docs: label early access features and hide ones not ready to document (#4855)

Labels early access features and hides features we aren't documenting
yet.

**Early access label:** Functions (4 pages, plus the GPU refresh,
checkpointing and adaptive execution pages), Secrets (3 pages), Tiered
Index Cache, Blob V2, and merge insert, which is labelled for Enterprise
only. Two sidebar titles are shorter so the tag fits.

**Hidden:** branching, JSON columns and geospatial.
- The JSON functions page and the agent branch-experiments tutorial go
in `.mintignore`.
- `tables/branching` stays reachable but out of the nav, as #4859 left
it, because SDK error messages link to it.
- Other pages' mentions of these features become draft comments, so
they're easy to restore.

**Secrets page:** drops references to manual KMS setup, which the
control plane handles. It keeps the encryption-at-rest facts, states the
Azure value-size limit inline, and points a 501 error to support.
This commit is contained in:
Brendan Clement
2026-11-05 00:13:02 -08:00
committed by GitHub
parent 071c9c541d
commit 1db0d3ca5f
24 changed files with 119 additions and 35 deletions
+7
View File
@@ -1 +1,8 @@
tools/cli.mdx
# Branching is Beta; not documented until early access. tables/branching.mdx
# stays reachable (out of the nav only): SDK error messages link to it.
# To restore: docs.json Tutorials > Feature Engineering, after tutorials/feature-engineering/materialized-views.
agent-branch-experiments.mdx
# JSON columns are Beta; not documented until early access.
# To restore: docs.json Documentation > SQL Reference > DQL, after sql/dql/vector-distance.
sql/dql/json-functions.mdx
+5 -1
View File
@@ -78,7 +78,6 @@ file format, so moving data from one edition to the other does not require a dat
| | [Database](/management/database) | ✅ | ✅ |
| | [Namespace](/management/namespace) | ✅ | ✅ |
| | [Table](/management/table) | ✅ | ✅ |
| | [Branches](/tables/branching) | ✅ | ✅ |
| | [View](/management/view) | ❌ | ✅ |
| | [Materialized view](/management/materialized-view) | ❌ | ✅ |
| | [Secret](/management/secret) | ❌ | ✅ |
@@ -104,6 +103,11 @@ file format, so moving data from one edition to the other does not require a dat
| | [OpenTelemetry](/deployment/opentelemetry) | ✅ | ✅ |
| | [Kubernetes](/enterprise/deployment/kubernetes) | ❌ | ✅ |
{/*
Draft until branching reaches early access; restore to the Management rows.
| | [Branches](/tables/branching) | ✅ | ✅ |
*/}
## Usage differences between Enterprise and Embedded {#usage-differences-between-enterprise-and-oss}
The [quickstart](/quickstart) guide shows both LanceDB Embedded connections and Enterprise `db://...`
+11 -3
View File
@@ -10,7 +10,7 @@ keywords: ["names", "naming", "table name", "namespace", "identifier", "limits"]
<span id="objects" />
Databases, namespaces, tables, views and branches all have names. Choose names
Databases, namespaces, tables and views all have names.{/* Draft until branching reaches early access: restore "Databases, namespaces, tables, views and branches all have names." */} Choose names
that describe their purpose, such as `documents`, `prod` or `training_2026`.
Consistent names make data easier to find and application code easier to read.
@@ -21,10 +21,14 @@ Use the following rules for names that work across LanceDB Embedded and LanceDB
| Table, materialized view, view | ASCII letters and digits, `-`, `_`, `.` | 1 to 128 characters |
| Namespace component | ASCII letters and digits, `-`, `_`, `.` | 1 to 64 characters; at most 10 components in a path |
| Database | ASCII letters and digits, `-`, `_`, `.`, with `/` between components | 1 to 64 characters for the whole name |
| Branch | Avoid `$` and `:`; use ASCII letters, digits, `-` and `_` for portability | Keep to 1 to 64 characters |
| Secret | ASCII letters and digits, `-`, `_`, `.` | 1 to 128 characters |
| Function | ASCII letters and digits, `_` | 1 to 128 characters |
{/*
Draft until branching reaches early access; restore after the Database row.
| Branch | Avoid `$` and `:`; use ASCII letters, digits, `-` and `_` for portability | Keep to 1 to 64 characters |
*/}
- Names and individual namespace or database components must not be empty.
