Skip to content

Input Configuration

Input specifications in gpux.yml.

Overview

The inputs section defines model input specifications.

inputs:
  - name: string          # Required: Input name
    type: string          # Required: Data type
    shape: [int]          # Optional: Tensor shape
    required: bool        # Optional: Required (default: true)
    max_length: int       # Optional: Maximum length
    description: string   # Optional: Description

Fields

name (required)

Input name matching the model's expected input.

  • Type: string
  • Required: Yes
inputs:
  - name: input_ids
  - name: attention_mask

type (required)

Data type of the input tensor.

  • Type: string
  • Required: Yes
  • Values: float32, float64, int32, int64, uint8, bool, string
inputs:
  - name: image
    type: float32
  - name: text
    type: string

shape

Tensor shape specification.

  • Type: list[int]
  • Required: No
  • Dynamic: Use -1 for dynamic dimensions
inputs:
  - name: image
    type: float32
    shape: [1, 3, 224, 224]  # Fixed shape
  - name: text
    type: int64
    shape: [1, -1]           # Dynamic length

required

Whether the input is required.

  • Type: boolean
  • Required: No
  • Default: true
inputs:
  - name: text
    type: string
    required: true
  - name: metadata
    type: string
    required: false

max_length

Maximum length for variable-length inputs.

  • Type: integer
  • Required: No
inputs:
  - name: text
    type: int64
    max_length: 512

description

Human-readable description.

  • Type: string
  • Required: No
inputs:
  - name: attention_mask
    type: int64
    description: Binary mask indicating valid tokens (1) and padding (0)

Examples

Image Input

inputs:
  - name: image
    type: float32
    shape: [1, 3, 224, 224]
    required: true
    description: RGB image tensor normalized to [0, 1]

Text Input

inputs:
  - name: input_ids
    type: int64
    shape: [1, 128]
    required: true
    description: Tokenized input IDs
  - name: attention_mask
    type: int64
    shape: [1, 128]
    required: true
    description: Attention mask for input

Multiple Inputs

inputs:
  - name: image
    type: float32
    shape: [1, 3, 224, 224]
  - name: text
    type: int64
    shape: [1, 77]

Alternative Syntax

Dict-style inputs:

inputs:
  input_ids:
    type: int64
    shape: [1, 128]
  attention_mask:
    type: int64
    shape: [1, 128]

See Also