4319f38c05
- Created new YAML files for audience size and audience type slots, defining their properties and annotations. - Added archived capacity slots including cubic meters, linear meters, item count, and descriptions, with appropriate URIs and ranges. - Introduced a template specificity slot for context-aware RAG filtering. - Consolidated capacity-related slots into a unified structure, including has_or_had_capacity, capacity_type, and capacity_value, with detailed descriptions and examples.
144 lines
4.6 KiB
YAML
144 lines
4.6 KiB
YAML
# Token - Token/unit count class
|
|
# Updated: 2026-01-17 per Rule 53/56 migration for cached_token
|
|
#
|
|
# Uses has_or_had_type with TokenType range instead of simple attribute.
|
|
|
|
id: https://nde.nl/ontology/hc/class/Token
|
|
name: token_class
|
|
title: Token Class
|
|
version: 1.1.0
|
|
|
|
prefixes:
|
|
linkml: https://w3id.org/linkml/
|
|
hc: https://nde.nl/ontology/hc/
|
|
schema: http://schema.org/
|
|
|
|
imports:
|
|
- linkml:types
|
|
- ../slots/has_or_had_identifier
|
|
- ../slots/has_or_had_label
|
|
- ../slots/has_or_had_description
|
|
- ../slots/has_or_had_type
|
|
- ../slots/has_or_had_quantity
|
|
- ./TokenType
|
|
- ./Quantity
|
|
|
|
default_prefix: hc
|
|
|
|
classes:
|
|
Token:
|
|
class_uri: schema:QuantitativeValue
|
|
description: |
|
|
Token count representation for LLM API usage tracking.
|
|
|
|
**Key Fields**:
|
|
|
|
| Field | Description | Example |
|
|
|-------|-------------|---------|
|
|
| has_or_had_type | Type of token | CachedTokenType |
|
|
| has_or_had_quantity | Count value | 50 (cached tokens) |
|
|
|
|
**Schema.org Alignment**:
|
|
|
|
Maps to `schema:QuantitativeValue` for numeric values with context.
|
|
|
|
**Use Cases**:
|
|
|
|
1. **Cached tokens**: Prompt tokens served from cache (reduced cost)
|
|
2. **Reasoning tokens**: Chain-of-thought tokens (GLM 4.7, o1)
|
|
3. **Input/Output tokens**: Standard prompt/completion counts
|
|
|
|
**Migration History**:
|
|
|
|
- v1.0.0: Simple class with token_type attribute
|
|
- v1.1.0: Updated to use has_or_had_type → TokenType per Rule 53/56
|
|
|
|
**Example (Cached Tokens)**:
|
|
|
|
```yaml
|
|
Token:
|
|
has_or_had_type:
|
|
has_or_had_identifier: hc:TokenType/CACHED
|
|
has_or_had_label: Cached Token
|
|
has_or_had_quantity:
|
|
quantity_value: 50
|
|
has_or_had_description: Tokens from provider cache
|
|
```
|
|
exact_mappings:
|
|
- schema:QuantitativeValue
|
|
slots:
|
|
- has_or_had_identifier
|
|
- has_or_had_label
|
|
- has_or_had_description
|
|
- has_or_had_type
|
|
- has_or_had_quantity
|
|
slot_usage:
|
|
has_or_had_identifier:
|
|
range: uriorcurie
|
|
required: false
|
|
examples:
|
|
- value: https://nde.nl/ontology/hc/token/llm-response-001-cached
|
|
description: Token count instance identifier
|
|
has_or_had_type:
|
|
range: TokenType
|
|
required: false
|
|
inlined: true
|
|
description: |
|
|
The type of token (INPUT, OUTPUT, CACHED, REASONING, TOTAL).
|
|
References TokenType taxonomy.
|
|
examples:
|
|
- value:
|
|
has_or_had_identifier: hc:TokenType/CACHED
|
|
has_or_had_label: Cached Token
|
|
description: Cached token type
|
|
- value:
|
|
has_or_had_identifier: hc:TokenType/REASONING
|
|
has_or_had_label: Reasoning Token
|
|
description: Reasoning/thinking token type
|
|
has_or_had_quantity:
|
|
range: Quantity
|
|
required: false
|
|
inlined: true
|
|
description: |
|
|
The numeric token count with optional metadata.
|
|
For simple use cases, prefer quantity_value directly.
|
|
examples:
|
|
- value:
|
|
quantity_value: 50
|
|
has_or_had_description: Tokens served from cache
|
|
description: Cached token count
|
|
has_or_had_label:
|
|
range: string
|
|
required: false
|
|
examples:
|
|
- value: Cached tokens for context window
|
|
has_or_had_description:
|
|
range: string
|
|
required: false
|
|
examples:
|
|
- value: 50 prompt tokens served from provider KV cache
|
|
comments:
|
|
- Token models LLM token counts with type and quantity
|
|
- Updated per Rule 53/56 to use has_or_had_type → TokenType
|
|
- Replaces simple token_type string attribute
|
|
see_also:
|
|
- https://platform.openai.com/docs/api-reference/chat/object#usage
|
|
- https://docs.z.ai/guides/capabilities/thinking-mode
|
|
examples:
|
|
- value:
|
|
has_or_had_identifier: https://nde.nl/ontology/hc/token/example-cached-001
|
|
has_or_had_type:
|
|
has_or_had_identifier: hc:TokenType/CACHED
|
|
has_or_had_label: Cached Token
|
|
has_or_had_quantity:
|
|
quantity_value: 50
|
|
has_or_had_description: Prompt tokens from cache
|
|
has_or_had_description: 50 cached tokens reducing API cost
|
|
description: Cached token count with full metadata
|
|
- value:
|
|
has_or_had_type:
|
|
has_or_had_identifier: hc:TokenType/REASONING
|
|
has_or_had_label: Reasoning Token
|
|
has_or_had_quantity:
|
|
quantity_value: 1250
|
|
description: Reasoning tokens from GLM 4.7 thinking mode
|