NVIDIA-NeMo/Gym is a library and infrastructure for evaluating and improving models and agents inside environments, where each environment defines tasks, agent interaction, verification, and execution state. It is for teams running reproducible evaluations or training at scale across settings such as code execution, tool calling, and sandboxes, and the catalogue entries provide skills and instructions for working with it.
Borrowing it
Nothing to install: this file belongs to NVIDIA-NeMo/Gym. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/NVIDIA-NeMo/Gym/main/.agents/skills/nemo-gym-pivot-datasets/SKILL.mdgit clone --depth 1 https://github.com/NVIDIA-NeMo/GymWrote 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/nvidia-nemo/gym/nemo-gym-pivot-datasets)<a href="https://agentmods.dev/skills/nvidia-nemo/gym/nemo-gym-pivot-datasets"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/gym/nemo-gym-pivot-datasets/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/nvidia-nemo/gym/nemo-gym-pivot-datasets"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/gym/nemo-gym-pivot-datasets.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00097 | $0.02076 |
| Opus 5 | $0.00048 | $0.01038 |
| Sonnet 5 | $0.00019 | $0.00415 |
| Haiku 4.5 | $0.00010 | $0.00208 |
Grade A, and why
nemo-gym-pivot-datasets 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 11d 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 — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Nemo Gym Pivot Datasets
Paper Reference
This skill operationalizes PivotRL: create local single-step pivot datasets from successful trajectories, prefer informative mixed-reward states, and train with verifier-based local rewards rather than exact trajectory imitation.
Invocation Check
Use this skill when the task is to turn existing agent trajectories or rollout artifacts into a Nemo Gym pivot dataset, or to validate whether a pivot JSONL/config pair can be used for single-step local RL or evaluation.
Before writing a converter, inspect representative source rows and the target resource server. Do not assume the source field names are the contract. Convert by reconstructing the semantic pieces needed by Gym's Responses-style row format.
Core Workflow
- Inspect the source data shape and count the candidate assistant decision points.
- Reconstruct model calls from the source's flattened output list, and cross-check the count against whatever per-trajectory model-call count the source records. A decision point is one whole model call, so this step decides what a row even is.
- Identify the semantic fields needed for each pivot:
- model-call input context before the pivot action
- available tools at that decision point
- expected assistant action
- reward/verifier target if it is separate from the demonstrated action
- optional provenance such as task id, source trajectory id, rollout id, uuid, depth, and original metadata
- Convert each accepted decision point into one pivot JSONL row.
- Generate or update the matching Gym config so the pivot-format JSONL can be used directly.
- Validate with the bundled validator and, when available, the target Gym resource-server models.
- Write metrics that make skipped rows, action types, tool names, depth, and provenance coverage easy to inspect, including the parallel / single / chat split and a batch-size histogram.
Row Shape
Read references/row-contract.md when implementing or reviewing a
converter. For single_step_tool_use_with_argument_comparison, the essential row fields are:
What ships with it
13 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.
- agents/openai.yaml 232 B
- references/config-training-and-agent-ref.md 8.0 KB
- references/conversion-patterns.md 5.8 KB
- references/rollout-artifact-pitfalls.md 9.2 KB
- references/row-contract.md 5.4 KB
- scripts/audit_pivot_turn_boundaries.py 22 KB runs code
- scripts/reference/chat_messages_to_pivot_dataset_reference.py 22 KB runs code
- scripts/reference/conversational_messages_to_pivot_dataset_reference.py 25 KB runs code
- scripts/reference/generic_pivot_dataset_reference.py 7.2 KB runs code
- scripts/reference/responses_output_to_pivot_dataset_reference.py 20 KB runs code
- scripts/reference/tool_messages_to_pivot_dataset_reference.py 27 KB runs code
- scripts/validate_pivot_dataset_sharded.py 16 KB runs code
- scripts/validate_pivot_dataset.py 15 KB runs code
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.
- 11d ago First seen · 177 lines · 97 tokens per session scan A 21144d656ebc
nemo-gym-pivot-datasets is a skill published in the GitHub repository NVIDIA-NeMo/Gym (1,181 stars, last pushed today), licensed Apache-2.0. It adds 97 tokens to every session and 2,076 once invoked, about $0.0005 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 2026-08-30.
Other skills, from other repositories
model_finetuning
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
model_finetuning
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
spark-training-gotchas
Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.
thinking-out-loud
A contract for what the agent does when a long, messy, stream-of-consciousness ramble arrives (usually voice dictation): act on nothing until the echo brief is approved. The echo audits the entire transfer, mission, locked decisions and constraints, open questions, flips and parked tangents, with the model's…
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.
hugging-face-model-trainer
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV…