Improve concepts docs and add docs for agent and resources server#796
Merged
bxyu-nvidia merged 10 commits intomainfrom Mar 2, 2026
Merged
Improve concepts docs and add docs for agent and resources server#796bxyu-nvidia merged 10 commits intomainfrom
bxyu-nvidia merged 10 commits intomainfrom
Conversation
cwing-nvidia
requested review from
bxyu-nvidia,
heatherlxd,
hwolff99 and
shashank3959
bxyu-nvidia
previously approved these changes
Mar 2, 2026
heatherlxd
reviewed
Mar 2, 2026
heatherlxd
reviewed
Mar 2, 2026
heatherlxd
reviewed
Mar 2, 2026
heatherlxd
reviewed
Mar 2, 2026
…arity Add new Training Approaches concept page explaining SFT vs RL, DPO, GRPO, and the rise of RLVR. Restructure About overview with concrete environment challenges and how NeMo Gym addresses each. Add "What is an Environment?" intro to Core Components (renamed to Environment Components). Reorder component presentation to Agent > Model > Resources to follow rollout flow. Signed-off-by: Chris Wing <cwing@nvidia.com>
Create dedicated Agent Server page with rollout lifecycle pseudocode, agent integration patterns (SimpleAgent, MiniSWEAgent), and tools guidance. Streamline the Agent tab in Environment Components to a concise overview linking to the new page. Add Agent Server to the landing page grid and sidebar navigation. Signed-off-by: Chris Wing <cwing@nvidia.com>
Make explicit in both Environment Components and Agent Server pages that the Agent is orchestration code which delegates all text generation to the Model server. Addresses user confusion from #773. Signed-off-by: Chris Wing <cwing@nvidia.com>
Replace inline implementation list and config path with a concise description and link to the dedicated Model Server page, matching the pattern used by the Agent tab. Signed-off-by: Chris Wing <cwing@nvidia.com>
Create dedicated Resources Server overview page with pseudocode, session management, tool implementations, verification logic, and examples (workplace_assistant, math_with_code). Streamline the Resources tab in Environment Components to match Agent/Model tabs. Add Resources Server card and toctree entry to docs index. Signed-off-by: Chris Wing <cwing@nvidia.com>
…onents Remove tab-set, merge component descriptions into linked bullets, and replace the architecture SVG with an ASCII diagram showing the rollout lifecycle across Agent, Model, and Resources servers. Signed-off-by: Chris Wing <cwing@nvidia.com>
Add seealso linking to Model Server Fields in the configuration reference, matching the pattern on Agent and Resources Server pages. Signed-off-by: Chris Wing <cwing@nvidia.com>
Add introductory paragraphs explaining the Model server's role and middleware purpose, consistent with Agent and Resources Server page structure. Signed-off-by: Chris Wing <cwing@nvidia.com>
Add rollout definition in GRPO context on training approaches page, link What's Next to training tutorials, and add tip on training tutorials page linking back to training approaches concepts. Signed-off-by: Chris Wing <cwing@nvidia.com>
Remove the What's Next section from training-approaches since the sidebar toctree already handles navigation between concept pages. Minor wording fixes in core-components and about/index. Signed-off-by: Chris Wing <cwing@nvidia.com>
cwing-nvidia
force-pushed
the
cwing/concepts-improvements
branch
from
4bb20d3 to
a46a83b
Compare
4 tasks
4 tasks
bxyu-nvidia
approved these changes
Mar 2, 2026
jsw-zorro
pushed a commit
to niletron/Gym
that referenced
this pull request
Apr 7, 2026
…IDIA-NeMo#796) ## Summary - Add Training Approaches page explaining SFT vs RL, DPO, GRPO, and RLVR - Rename "Core Components" to "Environment Components" with a "What is an Environment?" introduction and updated architecture diagram - Create dedicated Agent Server, Resources Server pages with pseudocode and examples - Improve Model Server page with conceptual intro and consistent seealso links to server config - Streamline Environment Components page: replace tab-set with linked bullets, replace SVG with ASCII diagram showing rollout lifecycle across servers - Improve About overview: replace motivation with more concrete challenges, and more concrete details on how NeMo Gym addresses - Add cross-links between training approaches, training tutorials, and rollout collection - Introduce rollout concept in GRPO section with glossary term reference ## Related Issues - Addresses NVIDIA-NeMo#773 (Model vs Agent confusion) - Partially addresses NVIDIA-NeMo#774 (rollout clarity and training cross-links) ## Test Plan - [ ] Build docs locally and verify all new pages render correctly - [ ] Verify all `{doc}`, `{ref}`, and `{term}` cross-links resolve - [ ] Check environment diagram renders properly in code block --------- Signed-off-by: Chris Wing <cwing@nvidia.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Related Issues
Test Plan
{doc},{ref}, and{term}cross-links resolve