Entity Fields Reference#

Review the following key fields for major entities in the NeMo Microservices Platform when using the NeMo Entity Store and NeMo Data Store microservice APIs.

Tip

For more information about APIs, see the API Reference.


Namespace Fields#

Namespaces are the top-level organizational unit for all other entities.

Field

Type

Required

Description

id

string

Yes

Unique identifier for the namespace

description

string

No

Description of the namespace’s purpose

custom_fields

object

No

User-defined metadata key-value pairs

ownership.created_by

string

No

ID of the user who created the namespace

ownership.access_policies

object

No

Access control policies for the namespace

Example Namespace Object in JSON Format
{
  "id": "research-team",
  "description": "Research team workspace for LLM experiments",
  "created_at": "2024-01-15T10:30:00Z",
  "updated_at": "2024-01-15T10:30:00Z",
  "custom_fields": {
    "department": "AI Research",
    "cost_center": "CC-123"
  },
  "ownership": {
    "created_by": "user-123",
    "access_policies": {
      "admins": ["team-leads"],
      "readers": ["all-researchers"]
    }
  }
}

Model Fields#

Models represent machine learning models that can be used for inference.

Field

Type

Required

Description

name

string

No

Unique name within the namespace

namespace

string

No

Namespace the model belongs to (defaults to “default”)

description

string

No

Description of the model

spec.num_parameters

integer

Yes

Number of model parameters

spec.context_size

integer

Yes

Maximum context size

spec.num_virtual_tokens

integer

Yes

Number of virtual tokens

spec.is_chat

boolean

Yes

Whether this is a chat model

base_model

string/object

No

Reference to another model used as base

precision

enum

No

Model precision (int8, bf16, fp16, fp32, bf16-mixed)

project

string

No

Associated project ID

custom_fields

object

No

User-defined metadata key-value pairs

Example Model Configuration in JSON Format
{
  "name": "llama-7b-chat",
  "namespace": "research-team",
  "description": "Fine-tuned LLaMA 7B chat model",
  "spec": {
    "num_parameters": 7000000000,
    "context_size": 4096,
    "num_virtual_tokens": 32000,
    "is_chat": true
  },
  "base_model": "meta-llama/Llama-2-7b",
  "precision": "bf16-mixed",
  "project": "chat-assistant",
  "custom_fields": {
    "training_epochs": 3,
    "evaluation_score": 0.85
  }
}

Model Precision Options#

  • int8: 8-bit integer precision

  • bf16: Brain floating-point format

  • fp16: 16-bit floating-point

  • fp32: 32-bit floating-point

  • bf16-mixed: Mixed precision with Brain floating-point

Dataset Fields#

Datasets represent data used for fine-tuning or evaluation.

Field

Type

Required

Description

name

string

No

Unique name within the namespace

namespace

string

No

Namespace the dataset belongs to (defaults to “default”)

description

string

No

Description of the dataset

files_url

string

Yes

Location of dataset files (URL)

format

string

No

Dataset format (e.g., SQuAD, BEIR)

project

string

No

Associated project ID

custom_fields

object

No

User-defined metadata key-value pairs

Example Dataset Configuration in JSON Format
{
  "name": "customer-support-data",
  "namespace": "research-team",
  "description": "Cleaned customer support conversations",
  "files_url": "s3://company-datasets/support-data/",
  "format": "jsonl",
  "project": "chat-assistant",
  "custom_fields": {
    "num_examples": 50000,
    "data_source": "Support Tickets Q1 2024",
    "preprocessing_version": "v2.1"
  }
}

Dataset URL Support#

The files_url field supports various storage locations:

  • NeMo Data Store

  • Hugging Face

  • S3

Project Fields#

Projects help organize related entities together.

Field

Type

Required

Description

name

string

No

Unique name within the namespace

namespace

string

No

Namespace the project belongs to (defaults to “default”)

description

string

No

Description of the project

custom_fields

object

No

User-defined metadata key-value pairs

Example Project Configuration in JSON Format
{
  "name": "chat-assistant",
  "namespace": "research-team",
  "description": "Customer support chat assistant development",
  "custom_fields": {
    "team_lead": "alice.smith",
    "target_completion": "2024-Q2",
    "priority": "high"
  }
}

Common Fields#

These fields are shared across most entities:

Field

Type

Description

created_at

datetime

When the entity was created

updated_at

datetime

When the entity was last updated

ownership.created_by

string

ID of the user who created the entity

ownership.access_policies

object

Access control policies

custom_fields

object

User-defined metadata key-value pairs