package mint.protos

Get desktop application:
View/edit binary Protocol Buffers messages

Configuration message for a constant learning rate.

Used in: LearningRate

• optional float learning_rate = 1

• optional int32 hidden_size = 1

• optional int32 kernel_size = 2

Configuration message for a cosine decaying learning rate as defined in https://github.com/tensorflow/models/blob/master/research/object_detection/utils/learning_schedules.py

Used in: LearningRate

• optional float learning_rate_base = 1

• optional uint32 total_steps = 2

• optional float warmup_learning_rate = 3

• optional uint32 warmup_steps = 4

• optional uint32 hold_base_rate_steps = 5

Used in: FACTModel

• optional string modality_a = 1

  The cross modal function.

• optional string modality_b = 2

• oneof model

  • Transformer transformer = 3

  • MLP mlp = 4

• optional CrossModalModel.CrossModalConcatDim cross_modal_concat_dim = 5

• optional MLP output_layer = 6

  Output MLP layer config.

• optional CrossModalModel.Preprocess preprocess = 7

Cross modal concatenate dimension

Used in: CrossModalModel

• DEFAULT_CONCAT = 0

• SEQUENCE_WISE = 1

• CHANNEL_WISE = 2

Preprocess each feature if necessary.

Used in: CrossModalModel

• DEFAULT_NONE = 0

• CONTRASTIVE = 1

  Prepare feature for mutual information loss.

Used in: Dataset

• oneof modality

  • GeneralModality general_modality = 2

Used in: TrainEvalPipelineConfig

• optional string name = 1

  Dataset name.

• optional string data_files = 2

  Path to the sstables.

• optional Dataset.WindowType window_type = 3

• optional string data_target_field = 4

  If the data_target_field is set, this is what will be parsed off of the input data to be used as the target.

• optional bool create_bert_masks = 5

  Params for creating BERT pretraining mask.

• optional Dataset.BERTMaskType bert_mask_type = 6

• repeated Preprocessor data_augmentation_options = 7

  Data augmentation options.

• optional bool sample_window = 8

  Window sampling Flag for whether to do the window sampling If this flag is set to false, then data_target_field must be set.

• optional int32 target_num_categories = 9

  How many output categories are in the input data. Required when the data_target_field is set.

• repeated DataModality modality = 10

• optional float input_length_sec = 11

  Input length in seconds

• optional float target_length_sec = 12

  Target length in seconds

• optional float target_shift_sec = 13

• optional float length_threshold_sec = 14

  Filter length in seconds

Used in: Dataset

• DEFAULT_MASK = 0

• CONTIGUOUS = 1

Used in: Dataset

• DEFAULT_WINDOW = 0

• BEGINNING = 1

• CENTER = 2

• RANDOM = 3

Message for configuring multi_modal cbt evaluation jobs.

Used in: TrainEvalPipelineConfig

• optional int32 batch_size = 1

• optional EvalMetric eval_metric = 2

  Metrics Options

Used in: EvalConfig

• oneof metric_oneof

  • MotionPredictionMetrics motion_prediction_metrics = 1

  • MotionGenerationMetrics motion_generation_metrics = 2

Configuration message for an exponentially decaying learning rate. See https://www.tensorflow.org/versions/master/api_docs/python/train/ \ decaying_the_learning_rate#exponential_decay

Used in: LearningRate

• optional float initial_learning_rate = 1

• optional uint32 decay_steps = 2

• optional float decay_factor = 3

• optional bool staircase = 4

• optional float burnin_learning_rate = 5

• optional uint32 burnin_steps = 6

• optional float min_learning_rate = 7

Used in: MultiModalModel

• repeated Modality modality = 1

• optional CrossModalModel cross_modal_model = 2

• optional string fk_path = 3

A preprocessor for FACT model with AIST++ dataset.

Used in: ModalityPreprocessor, Preprocessor

(message has no fields)

Used in: DataModality

• optional string feature_name = 1

• optional int32 dimension = 2

• optional int32 sample_rate = 3

  How many data samples there are per second for this modality

• optional int32 resize = 4

  Options for images.

