Skip to content

feat: document --agent-type CLI flag and agent swapping mechanism - #2793

Merged
marta-sd merged 9 commits into
mainfrom
martas/agent-selector-docs
Sep 4, 2026
Merged

feat: document --agent-type CLI flag and agent swapping mechanism#2793
marta-sd merged 9 commits into
mainfrom
martas/agent-selector-docs

Conversation

@marta-sd

@marta-sd marta-sd commented Aug 26, 2026

Copy link
Copy Markdown
Contributor

This PR documents the new --agent CLI flag and agent swapping mechanism. It should be merge as the last as it documents features that are not yet in themain branch.

The entire PR stack will introduce the following changes:

  1. config composition
  2. --agent-type flag in the CLI for selecting the agent
  3. safeguard preventing unsupported env <> agent pairings
  4. aligning with dataset changes introduced in epic epic: dataset preparation #1338
  5. docs update (this PR)

@copy-pr-bot

copy-pr-bot Bot commented Aug 26, 2026

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@github-actions

github-actions Bot commented Aug 26, 2026

Copy link
Copy Markdown
Contributor

@marta-sd
marta-sd requested review from lbliii and sephmard August 26, 2026 13:29
@marta-sd marta-sd linked an issue Aug 27, 2026 that may be closed by this pull request
sephmard
sephmard previously approved these changes Aug 27, 2026

@sephmard sephmard left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approving. I also think that we need to include an update in the gym/evaluation tab. I'd suggest that this is its own new subtab under Evaluate: Harness.

@marta-sd
marta-sd dismissed sephmard’s stale review August 28, 2026 09:40

The merge-base changed after approval.

@marta-sd marta-sd changed the title martas/agent selector docs feat: document --agent-type CLI flag and agent swapping mechanism Aug 28, 2026
@marta-sd
marta-sd force-pushed the martas/agent-selector-docs branch from 9ffc9b8 to 5e8ecfc Compare August 28, 2026 10:29
@copy-pr-bot

copy-pr-bot Bot commented Aug 28, 2026

Copy link
Copy Markdown

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

@marta-sd

Copy link
Copy Markdown
Contributor Author

Approving. I also think that we need to include an update in the gym/evaluation tab. I'd suggest that this is its own new subtab under Evaluate: Harness.

@sephmard I added a dedicated page in 3a4c2b3

@github-actions github-actions Bot added the sla:review-overdue Review response is over the one-business-day SLA label Aug 28, 2026
Glorf
Glorf previously approved these changes Aug 28, 2026

@sephmard sephmard left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This page is about agentic evaluation, but "The composed instance is renamed" example (gpqa_mcqa_simple_agent/hermes_agent) isn't agentic — gpqa / gsm8k uses the mcqa resources server, which is plain multiple-choice verification, no tools.

Please swap it for working agentic example — the intro's own terminal_bench_2_1/opencode + terminus_2_agent composition works directly here.

Comment on lines +100 to +101
# Run GPQA with the Hermes harness instead of the one its config names
gym eval run --agent-type hermes_agent --benchmark gpqa --model-type vllm_model

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

GPQA isnt the best example, as its non-agentic

Comment thread fern/versions/latest/pages/evaluation/harness.mdx Outdated
Comment thread fern/versions/latest/pages/evaluation/harness.mdx Outdated
Comment on lines +32 to +46
## The composed instance is renamed

Composition replaces the environment's agent and renames the server instance after the harness that
runs it, so metrics and `gym env status` report what actually ran. Composing `hermes_agent` onto
`gpqa_mcqa_simple_agent` produces `gpqa_mcqa_hermes_agent`.

That name is what you use to route rollouts when the servers are already up:

```bash
gym env start --agent-type hermes_agent --benchmark gpqa --model-type vllm_model
gym eval run --no-serve --agent gpqa_mcqa_hermes_agent --input rows.jsonl
```

End-to-end rollouts collection (without the `--no-serve` flag) keeps working — rows stamped with the
pre-swap instance are re-routed for you under the hood.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

"The composed instance is renamed" example (gpqa_mcqa_simple_agent/hermes_agent) isn't agentic (plain MCQA, no tools); suggest swapping to the intro's own terminal_bench_2_1/opencode + terminus_2_agent composition, both for accuracy-to-theme and to cut the page down to one running example instead of three.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Replacing it requires introducing changes to the configs - otherwise the composed name was completely unreadable. This change requires approval from eng team (ideally @bxyu-nvidia) because it touches configs that are in the critical path.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, thanks for the context

@marta-sd

marta-sd commented Sep 3, 2026

Copy link
Copy Markdown
Contributor Author

This page is about agentic evaluation, but "The composed instance is renamed" example (gpqa_mcqa_simple_agent/hermes_agent) isn't agentic — gpqa / gsm8k uses the mcqa resources server, which is plain multiple-choice verification, no tools.

Please swap it for working agentic example — the intro's own terminal_bench_2_1/opencode + terminus_2_agent composition works directly here.

Due to what's in configs the name would be "opencode_sandboxed_agent_terminal_bench_2_1_terminus_2_sandboxed_agent". I need to update configs for this to make sense.

marta-sd and others added 9 commits September 3, 2026 09:40
Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
Co-authored-by: Seph Mard <seph.mard@gmail.com>
Signed-off-by: Marta Stepniewska-Dziubinska <marta-sd@users.noreply.github.com>
Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
Co-authored-by: Seph Mard <seph.mard@gmail.com>
Signed-off-by: Marta Stepniewska-Dziubinska <marta-sd@users.noreply.github.com>
Server instances are named <benchmark>_<implementation>, but the sandbox
benchmarks led with the harness. That also broke the composed instance
name: the rename strips the agent type only as a suffix, so swapping a
harness in appended the new type instead of replacing the old one.

Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>
The rename and pairing examples used gpqa/gsm8k, which score multiple
choice with no tools, on a page about agentic evaluation. All the sample
output is now copied from real runs of the intro's own swap.

Signed-off-by: Marta Stepniewska-Dziubinska <martas@nvidia.com>

@sephmard sephmard left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

approving docs changes - @bxyu-nvidia reviewed the config changes

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

r0.6.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge. sla:review-overdue Review response is over the one-business-day SLA

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Add docs for the --agent-type CLI flag

5 participants