | ACM Blue (#0071BC) | Dark Gray (#333333) | White |\n| NeurIPS | Purple (#7B2D8E) | Gold (#F0AD00) | White |\n| ICML | Teal (#008080) | Orange (#FF6600) | White |\n| Generic | Dark Gray (#333333) | Blue (#0066CC) | White |\n\n---\n\n## Workflow\n\n### Step 1: Content Extraction\n```text\n- Read the compiled paper (PDF or LaTeX source)\n- Identify: thesis, contributions, architecture figure, key eval figures\n- Note the talk type and duration\n```\n\n### Step 2: Outline Generation\n```text\n- Select the appropriate slide structure template (above)\n- Map paper sections to slide groups\n- Allocate time per slide group\n```\n\n### Step 3: Slide-by-Slide Generation\n```text\n- Generate Beamer source slide by slide\n- Add speaker notes per slide\n- Include figures from paper (copy to slides\u002F directory)\n- Generate python-pptx script for PPTX version\n```\n\n### Step 4: Review and Polish\n```text\n- Check total slide count matches talk duration\n- Verify all figures are readable at presentation resolution\n- Run Beamer compilation: latexmk -pdf slides.tex\n- Run PPTX generation: python3 generate_slides.py\n- Review speaker notes for timing and transitions\n```\n\n### Quick Checklist\n- [ ] Slide count appropriate for talk type\u002Fduration\n- [ ] Title slide has correct authors, affiliations, venue\n- [ ] Architecture diagram included and clearly labeled\n- [ ] Key eval figures annotated with takeaways\n- [ ] Speaker notes include timing markers\n- [ ] Transitions between sections are smooth\n- [ ] Demo slide has recorded backup\n- [ ] Thank-you slide includes paper link \u002F QR code\n- [ ] Font sizes ≥ 24pt for readability from back of room\n- [ ] Consistent color scheme throughout\n\n---\n\n## Common Issues and Solutions\n\n| Issue | Solution |\n|-------|----------|\n| Too many slides for time limit | Cut details, keep one figure per point |\n| Slides feel like paper paragraphs | Use bullet points (≤ 6 per slide), let figures tell the story |\n| Audience lost during design section | Add architecture walkthrough with progressive reveal |\n| Evaluation slides overwhelming | Show 2–3 strongest figures, put rest in backup slides |\n| Speaker notes too long | Target 3–4 sentences per slide, focus on transitions |\n| Beamer compilation fails | Check figure paths, use `\\graphicspath{{figures\u002F}}` |\n| PPTX looks different from Beamer | Adjust python-pptx font sizes and margins manually |\n\n---\n\n## References\n\n- [references\u002Fslide-templates.md](references\u002Fslide-templates.md) — Complete Beamer template code and python-pptx generation script\n- Mike Dahlin, \"Giving a Conference Talk\" — https:\u002F\u002Fwww.cs.utexas.edu\u002F~dahlin\u002Fprofessional\u002FgoodTalk.pdf\n",{"name":5,"tags":23,"author":34,"license":14,"version":35,"description":7,"dependencies":36},[24,25,26,27,28,29,30,31,32,33],"Presenting Conference Talks","Beamer","PPTX","Slides","Speaker Notes","OSDI","SOSP","ASPLOS","NeurIPS","ICML","Orchestra Research","1.0.0",[37],"python-pptx>=0.6.21","main","46e03791269608c5ac886a64c08f5d94147e8786","labs\u002F.codex\u002Fskills\u002Fpresenting-conference-talks","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox","https:\u002F\u002Fgithub.com\u002Fbeyondai","2026-11-08 04:18:42",{"source":8,"skillName":5,"commands":45,"prompt":55,"zipAvailable":56},[46,49,52],{"tool":47,"command":48},"npx","npx skills add beyondai\u002Fsandbox --skill presenting-conference-talks",{"tool":50,"command":51},"bunx","bunx skills add beyondai\u002Fsandbox --skill presenting-conference-talks",{"tool":53,"command":54},"pnpm","pnpm dlx skills add beyondai\u002Fsandbox --skill presenting-conference-talks","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.codex\u002Fskills\u002Fpresenting-conference-talks\u002FSKILL.md",true,[58],{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Fslide-templates.md","REFERENCE","参考文档",15279,"https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.codex\u002Fskills\u002Fpresenting-conference-talks\u002Freferences\u002Fslide-templates.md",[65,72,79,86,93,100],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":18},"01M7D3GMPGJZRG8EPY404YF8TR","weights-and-biases","Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Fweights-and-biases\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M7D3GFZH440V2J532BSZMPVP","serving-llms-vllm","Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency\u002Fthroughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ\u002FAWQ\u002FFP8), and tensor parallelism.","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Fvllm\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M7D3GAE8MQCPGF72EP9G9GKY","verl-rl-training","Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Fverl\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M7D3G7KKNE5RGNYJNXV4F0MM","unsloth","Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA\u002FQLoRA optimization","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Funsloth\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":18},"01M7D3G2RCZQ08GS0KAW5CBWSC","fine-tuning-with-trl","Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO\u002FGRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Ftrl-fine-tuning\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":18},"01M7D3FX08BPB172CYGM597NZF","torchforge-rl-training","Provides guidance for PyTorch-native agentic RL using torchforge, Meta's library separating infra from algorithms. Use when you want clean RL abstractions, easy algorithm experimentation, or scalable training with Monarch and TorchTitan.","https:\u002F\u002Fgithub.com\u002Fbeyondai\u002Fsandbox\u002Fblob\u002Fmain\u002Flabs\u002F.opencode\u002Fskills\u002Ftorchforge\u002FSKILL.md",[],[]]