- Table, view, namespace and database components cannot be `.` or `..`.
- Secret names cannot be `.` or `..`.
@@ -60,5 +64,9 @@ The empty path `[]` identifies the root namespace.
## Learn more {#next}
- [Namespaces](/management/namespace): create namespaces and address tables.
- [Branches](/tables/branching): name and manage separate lines of table history.
- [Limits](/basics/limits): check other constraints before designing your schema.
{/*
Draft until branching reaches early access; restore after Namespaces.
- [Branches](/tables/branching): name and manage separate lines of table history.
*/}
+3
View File
@@ -1,5 +1,6 @@
---
title: "Create a Function"
tag: "Early access"
sidebarTitle: "Create Function"
description: "Register Python as a column: CPU, GPU, and remote-service Functions, called once per row or once per batch."
icon: code
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
Some curation steps cannot be written as a SQL expression: an embedding
model, an image decode, a call to an external service. A Function is a
Python callable registered with the database, versioned, and then applied as
+4 -1
View File
@@ -1,6 +1,7 @@
---
title: "GPU Function Execution"
sidebarTitle: "GPU Function Execution"
tag: "Early access"
sidebarTitle: "GPU Functions"
description: "How a GPU Function's worker gets scheduled onto a GPU-capable node, and what happens when none is free."
icon: microchip
keywords: ["gpu", "gpu class", "scheduling", "function", "udf", "kubernetes", "node selector"]
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
[Registering a GPU Function](/curation/create-function#gpu) sets `gpu=True`
and an optional `gpu_class`; this page covers what happens after that, when
a job actually needs a worker. An operator configures one or more named
+3
View File
@@ -1,5 +1,6 @@
---
title: "Vend a Secret to a Function"
tag: "Early access"
sidebarTitle: "Vend Secret"
description: "Bind a Secret to a Function by reference, and how its value reaches running code during SQL evaluation and refresh."
icon: key
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
A Function that calls an external service needs a credential without
putting it in the function's source -- see
[Create a Function](/curation/create-function#remote) for how to register
+1 -3
View File
@@ -590,7 +590,6 @@
"sql/dql/dedupe",
"sql/dql/duplicate-pairs",
"sql/dql/vector-distance",
"sql/dql/json-functions",
"sql/dql/blob-descriptor",
"sql/dql/generate-series",
"sql/dql/explain"
@@ -708,8 +707,7 @@
"group": "Feature Engineering",
"pages": [
"tutorials/feature-engineering/feature-engineering-101",
"tutorials/feature-engineering/materialized-views",
"agent-branch-experiments"
"tutorials/feature-engineering/materialized-views"
]
},
{
@@ -1,5 +1,6 @@
---
title: "Adaptive Execution"
tag: "Early access"
sidebarTitle: "Adaptive Execution"
description: "How a Function refresh's part size, resource class, worker count and checkpoint frequency adjust themselves to observed behavior."
icon: "gauge-high"
@@ -8,6 +9,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
When a [computed column](/feature-engineering/add-computed-column) or
[materialized view](/feature-engineering/materialized-views) is filled by
a registered [Function](/curation/create-function), LanceDB Enterprise
+3
View File
@@ -1,5 +1,6 @@
---
title: "Function Refresh Checkpointing"
tag: "Early access"
sidebarTitle: "Checkpointing"
description: "How a Function refresh's progress survives a worker failure or restart, part by part."
icon: "floppy-disk"
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
This page is about checkpointing a [Function refresh](/enterprise/adaptive-execution)
specifically -- not [PyTorch training checkpointing](/training/checkpointing),
a different, unrelated mechanism covered elsewhere.
