Struct ModelsResource 

pub struct ModelsResource<'c> { /* private fields */ }

client.models() — B-112 embedded ML model registry.

Upload ONNX models that the streaming ml_predict operator scores events against, in-process on the Pulse engine (no model-server hop). Models are org-scoped; upload / delete require the ADMIN role.

Example

use pulse_client::{PulseClient, ModelUpload};
use std::collections::BTreeMap;

let mut input = BTreeMap::new();
input.insert("amount".to_string(), "float".to_string());
input.insert("country".to_string(), "string".to_string());

client
    .models()
    .upload(
        ModelUpload::from_path("fraud-classifier", "./model.onnx")
            .input_schema(input),
    )
    .await?;

Implementations

impl ModelsResource<'_>

pub async fn upload(&self, upload: ModelUpload) -> Result<Value, PulseError>

POST /api/pulse/ml-models — upload (or replace) a model.

Sent as multipart/form-data: a file part named model carrying the bytes, plus text parts name, runtime, and (when set) inputSchema / outputSchema as JSON strings. Replacing an existing name hot-swaps the model with no agent restart.

Returns the persisted model metadata (name, runtime, sha256, version, …).

Errors

• PulseError::InvalidConfig if name is blank, if neither or both of path/data are set, or if the model bytes are empty.
• PulseError::Transport if reading the file at path fails.

pub async fn list(&self) -> Result<Vec<Value>, PulseError>

GET /api/pulse/ml-models — models registered for the caller’s org.

pub async fn get(&self, name: &str) -> Result<Value, PulseError>

GET /api/pulse/ml-models/{name} — metadata for one model.

pub async fn delete(&self, name: &str) -> Result<(), PulseError>

DELETE /api/pulse/ml-models/{name} — remove a model (ADMIN).

Trait Implementations

impl<'c> Debug for ModelsResource<'c>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.