共用方式為


ImageModelDistributionSettings interface

分散運算式,以掃掠模型設定的值。 一些範例包括:

ModelName = "choice('seresnext', 'resnest50')";
LearningRate = "uniform(0.001, 0.01)";
LayersToFreeze = "choice(0, 2)";
```</example>
All distributions can be specified as distribution_name(min, max) or choice(val1, val2, ..., valn)
where distribution name can be: uniform, quniform, loguniform, etc
For more details on how to compose distribution expressions please check the documentation:
https://docs.microsoft.com/en-us/azure/machine-learning/how-to-tune-hyperparameters
For more information on the available settings please visit the official documentation:
https://docs.microsoft.com/en-us/azure/machine-learning/how-to-auto-train-image-models.

屬性

amsGradient

當優化器為 『adam』 或 'adamw' 時啟用AMSGrad。

augmentations

使用擴增的設定。

beta1

優化器為 'adam' 或 'adamw' 時的 'beta1' 值。 必須是範圍 [0, 1] 中的浮點數。

beta2

優化器為 'adam' 或 'adamw' 時的 'beta2' 值。 必須是範圍 [0, 1] 中的浮點數。

distributed

是否要使用散發器定型。

earlyStopping

在定型期間啟用早期停止邏輯。

earlyStoppingDelay

追蹤主要計量改善之前,要等待的最小 Epoch 或驗證評估數目,以便進行早期停止。 必須是正整數。

earlyStoppingPatience

停止執行之前,沒有主要計量改進的最小 Epoch 或驗證評估數目。 必須是正整數。

enableOnnxNormalization

匯出 ONNX 模型時啟用正規化。

evaluationFrequency

評估驗證數據集以取得計量分數的頻率。 必須是正整數。

gradientAccumulationStep

漸層累積表示執行已設定數目的 「GradAccumulationStep」 步驟,而不更新模型權數,同時累積這些步驟的漸層,然後使用累積的漸層來計算權數更新。 必須是正整數。

layersToFreeze

要凍結模型的圖層數目。 必須是正整數。 例如,將 2 當做 'seresnext' 的值傳遞表示凍結 layer0 和 layer1。 如需支援模型的完整清單,以及層次凍結的詳細資料,請參閱:https://docs.microsoft.com/en-us/azure/machine-learning/how-to-auto-train-image-models

learningRate

初始學習速率。 必須是範圍 [0, 1] 中的浮點數。

learningRateScheduler

學習速率排程器的類型。 必須是 『warmup_cosine』 或 'step'。

modelName

要用於定型的模型名稱。 如需可用模型的詳細資訊,請流覽官方檔:https://docs.microsoft.com/en-us/azure/machine-learning/how-to-auto-train-image-models

momentum

優化器為 『zn』 時的動量值。 必須是範圍 [0, 1] 中的浮點數。

nesterov

在優化器為 『zn』 時啟用 nesterov。

numberOfEpochs

定型 Epoch 的數目。 必須是正整數。

numberOfWorkers

數據載入器背景工作角色的數目。 必須是非負整數。

optimizer

優化工具的類型。 必須是 『zn』、『adam』或 『adamw』。

randomSeed

使用決定性定型時要使用的隨機種子。

stepLRGamma

學習速率排程器為 『step』 時的 gamma 值。 必須是範圍 [0, 1] 中的浮點數。

stepLRStepSize

學習速率排程器為 『step』 時的步驟大小值。 必須是正整數。

trainingBatchSize

定型批次大小。 必須是正整數。

validationBatchSize

驗證批次大小。 必須是正整數。

warmupCosineLRCycles

學習速率排程器為「warmup_cosine」時,餘弦迴圈的值。 必須是範圍 [0, 1] 中的浮點數。

warmupCosineLRWarmupEpochs

學習速率排程器為「warmup_cosine」時,熱身 Epoch 的值。 必須是正整數。

weightDecay

優化器為 'zn'、'adam' 或 'adamw' 時的權數衰變值。 必須是範圍[0, 1] 中的浮點數。

屬性詳細資料

amsGradient

當優化器為 『adam』 或 'adamw' 時啟用AMSGrad。

amsGradient?: string

屬性值

string

augmentations

使用擴增的設定。

augmentations?: string

屬性值

string

beta1