@@ -1,5 +1,6 @@
---
title: "GPU Refresh Execution"
tag: "Early access"
sidebarTitle: "GPU Refresh Execution"
description: "A computed column or materialized view filled by a GPU Function schedules its refresh workers the same way GPU Function execution does in Data Curation."
icon: microchip
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
A [computed column](/feature-engineering/add-computed-column) or
[materialized view](/feature-engineering/materialized-views) filled by a
registered [Function](/curation/create-function) with `gpu=True` refreshes
+4 -1
View File
@@ -1,6 +1,7 @@
---
title: "Tiered Index Cache"
sidebarTitle: "Tiered Index Cache"
tag: "Early access"
sidebarTitle: "Index Cache"
description: "How LanceDB Enterprise caches deserialized index structures across memory and local disk."
icon: "layer-group"
availability:
@@ -8,6 +9,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
The tiered index cache holds deserialized vector, full-text and scalar
index structures in memory, backed by a local-disk tier beneath it. When
memory fills, the cache serializes evicted entries and spills them to disk
+5 -1
View File
@@ -23,7 +23,7 @@ time stitching infrastructure together and more time improving datasets, testing
## Accelerate the data flywheel for model research {#accelerate-the-data-flywheel-for-model-research}
Keep your AI data in one table, and give each agent-driven experiment its own isolated branch.
Keep your AI data in one table.{/* Draft until branching reaches early access: restore ", and give each agent-driven experiment its own isolated branch" before the period. */}
Find the failure, fix the data, and retrain.
<Steps>
@@ -31,11 +31,15 @@ Find the failure, fix the data, and retrain.
Start from one failed grasp or evaluation example. [Search](/search/) multimodal data, metadata, and embeddings
together, retrieve similar examples, and [curate](/curation) them into a targeted training subset.
</Step>
{/*
Draft until branching reaches early access.
<Step title="Data sandboxes for safe agentic experimentation">
Don't let an agent silently corrupt your data and ruin a training run. Give each agent its own
[branch](/tables/branching), a writable fork of the table, to experiment on, and review the changes before
they reach `main`. See [Run experiments on branches](/agent-branch-experiments).
</Step>
*/}
<Step title="A single dataset for every modality">
Keep episodes, frames, metadata, features, and embeddings together in one schema-enforced table, without
pipelines to keep them in sync. Table [versions](/management/table#versioning) let you trace each training
+2
View File
@@ -30,6 +30,8 @@ import {
import { PyWriteWorkflow, TsWriteWorkflowBody, TsWriteWorkflowTail, RsWriteWorkflow } from '/snippets/ingestion_examples.mdx';
<Badge color="purple">Early access in Enterprise</Badge>
Use merge insert when incoming records should be compared with existing rows.
Combine update and insert policies for an **upsert**. Use [append](/ingestion/insert)
when every incoming row should simply be added.
+10 -4
View File
@@ -37,7 +37,7 @@ retention period comes from its configuration instead.
The latest version is never removed, whatever you pass. `timedelta(0)` makes
every older version eligible for removal, but cleanup still keeps tagged
versions and the files branches still read, as the warning below describes.
versions{/* Draft until branching reaches early access: restore " and the files branches still read" */}, as the warning below describes.
<Warning>
Pruning is what makes [time travel](/management/table#versioning) finite: an
@@ -47,12 +47,18 @@ whatever its age, when it is:
- the **latest** version;
- a **tagged** version.
{/*
Draft until branching reaches early access.
Cleanup also keeps the files a **branch** still reads, and the version the
branch was forked from while the branch's remaining versions read the parent's
files. That does not keep the fork for good: once the branch has replaced the
data it inherited and its own older versions are pruned, cleanup prunes the
fork version like any other. To keep a version recoverable, tag it, subject to
the policy below, or keep the window wide enough to cover it.
fork version like any other.
*/}
To keep a version recoverable, tag it, subject to the policy below, or keep
the window wide enough to cover it.
Under the default policy, a tag can also stop cleanup. When a tagged version is
older than the window and is not the latest, cleanup refuses to run: it raises
@@ -75,7 +81,7 @@ need before running this, not after.
[Managed blobs](/management/table#managed-blobs) store out-of-line payloads
under the table's `_blobs/` directory. Cleanup follows the references from
retained snapshots, including protected tags and branches, before selecting
retained snapshots, including protected tags{/* Draft until branching reaches early access: restore " and branches" */}, before selecting
objects for deletion. Replacing a blob or deleting a row makes its old payload
eligible only after no retained snapshot references it.
