@hoff97/tensor-js/dist/lib/onnx/nodes/constantOfShape.js

PyTorch like deep learning inferrence library

var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
    function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
    return new (P || (P = Promise))(function (resolve, reject) {
        function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
        function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
        function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
        step((generator = generator.apply(thisArg, _arguments || [])).next());
    });
};
import { CPUTensor } from '../../tensor/cpu/tensor';
import { getSize } from '../../util/shape';
import { OnnxNode } from '../node';
import { createTensor } from '../util';
// This does not support gradients right now, mainly because
// the forward pass needs to directly access the constant value
export class ConstantOfShapeNode extends OnnxNode {
    constructor(attributes, inputs, outputs, constants, onnxVersion, mode) {
        super(attributes, inputs, outputs, constants, onnxVersion, mode);
        if (onnxVersion < 11) {
            const tensor = this.getAttributeTensor('value');
            if (tensor !== null && tensor !== undefined) {
                this.tensor = createTensor(tensor);
            }
        }
    }
    forward(inputs) {
        return __awaiter(this, void 0, void 0, function* () {
            const _shape = inputs[0];
            if (this.onnxVersion < 11 && this.tensor !== undefined) {
                if (!(_shape instanceof CPUTensor)) {
                    throw new Error('ConstantOfShape needs cpu tensor as shape tensor');
                }
                const shape = new Array(_shape.size);
                for (let i = 0; i < _shape.size; i++) {
                    shape[i] = _shape.get(i);
                }
                const size = getSize(shape);
                const values = new Float32Array(size).fill(this.tensor.get(0));
                return [new CPUTensor(shape, values, this.tensor.dtype)];
            }
            throw new Error(`ConstantOfShape not implemented for onnx version ${this.onnxVersion}`);
        });
    }
    getType() {
        return 'ConstantOfShape';
    }
    delete() {
        if (this.tensor !== undefined) {
            this.tensor.delete();
        }
    }
}
//# sourceMappingURL=constantOfShape.js.map