> ## Documentation Index > Fetch the complete documentation index at: https://docs.oxen.ai/llms.txt > Use this file to discover all available pages before exploring further. # Create an evaluation > Run model inference over a dataset in the repository. The resource path (branch and file) is encoded in the URL, e.g. `POST /api/repos/{ns}/{repo}/evaluations/main/datasets/training.parquet`. ## OpenAPI ````yaml https://dev.hub.oxen.ai/api/_spec/oxen_hub_api.json post /api/repos/{namespace}/{repo_name}/evaluations/*resource_path openapi: 3.0.0 info: title: oxen version: 0.224.2 servers: - url: https://dev.hub.oxen.ai variables: {} security: [] tags: [] paths: /api/repos/{namespace}/{repo_name}/evaluations/*resource_path: post: tags: - evaluations summary: Create an evaluation description: >- Run model inference over a dataset in the repository. The resource path (branch and file) is encoded in the URL, e.g. `POST /api/repos/{ns}/{repo}/evaluations/main/datasets/training.parquet`. operationId: OxenApiWeb.Controllers.EvaluationController.create_evaluation parameters: [] requestBody: content: application/json: schema: $ref: '#/components/schemas/CreateEvaluationRequest' description: Create evaluation request required: false responses: '200': content: application/json: schema: $ref: '#/components/schemas/EvaluationResponse' description: Evaluation created '400': content: application/json: schema: type: object description: Invalid parameters callbacks: {} security: - authorization: [] components: schemas: CreateEvaluationRequest: description: > Request payload to create an evaluation. The resource path (branch + file path) is specified in the URL after `/evaluations/`, for example: `POST /api/repos/{namespace}/{repo}/evaluations/main/datasets/training.parquet`. properties: auto_commit: default: false description: >- If true, automatically commit the results when the evaluation completes type: boolean batch_size: default: 10 description: Number of rows to process per inference batch type: integer commit_message: description: Commit message used when auto_commit is true nullable: true type: string input_type: description: Type of input data enum: - text - image - video type: string is_sample: default: false description: >- If true, only evaluate a subset of rows. If false, evaluate all rows. type: boolean model_id: description: ID of the model to run inference with format: uuid type: string name: description: Human-readable name for the evaluation type: string output_type: description: Type of output produced by the model enum: - text - image - video - embeddings type: string prompt: description: >- Prompt template sent to the model. Use `{column_name}` placeholders to inject values from each row. type: string sample_count: default: 10 description: Number of rows to sample when is_sample is true type: integer target_branch: description: Branch where evaluation results are committed type: string target_column: description: Column where the model output will be written type: string target_path: description: File path for the output data frame type: string required: - prompt - input_type - output_type - model_id - target_column title: CreateEvaluationRequest type: object EvaluationResponse: description: Standard response wrapper for a single evaluation. properties: evaluation: anyOf: - $ref: '#/components/schemas/Evaluation' description: Evaluation resource status: description: High-level status string (for example, 'success'). type: string status_message: description: Human-readable status message (for example, 'resource_found'). type: string title: EvaluationResponse type: object Evaluation: description: >- An evaluation resource representing a model inference run over a dataset. properties: cancelled_at: format: date_time nullable: true type: string completed_at: format: date_time nullable: true type: string completion_tokens_used: description: Completion tokens consumed type: integer created_by: description: User who created the evaluation nullable: true properties: id: type: string image: nullable: true type: string name: nullable: true type: string username: type: string type: object credits_used: description: Credits consumed format: float type: number error_message: description: Error details if status is errored nullable: true type: string id: description: Evaluation ID format: uuid type: string input_type: enum: - text - image - video type: string inserted_at: format: date_time type: string is_sample: type: boolean model: additionalProperties: true description: Model used for this evaluation nullable: true type: object name: description: Evaluation name nullable: true type: string output_type: enum: - text - image - video - embeddings type: string progress: description: Processing progress properties: processed: description: Number of rows processed so far type: integer total: description: Total number of rows to process type: integer type: object prompt: description: Prompt template used for inference type: string prompt_tokens_used: description: Prompt tokens consumed type: integer repository_id: type: string resource: description: Source data resource nullable: true properties: path: description: Path inside the repository type: string version: description: Commit ID or version type: string type: object sample_count: type: integer started_at: format: date_time nullable: true type: string status: description: Current status enum: - queued - running - committing - completed - errored - cancelled type: string target_branch: nullable: true type: string target_column: nullable: true type: string target_path: nullable: true type: string tokens_used: description: Total tokens (prompt + completion) type: integer title: Evaluation type: object securitySchemes: authorization: scheme: bearer type: http ````