+3
View File
@@ -1,5 +1,6 @@
---
title: "Functions"
tag: "Early access"
sidebarTitle: "Function"
description: "Register a Python function once and apply it as a column, for work SQL cannot express."
icon: code
@@ -23,6 +24,8 @@ import {
} from '/snippets/function.mdx';
import { PyConnectEnterprisePlaceholders as FunctionConnect } from '/snippets/connection.mdx';
<Badge color="purple">Early access</Badge>
Some columns cannot be written as a SQL expression: an embedding, a model call,
an image decode. A function is a Python callable registered with the database,
versioned, and then applied as a column. The code runs where the data is
+8 -7
View File
@@ -1,5 +1,6 @@
---
title: "Secret"
tag: "Early access"
sidebarTitle: "Secret"
description: "Create, rotate, list, describe and drop a named credential that Functions bind by reference."
icon: key
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
A **Secret** is a catalog object, like a table or a namespace: a named
credential, such as a model provider's API key, stored once and addressed by
name. A [Function](/curation/create-function) binds a Secret by reference
@@ -109,12 +112,10 @@ Every Secret's value is encrypted directly with your deployment's key in AWS
KMS, Google Cloud KMS, or Azure Key Vault -- there is no separate application
data key to manage. Storage, backups and replicas hold only ciphertext.
Decryption goes through your KMS, which records it in your cloud's audit
log. A deployment without a KMS key does not offer Secrets; see
[KMS integration](/enterprise/deployment/kms).
log.
<Badge color="blue">BYOC</Badge> You are responsible for the network egress
policy for the jobs that run Functions, and for scoping the Function
identity's cloud role to the KMS key so it can decrypt but not encrypt.
policy for the jobs that run Functions.
### Redaction {#redaction}
@@ -131,7 +132,7 @@ that holds that credential.
| Item | Limit |
| --- | --- |
| Secret name | 1--128 characters from `A-Z a-z 0-9 _ . -`. `$` is reserved as the namespace delimiter. |
| Value | One nonempty UTF-8 string, at most 4,096 bytes. Azure Key Vault RSA keys allow less -- see [KMS integration](/enterprise/deployment/kms). |
| Value | One nonempty UTF-8 string, at most 4,096 bytes. On Azure, at most 158, 286 or 414 bytes for an RSA-2048, RSA-3072 or RSA-4096 key. |
| Bindings per Function | 16 |
| Environment variable name | `[A-Za-z_][A-Za-z0-9_]*`, and not [reserved](#reserved) |
| Variables per Function | Each bound once. Two variables may bind the same Secret. |
@@ -157,8 +158,8 @@ refreshing one: the Secret was never created in that namespace, was dropped,
or its name is misspelled in the binding. Create it, or register a new
Function version with the right name.
**501 from a Secret API call**: the deployment has no KMS key configured, so
Secrets are disabled. See [KMS integration](/enterprise/deployment/kms).
**501 from a Secret API call**: Secrets aren't enabled for this deployment.
Contact [LanceDB support](/deployment/support).
## Privileges {#privileges}
+12 -6
View File
@@ -143,8 +143,8 @@ import {
A table stores data in rows and columns and keeps a history of committed
versions. Queries read table data, writes update it, and indexes accelerate
searches over its columns. Branches let you make changes separately from the
main table history.
searches over its columns.{/* Draft until branching reaches early access: restore "Branches let you make changes separately from the
main table history." */}
This page covers table management, data types, schema changes and version
history. For loading records, see [Ingestion](/ingestion). For changes to
@@ -161,7 +161,11 @@ In LanceDB Enterprise, tables are also created and dropped in SQL with [`CREATE
| **Columns** | including vectors, blobs, and [computed](/feature-engineering/add-computed-column) columns |
| **Versions** | every write commits one, so reads can [travel back](/management/table#versioning) |
| **Indexes** | vector, full-text and scalar, built over its columns |
{/*
Draft until branching reaches early access.
| **Branches** | isolated lines of change you can [merge or discard](/tables/branching) |
*/}
## Create a table {#create}
@@ -326,7 +330,7 @@ as it is to the format.
| String and binary | `utf8`, `large_utf8`, `binary`, `large_binary`, `fixed_size_binary(n)` |
| Date and time | `date32`, `date64`, `time32`, `time64`, `timestamp`, `duration` |
| Nested | `struct`, `list`, `large_list`, `fixed_size_list`, `map` |
| Lance extensions | vectors, [JSON](/sql/data-types#json), [blobs](#multimodal), bfloat16 |
| Lance extensions | vectors, [blobs](#multimodal), bfloat16{/* Draft until JSON columns reach early access: restore "[JSON](/sql/data-types#json), " after "vectors, " */} |
A vector column is a `fixed_size_list<float32, dim>`, an ordinary Arrow type,
which is why it is selectable and joinable but why arithmetic over it is list
@@ -345,9 +349,7 @@ file, up to 64 KiB by default, and larger values are stored outside it.
