Home

json2vec is for predictive modeling on records that are not naturally flat. Customers have transactions, orders have line items, sessions have clickstream events, devices recur across histories, and every level can carry signal.

Most ML pipelines handle that shape by flattening it first: build rolling aggregates, hand-pick windows, maintain feature stores, and then train a model on one fixed row. That can work, but it makes representation a separate engineering system. json2vec takes the opposite path: describe the structured record, and the schema becomes the model.

The framework is designed for large production datasets, not just toy nested examples: training and batch inference over billions of observations, with throughput-oriented data paths and logging for pipelines that can process 100k+ observations per second on appropriately sized hardware. The examples in these docs stay deliberately small so the modeling loop is easy to inspect.

Core Idea

A json2vec schema is both a data contract and an architecture blueprint.

• Leaf fields such as Number, Category, Set, Entity, Text, and Vector become datatype-specific tensorfields.
• Branch nodes define shared contexts for child fields, with optional local attention and pooling before the representation flows upward.
• Targets, masks, pruning, and embeddings are configured on the same schema tree.
• Prediction output is keyed by schema address, so decoded values and embeddings remain attached to the part of the record that produced them.

That lets one model surface support supervised prediction, self-supervised reconstruction, embedding export, schema mutation, field importance, and serving without rebuilding the data representation for each workflow.

Execution Model

json2vec builds Lightning-compatible models. The schema defines the model tree, typed losses, prediction outputs, and embeddings; Lightning runs the fit, validation, test, and prediction loops, including device placement, callbacks, logging, checkpointing, and distributed execution.

A Schema Defines A Model

The generated root schema node is named record by default. This example names it order, which changes output addresses such as order/returned but does not change the source keys read from each input record.

import json2vec as jv

model = jv.Model(
    d_model=64,
    n_layers=2,
    n_heads=4,
    embed=True,
    customer_tier=jv.Category(size=16),
    returned=jv.Category(target=True, size=2),
    line_items=jv.Branch(
        embed=True,
        length=32,
        sku=jv.Category(size=2048),
        quantity=jv.Number,
        price=jv.Number,
    ),
)

model

Model780.5K params · 3.0 MB

▶nodes ModuleDict780.5K

▶record/customer_tier NodeModule52.9K

▶embedder Embedder1.3K

vocab OnlineVocabularyModel

▶embeddings ModuleDict1.3K

state Embedding 5, 64320

dtype float32
device cpu

content Embedding 16, 641.0K

dtype float32
device cpu

▶counters ModuleDict0

state Counter

content Counter

▶decoder Decoder51.5K

▶pool LearnedQueryCrossAttention50.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶linears ModuleDict1.4K

state Linear in_features=64, out_features=5, bias=True325

in_features 64
out_features 5
bias True

tensor

dtype float32
device cpu

content Linear in_features=64, out_features=16, bias=True1.0K

in_features 64
out_features 16
bias True

tensor

dtype float32
device cpu

▶record/returned NodeModule51.1K

▶embedder Embedder448

vocab OnlineVocabularyModel

▶embeddings ModuleDict448

state Embedding 5, 64320

dtype float32
device cpu

content Embedding 2, 64128

dtype float32
device cpu

▶counters ModuleDict0

state Counter

content Counter

▶decoder Decoder50.6K

▶pool LearnedQueryCrossAttention50.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶linears ModuleDict455

state Linear in_features=64, out_features=5, bias=True325

in_features 64
out_features 5
bias True

tensor

dtype float32
device cpu

content Linear in_features=64, out_features=2, bias=True130

in_features 64
out_features 2
bias True

tensor

dtype float32
device cpu

▶record/line_items/sku NodeModule317.0K

▶embedder Embedder131.4K

vocab OnlineVocabularyModel

▶embeddings ModuleDict131.4K

state Embedding 5, 64320

dtype float32
device cpu

content Embedding 2048, 64131.1K

dtype float32
device cpu

▶counters ModuleDict0

state Counter

content Counter

▶decoder Decoder185.6K

▶pool LearnedQueryCrossAttention52.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶linears ModuleDict133.4K

state Linear in_features=64, out_features=5, bias=True325

in_features 64
out_features 5
bias True

tensor

dtype float32
device cpu

content Linear in_features=64, out_features=2048, bias=True133.1K

in_features 64
out_features 2048
bias True

tensor

dtype float32
device cpu

▶record/line_items/quantity NodeModule54.6K

▶embedder Embedder2.0K

embeddings Embedding 5, 64320

dtype float32
device cpu

counter Counter

linear Linear in_features=26, out_features=64, bias=True1.7K

in_features 26
out_features 64
bias True

tensor

dtype float32
device cpu

normalizer GlobalOnlineNormalizer

▶decoder Decoder52.6K

▶pool LearnedQueryCrossAttention52.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

classification Linear in_features=64, out_features=5, bias=True325

in_features 64
out_features 5
bias True

tensor

dtype float32
device cpu

regression Linear in_features=64, out_features=1, bias=True65

in_features 64
out_features 1
bias True

tensor

dtype float32
device cpu

▶record/line_items/price NodeModule54.6K

▶embedder Embedder2.0K

embeddings Embedding 5, 64320

dtype float32
device cpu

counter Counter

linear Linear in_features=26, out_features=64, bias=True1.7K

in_features 26
out_features 64
bias True

tensor

dtype float32
device cpu

normalizer GlobalOnlineNormalizer

▶decoder Decoder52.6K

▶pool LearnedQueryCrossAttention52.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