優化器為 'adam' 或 'adamw' 時的 'beta1' 值。 必須是範圍 [0, 1] 中的浮點數。

beta1?: string

屬性值

string

beta2

優化器為 'adam' 或 'adamw' 時的 'beta2' 值。 必須是範圍 [0, 1] 中的浮點數。

beta2?: string

屬性值

string

distributed

是否要使用散發器定型。

distributed?: string

屬性值

string

earlyStopping

在定型期間啟用早期停止邏輯。

earlyStopping?: string

屬性值

string

earlyStoppingDelay

追蹤主要計量改善之前,要等待的最小 Epoch 或驗證評估數目,以便進行早期停止。 必須是正整數。

earlyStoppingDelay?: string

屬性值

string

earlyStoppingPatience

停止執行之前,沒有主要計量改進的最小 Epoch 或驗證評估數目。 必須是正整數。

earlyStoppingPatience?: string

屬性值

string

enableOnnxNormalization

匯出 ONNX 模型時啟用正規化。

enableOnnxNormalization?: string

屬性值

string

evaluationFrequency

評估驗證數據集以取得計量分數的頻率。 必須是正整數。

evaluationFrequency?: string

屬性值

string

gradientAccumulationStep

漸層累積表示執行已設定數目的 「GradAccumulationStep」 步驟,而不更新模型權數,同時累積這些步驟的漸層,然後使用累積的漸層來計算權數更新。 必須是正整數。

gradientAccumulationStep?: string

屬性值

string

layersToFreeze

要凍結模型的圖層數目。 必須是正整數。 例如,將 2 當做 'seresnext' 的值傳遞表示凍結 layer0 和 layer1。 如需支援模型的完整清單,以及層次凍結的詳細資料,請參閱:https://docs.microsoft.com/en-us/azure/machine-learning/how-to-auto-train-image-models

layersToFreeze?: string

屬性值

string

learningRate

初始學習速率。 必須是範圍 [0, 1] 中的浮點數。

learningRate?: string

屬性值

string

learningRateScheduler

學習速率排程器的類型。 必須是 『warmup_cosine』 或 'step'。

learningRateScheduler?: string

屬性值

string

modelName

要用於定型的模型名稱。 如需可用模型的詳細資訊,請流覽官方檔:https://docs.microsoft.com/en-us/azure/machine-learning/how-to-auto-train-image-models

modelName?: string

屬性值

string

momentum

優化器為 『zn』 時的動量值。 必須是範圍 [0, 1] 中的浮點數。

momentum?: string

屬性值

string

nesterov

在優化器為 『zn』 時啟用 nesterov。

nesterov?: string

屬性值

string

numberOfEpochs

定型 Epoch 的數目。 必須是正整數。

numberOfEpochs?: string

屬性值

string

numberOfWorkers

數據載入器背景工作角色的數目。 必須是非負整數。

numberOfWorkers?: string

屬性值

string

optimizer

優化工具的類型。 必須是 『zn』、『adam』或 『adamw』。

optimizer?: string

屬性值

string

randomSeed

使用決定性定型時要使用的隨機種子。

randomSeed?: string

屬性值

string

stepLRGamma

學習速率排程器為 『step』 時的 gamma 值。 必須是範圍 [0, 1] 中的浮點數。

stepLRGamma?: string

屬性值

string

stepLRStepSize

學習速率排程器為 『step』 時的步驟大小值。 必須是正整數。

stepLRStepSize?: string

屬性值

string

trainingBatchSize

定型批次大小。 必須是正整數。

trainingBatchSize?: string

屬性值

string

validationBatchSize

驗證批次大小。 必須是正整數。

validationBatchSize?: string

屬性值

string

warmupCosineLRCycles

學習速率排程器為「warmup_cosine」時,餘弦迴圈的值。 必須是範圍 [0, 1] 中的浮點數。

warmupCosineLRCycles?: string

屬性值

string

warmupCosineLRWarmupEpochs

學習速率排程器為「warmup_cosine」時,熱身 Epoch 的值。 必須是正整數。

warmupCosineLRWarmupEpochs?: string

屬性值

string

weightDecay

優化器為 'zn'、'adam' 或 'adamw' 時的權數衰變值。 必須是範圍[0, 1] 中的浮點數。

weightDecay?: string

屬性值

string