### Blob columns {#multimodal-blob-v2}
<Warning>
Blob V2 is in early access.
</Warning>
<Badge color="purple">Early access</Badge>
A `blob` column keeps metadata separate from the payload bytes, so queries can
filter rows without loading the blobs.
@@ -1490,6 +1492,9 @@ Deleting a tag only removes the label, not the version it points to. After
deletion, the underlying table version becomes eligible for cleanup again.
</Note>
{/*
Draft until branching reaches early access.
### Branches {#versioning-branches}
Beyond linear history, LanceDB also supports **branches**: isolated, writable
@@ -1499,6 +1504,7 @@ writable history, making it ideal for experiments, backfills, and migrations
that you want to keep separate from production reads on `main`.
Branches are covered in their own guide: see [Branches](/tables/branching).
*/}
### Delete Data From the Table {#versioning-delete-data-from-the-table}
+2 -2
View File
@@ -70,8 +70,8 @@ returning a result that omits un-compacted rows.
## Limitations {#limitations}
<Note>
Support for all table operations on RTE tables is coming soon. Time travel,
branching, `count_rows`, row IDs (`with_row_id`), and `TABLESAMPLE` are not yet
Support for all table operations on RTE tables is coming soon. Time travel,{/* Draft until branching reaches early access: restore " branching," */}
`count_rows`, row IDs (`with_row_id`), and `TABLESAMPLE` are not yet
available.
</Note>
+9 -1
View File
@@ -3,7 +3,7 @@ title: "SQL data types"
sidebarTitle: "Data Types"
description: "How Lance columns appear to SQL, including the types SQL has no name for."
icon: shapes
keywords: ["sql types", "arrow", "cast", "pseudo-columns", "vector", "json"]
keywords: ["sql types", "arrow", "cast", "pseudo-columns", "vector"]
availability:
embedded: unavailable
enterprise: available
@@ -84,6 +84,9 @@ calculations. SQL vector search depends on the Enterprise server version and que
plan; see [SQL](/basics/sql#no-vector-sql).
</Warning>
{/*
Draft until JSON columns reach early access.
### JSON {#json}
Stored as JSONB in `large_binary` with the `lance.json` extension name. It reads
@@ -98,6 +101,7 @@ WHERE json_get_int(data, 'status') = 200
`json_get`, its typed variants, and `json_extract` are **index-accelerated** by a
JSON scalar index. See [JSON functions](/sql/dql/json-functions).
*/}
### bfloat16 {#bfloat16}
@@ -105,6 +109,9 @@ Sixteen-bit floats in the bfloat16 layout, stored as `fixed_size_binary(2)`.
Useful as a vector element type when halving the memory matters more than the
precision does.
{/*
Draft: geospatial queries and indexes are unsupported in LanceDB Enterprise.
## Geospatial {#geospatial}
<Warning>
@@ -118,6 +125,7 @@ spatial index is gated the same way.
Geometry can still be stored (it is binary, and binary columns work), but
nothing interprets it. Filter on a bounding box held in ordinary numeric
columns if you need spatial predicates today.