classification Linear in_features=64, out_features=5, bias=True325

in_features 64
out_features 5
bias True

tensor

dtype float32
device cpu

regression Linear in_features=64, out_features=1, bias=True65

in_features 64
out_features 1
bias True

tensor

dtype float32
device cpu

▶record NodeModule150.1K

▶encoder BranchEncoder150.1K

▶encoder ModuleList100.0K

▶0 RotaryTransformerEncoderLayer50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

▶1 RotaryTransformerEncoderLayer50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

▶pool LearnedQueryCrossAttention50.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶record/line_items NodeModule100.2K

▶encoder BranchEncoder100.2K

▶encoder ModuleList50.0K

▶0 RotaryTransformerEncoderLayer50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

▶pool LearnedQueryCrossAttention50.2K

▶blocks ModuleList50.0K

▶0 CrossAttentionBlock50.0K

attention_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

ffn_norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

▶attention RotaryMultiheadAttention16.6K

q_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

k_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

v_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

out_proj Linear in_features=64, out_features=64, bias=True4.2K

in_features 64
out_features 64
bias True

tensor

dtype float32
device cpu

rotary RotaryEmbedding

▶ffn Sequential33.1K

0 Linear in_features=64, out_features=256, bias=True16.6K

in_features 64
out_features 256
bias True

tensor

dtype float32
device cpu

1 GELU approximate='none'

approximate 'none'

2 Dropout p=0.0, inplace=False

p 0.0
inplace False

3 Linear in_features=256, out_features=64, bias=True16.4K

in_features 256
out_features 64
bias True

tensor

dtype float32
device cpu

4 Dropout p=0.0, inplace=False

p 0.0
inplace False

norm LayerNorm (64,) eps=1e-05, elementwise_affine=True, bias=True128

eps 1e-05
elementwise_affine True
bias True

tensor

dtype float32
device cpu

Module typesWeightActivationNormalizationRegularizationTrainableFrozen / no params

This model reads records shaped like:

{
    "customer_tier": "gold",
    "line_items": [
        {"sku": "A12", "quantity": 2, "price": 19.99},
        {"sku": "B07", "quantity": 1, "price": 45.50},
    ],
    "returned": "false",
}

{"customer_tier": "gold", "line_items": [{"sku": "A12", "quantity": 2, "price": "text/plain+float:19.99"}, {"sku": "B07", "quantity": 1, "price": "text/plain+float:45.5"}], "returned": "false"}

The model learns from the order as a structured object. The line_items branch has its own repeated context, returned is withheld and decoded as a supervised target, and embed=True asks prediction to emit embeddings at configured addresses.

What Is Different

• Hierarchical context encoding: child records interact locally before their representation flows upward. A login session, transaction list, or order item list can keep its own window instead of competing inside one flat sequence.
• Extensible datatypes: each field type owns validation, tensorization, missing-state handling, masking, decoding, loss, metrics, and output writing.
• Unified training roles: target=True, p_prune, and p_mask all use the same reconstruction path, so supervised and self-supervised objectives share the model tree.
• Embedding trees: embeddings can come from the root, branches, or selected leaves, making branch-level retrieval and diagnostics possible.
• Schema evolution: models are completely mutable and can be updated, extended, deleted, reset, or temporarily overridden after construction.
• Scale-oriented runtime: online preprocessing state, streaming data modules, throughput logging, and checkpointed tensorfield state are designed for high-volume training and prediction jobs.
• One path for training and serving: queries, preprocessors, tensorization, model execution, prediction writing, and postprocessors stay in the same configured path.

When It Fits

Use json2vec when relationships inside the record matter: account histories, fraud or risk snapshots, order and fulfillment events, flight itineraries, operations telemetry, user sessions, repeated measurements, or any mixed datatype object where flattening would discard useful structure.

Use a simpler tabular model when flattening loses no meaningful context. The point is not to replace every table. The point is to model nested business data without making a feature table the only representation the model can see.

Where To Start

• New to the package: Getting Started
• Read the schema as a tree: Model Tree
• Bind source records to fields: Query Paths
• Short API map: AI / Expert Quickstart
• Modeling rationale: Why json2vec
• Built-in fields: Data Types
• Learning modes and embeddings: Learning Modes & Embeddings
• Windowed reconstruction in branches: Dynamic Masking
• Training execution: Training With Lightning
• In-memory and streaming inputs: Data Modules
• Offline prediction jobs: Batch Inference
• Serving-time output shaping: Postprocessors
• Input ablation diagnostics: Field Importance
• Applied risk example: Device Tenure

Learning Path

For a first pass, follow this sequence:

Getting Started -> Model Tree -> Query Paths -> Data Types -> Dynamic Masking -> Training With Lightning -> Data Modules -> Batch Inference

Community

Join the json2vec Discord for questions, design discussion, and release notes.