Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add guliqianxun/research-skills --skill torch-model-designgit clone --depth 1 https://github.com/guliqianxun/research-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/guliqianxun/research-skills/torch-model-design)<a href="https://agentmods.dev/skills/guliqianxun/research-skills/torch-model-design"><img src="https://agentmods.dev/badge/skills/guliqianxun/research-skills/torch-model-design/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/guliqianxun/research-skills/torch-model-design"><img src="https://agentmods.dev/badge/skills/guliqianxun/research-skills/torch-model-design.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00205 | $0.02184 |
| Opus 5 | $0.00102 | $0.01092 |
| Sonnet 5 | $0.00041 | $0.00437 |
| Haiku 4.5 | $0.00020 | $0.00218 |
Grade A, and why
torch-model-design scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 192 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PyTorch Model Design
Implementation-focused skill for ML model development in PyTorch — architecture design, profiling, distributed training, and production inference.
How to Use This Skill
Six reference files cover distinct concerns. Read only what the current task needs.
| Reference | When to Read |
|---|---|
| references/principles.md | Start here. Engineering principles: computation graph, memory model, precision, training stability, parallelism, inference vs training, multimodal conventions, arithmetic intensity / roofline model, data loading at scale, reproducibility, loss function patterns |
| references/architecture.md | Implementing a model: attention variants (MHA/GQA/MQA/sliding window), FFN variants (SwiGLU/MoE), position encoding (RoPE), block assembly, initialization, gradient debugging |
| references/profiling-optimization.md | Making it fast: MFU, FLOPs counting, memory analysis, AMP (bf16/fp16/fp8), FlashAttention, torch.compile, gradient checkpointing, sequence packing, benchmarking |
| references/distributed-training.md | Multi-GPU/multi-node: DDP, FSDP2, tensor parallel, pipeline parallel, DeviceMesh, 2D/3D parallelism, context parallel, distributed checkpointing |
| references/multimodal-temporal.md | Multimodal and temporal models: modality tokenization (text/image/audio/video), cross-attention vs interleaved, modality alignment, temporal patterns, causal masking, streaming inference, memory budgeting |
| references/inference-deployment.md | Production inference: KV cache, quantization (INT8/FP8/GPTQ/AWQ via torchao), speculative decoding, continuous batching, torch.export |
Phase 1: Architecture Design
Decision: Graph Mode
Does the model have data-dependent control flow?
(early exit, variable-length loops, conditional layers)
├─ YES → Eager mode. torch.compile(dynamic=True) may work
│ but test carefully — graph breaks eliminate most benefit.
└─ NO ↓
Research iteration mode (architecture changes frequently)?
├─ YES → Start eager. Compile when architecture stabilizes.
│ Compiled model debugging is significantly harder.
└─ NO ↓
Production / throughput-critical?
└─ YES → torch.compile(mode="max-autotune").
Invest in eliminating graph breaks first.
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 192 lines · 205 tokens per session scan A 3a08a1e0146d
torch-model-design is a skill published in the GitHub repository guliqianxun/research-skills (5 stars, last pushed 4mo ago), licensed MIT. It adds 205 tokens to every session and 2,184 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora and retrieve grounded contexts using the Google GenAI SDK. Use when listing RAG corpora or files, inspecting a corpus, retrieving contexts, or generating content grounded in a RAG corpus. Do not use for standard database queries (use SQL/Spanner skills), Google…
agent-platform-model-registry
Agent Platform Model Registry Management. Use when you need to upload, list, describe, update, or delete machine learning models (and their versions) in the Agent Platform Model Registry. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform models.
foundry-config-setup
Resolve missing setup caused by a hardcoded Foundry project endpoint or model in a sample. Use when a sample fails because it uses a placeholder/hardcoded projectendpoint (for example "https://your-project.services.ai.azure.com") or a hardcoded model instead of reading them from the environment.
google-cloud-solution-agentic-analytics-spark-knowledge-catalog
Discovers requirements and generates guidance to design and deploy a governed, secure agentic-analytics solution for data that's distributed across Google Cloud, other cloud providers, or on-premises. Data that's outside Google Cloud (such as data from Databricks, Snowflake, Salesforce, SAP, or Oracle systems) is…
training-check
Interactively monitor training metrics from the current Codex session, periodically checking WandB or fallback logs for NaN, divergence, plateaus, and broken runs.
nemo-automodel-launcher-config
Configure NeMo AutoModel job launches for interactive runs, Slurm clusters, and SkyPilot cloud execution.