*/}
## Pseudo-columns {#pseudo-columns}
+3
View File
@@ -1,5 +1,6 @@
---
title: "Function"
tag: "Early access"
sidebarTitle: "Function"
description: "Create, describe, alter, list, and drop registered functions with SQL."
icon: function
@@ -9,6 +10,8 @@ availability:
enterprise: available
---
<Badge color="purple">Early access</Badge>
[SQL runs in LanceDB Enterprise.](/sql)
## CREATE FUNCTION {#create}
+2 -1
View File
@@ -1,12 +1,13 @@
---
title: "Secret"
tag: "Early access"
sidebarTitle: "Secret"
description: "CREATE, ALTER, DROP, SHOW and DESCRIBE SECRET, and granting privileges on Secrets."
icon: key
keywords: ["secret", "create secret", "alter secret", "drop secret", "show secrets", "describe secret", "grant", "sql"]
---
<Badge color="red">Enterprise</Badge> [SQL runs in LanceDB Enterprise.](/sql)
<Badge color="purple">Early access</Badge> <Badge color="red">Enterprise</Badge> [SQL runs in LanceDB Enterprise.](/sql)
A [Secret](/management/secret) is a named credential that Functions bind by
reference. These statements store, rotate, list, describe and drop Secrets.
+6 -2
View File
@@ -15,7 +15,7 @@ plus what Lance adds.
| | |
| --- | --- |
| [Scalar functions](/sql/dql/scalar-functions) | per row: math, string, regex, date/time, array, struct, JSON |
| [Scalar functions](/sql/dql/scalar-functions) | per row: math, string, regex, date/time, array, struct{/* Draft until JSON columns reach early access: restore ", JSON" */} |
| [Aggregate functions](/sql/dql/aggregation#functions) | per group: `count`, `sum`, statistical, approximate |
| [Window functions](/sql/dql/window#functions) | per window: ranking, offset, running aggregates |
| [Table functions](#table-functions) | produce a relation rather than a value |
@@ -45,13 +45,17 @@ see [Data types](/sql/data-types#pseudo-columns).
## What Lance accelerates {#accelerated}
Two function families use an index rather than scanning, which is the difference
`contains_tokens` uses an index rather than scanning, which is the difference
between a fast predicate and a slow one:
| | Index used |
| --- | --- |
| [`contains_tokens`](/sql/dql/contains-tokens) | full-text index |
{/*
Draft until JSON columns reach early access; restore "Two function families" above.
| [`json_get`, typed variants, `json_extract`](/sql/dql/json-functions) | JSON scalar index |
*/}
Confirm with [`EXPLAIN`](/sql/dql/explain): an accelerated predicate shows an
index node rather than a `LanceScan` followed by a filter.
+9 -1
View File
@@ -3,7 +3,7 @@ title: "Scalar functions"
sidebarTitle: "Scalar Functions"
description: "The per-row function set: math, string, regex, date/time, array, struct, and what Lance adds."
icon: calculator
keywords: ["scalar functions", "string functions", "math functions", "regex", "array functions", "json"]
keywords: ["scalar functions", "string functions", "math functions", "regex", "array functions"]
availability:
embedded: unavailable
enterprise: available
@@ -133,7 +133,14 @@ reference page:
| [`contains_tokens`](/sql/dql/contains-tokens) | index-accelerated keyword predicate, usable in `WHERE` |
| [`ivf_partition`](/sql/dql/ivf-partition) | assign each row to its trained IVF index partition |
| [`blob_descriptor`](/sql/dql/blob-descriptor) | a BlobV2 column's payload location, without reading it |
{/*
Draft until JSON columns reach early access.
| [JSON accessors](/sql/dql/json-functions) | `json_get` and its typed variants, `json_extract`, and more |
*/}
{/*
Draft: geospatial queries and indexes are unsupported in LanceDB Enterprise.
### Geospatial {#geo}
@@ -143,6 +150,7 @@ Lance defines `st_*` functions on GeoArrow, but they sit behind a build feature
feature rather than an unknown-function error, and the spatial index is gated
the same way.
</Warning>
*/}
## Additional scalar functions {#escape-hatch}
+1 -1
View File
@@ -21,7 +21,7 @@ are runnable in Colab; the rest are written up here.
Retrieval-augmented generation and agent applications.
</Card>
<Card title="Feature engineering" icon="wand-magic-sparkles" href="/tutorials/feature-engineering/feature-engineering-101">
Deriving columns, materialized views, experimenting on branches.
Deriving columns and materialized views.{/* Draft until branching reaches early access: restore "Deriving columns, materialized views, experimenting on branches." */}
</Card>
<Card title="Training" icon="graduation-cap" href="/training/object-detection">
Curating data and training models directly on a table.