Coming from Neptune?
GoodSeed makes it easy to transition from Neptune. Whether you want to browse your existing runs, permanently export your data, or swap out the Neptune SDK entirely — we've got you covered.
View your Neptune runs #
Curious what GoodSeed feels like? Connect your Neptune project and browse your existing runs in the GoodSeed viewer — no data migration needed. Your API token stays in your browser cookies and is never sent to our servers.
Export data from Neptune #
Permanently migrate your Neptune runs into local GoodSeed storage using neptune-exporter. For full documentation, see the Neptune migration guide.
1. Install neptune-exporter
git clone https://github.com/neptune-ai/neptune-exporter
cd neptune-exporter
uv sync --extra goodseedCopied!2. Export your data to disk
Set your Neptune API token via NEPTUNE_API_TOKEN or the --api-token flag.
uv run neptune-exporter export \
-p "workspace/project" \
--exporter neptune2 \
--data-path ./exports/dataCopied!Neptune 2.x also supports --runs-query for NQL filtering. See the exporter docs for details.
uv run neptune-exporter export \
-p "workspace/project" \
--exporter neptune3 \
--data-path ./exports/dataCopied!3. Load into GoodSeed
uv run neptune-exporter load \
--loader goodseed \
--data-path ./exports/dataCopied!Your runs are now in ~/.goodseed. View them with goodseed serve.
| Option | Description |
|---|---|
--goodseed-project | Override project name (default: Neptune project ID) |
--goodseed-home | Custom data directory (default: ~/.goodseed) |
--step-multiplier | Scale decimal Neptune steps to integers (e.g. 1000) |
Package interface mapping #
GoodSeed provides drop-in replacements for the Neptune Python SDK. Change one import line and your existing training scripts work with GoodSeed — no other code changes needed.
import goodseed.neptune as neptune
run = neptune.init_run(name="baseline", mode="async")
run["params/lr"] = 0.001
run["train/loss"].log(0.5, step=1)
run["sys/tags"].add("production")
run.stop()Copied!import goodseed.neptune_scale as neptune_scale
run = neptune_scale.Run(project="workspace/project", run_id="my-run")
run.log_metrics({"loss": 0.5}, step=1)
run.log_configs({"lr": 0.001})
run.add_tags(["baseline"])
run.close()Copied!