Skip to content

Commit 3c987c4

Browse files
committed
docs(data-designer): address Knepper review on submit-only seeds/providers
Clarify that local file/DataFrame seeds remain library-valid but not on the platform path, and that remote validate/submit is IGW-provider-only. Signed-off-by: Matt Kornfield <mkornfield@nvidia.com>
1 parent 1dd1c19 commit 3c987c4

2 files changed

Lines changed: 9 additions & 8 deletions

File tree

docs/data-designer/tutorials/seeding.mdx

Lines changed: 7 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -11,15 +11,16 @@ This tutorial demonstrates how to use external datasets as seed data for synthet
1111

1212
For more detail about seed dataset behavior, see the [open-source library's version](https://docs.nvidia.com/nemo/datadesigner/v0.9.1/tutorials/seeding-with-an-external-dataset) of this tutorial.
1313

14-
## Seed Sources by Execution Mode
14+
## Seed Sources on the Platform
1515

16-
Seed source support depends on platform resource availability:
16+
This tutorial covers platform execution (`nemo data-designer … submit` and the SDK). Seed support there is narrower than the open-source Data Designer library:
1717

18-
| Seed source | CLI `submit` / SDK today | Use case |
19-
|-------------|--------------------------|----------|
20-
| **Local files or DataFrames** | Not supported | Upload local files to Filesets first. |
21-
| **HuggingFace** | Supported | Publicly available datasets or private HuggingFace datasets. |
18+
| Seed source | Platform (`submit` / SDK) | Notes |
19+
|-------------|---------------------------|-------|
20+
| **Local files or DataFrames** | Not supported on the platform path | Still valid in the [open-source library](https://docs.nvidia.com/nemo/datadesigner/v0.9.1/tutorials/seeding-with-an-external-dataset). For platform jobs, upload files to a Fileset first (see Step 1 below). |
21+
| **HuggingFace** | Supported | Public or private HuggingFace datasets (private needs a Secrets API token). |
2222
| **Files API Filesets** | Supported | Shared seed data stored through the Files API. |
23+
2324
### HuggingFace Datasets
2425

2526
Use `HuggingFaceSeedSource` to load data from HuggingFace:

plugins/nemo-data-designer/README.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -27,8 +27,8 @@ The exit code is `0` only when every requested context validates cleanly. JSON o
2727

2828
### Local vs. remote
2929

30-
- **Local** checks library-level compatibility: the engine compiles the config and resolves model providers. Providers can be defined locally **or** referenced by name from the Inference Gateway — both are first-class.
31-
- **Remote** mirrors what `nemo data-designer <preview|create> submit` accepts: unsupported seed types and `tool_configs` are rejected, IGW providers are resolved against the platform, Files-service seeds are looked up, and Nemotron Personas filesets are checked. The remote pass is a client-side simulation of those checks; it does not contact the data-designer service.
30+
- **Local** (`--execution-context local`) checks open-source library compatibility: the engine compiles the config and resolves model providers the library understands (including locally defined providers). This is a library/config check, not a platform execution path — there is no `nemo data-designer … run` verb.
31+
- **Remote** (`--execution-context remote`) mirrors what `nemo data-designer <preview|create> submit` accepts on the platform: local seed types and `tool_configs` are rejected, **only Inference Gateway providers** are accepted, Files-service seeds are looked up, and Nemotron Personas filesets are checked. The remote pass is a client-side simulation of those checks; it does not contact the data-designer service.
3232

3333
### Programmatic use
3434

0 commit comments

Comments
 (0)