• optional int32 crop_size = 5

Configuration message for optimizer learning rate.

Used in: TrainConfig

• oneof learning_rate

  • ConstantLearningRate constant_learning_rate = 1

  • ExponentialDecayLearningRate exponential_decay_learning_rate = 2

  • ManualStepLearningRate manual_step_learning_rate = 3

  • CosineDecayLearningRate cosine_decay_learning_rate = 4

Used in: CrossModalModel, ModalityModel

• optional float initializer_range = 1

• optional string hidden_act = 2

• optional int32 out_dim = 3

  Output dimension.

Configuration message for a manually defined learning rate schedule.

Used in: LearningRate

• optional float initial_learning_rate = 1

• repeated ManualStepLearningRate.LearningRateSchedule schedule = 2

• optional bool warmup = 3

  Whether to linearly interpolate learning rates for steps in [0, schedule[0].step].

Used in: ManualStepLearningRate

• optional uint32 step = 1

• optional float learning_rate = 2

Used in: FACTModel

• optional string feature_name = 1

• optional int32 feature_dim = 2

• optional int32 sequence_length = 3

• optional ModalityInputConfig input_config = 4

• repeated ModalityPreprocessor preprocessor = 5

• repeated ModalityModel model = 6

Used in: Modality

• optional bool use_look_ahead_mask = 2

Used in: Modality

• oneof model

  • Transformer transformer = 1

  • MLP mlp = 2

Used in: Modality

• oneof preprocessor

  • FACTPreprocessor fact_preprocessor = 1

3D Motion Generation Metrics.

Used in: EvalMetric

• repeated float pck_thresholds = 1

• optional int32 num_joints = 2

3D Motion Prediction Metrics.

Used in: EvalMetric

• optional bool add_positional_metrics = 1

• repeated float pck_thresholds = 3

Used in: TrainEvalPipelineConfig

• oneof model

  • FACTModel fact_model = 1

Message for preprocessing the data.

Used in: Dataset

• oneof preprocessor

  • FACTPreprocessor fact_preprocessor = 1

Message for configuring multi_modal training jobs.

Used in: TrainEvalPipelineConfig

• optional int32 num_steps = 1

• optional int32 batch_size = 2

• optional bool use_bfloat16 = 3

  Whether to use bfloat16 for training (on TPU).

• optional LearningRate learning_rate = 4

  Learning rate.

• optional float grad_clip_norm = 5

• optional string fine_tune_checkpoint = 6

  Checkpoint to restore variables from.

• optional TrainConfig.CheckpointType fine_tune_checkpoint_type = 7

Options for checkpoint type. DEFAULT will load all the variables. MOTION_MODEL will only load the variables for motion model.

Used in: TrainConfig

• DEFAULT = 0

Convenience message for configuring a training and eval pipeline. Allows all of the pipeline parameters to be configured from one file.

• optional MultiModalModel multi_modal_model = 1

• optional TrainConfig train_config = 2

• optional Dataset train_dataset = 3

• optional EvalConfig eval_config = 4

• optional Dataset eval_dataset = 5

Used in: CrossModalModel, ModalityModel

• optional int32 hidden_size = 1

  Modality specific parameters.

• optional int32 num_hidden_layers = 2

• optional int32 num_attention_heads = 3

• optional int32 max_position_embeddings = 4

• optional int32 intermediate_size = 5

  Common params for transformer.

• optional string hidden_act = 6

• optional float hidden_dropout_prob = 7

• optional float attention_probs_dropout_prob = 8

• optional float initializer_range = 9

• optional string masked_loss_type = 10

• optional bool add_spatial_attention = 11

  Parameters for spatial transformer.

• optional int32 sp_hidden_size = 12

• optional int32 sp_num_attention_heads = 13

• optional int32 sp_num_hidden_layers = 14

• optional bool add_cls_token = 15

  Whether or not concat the input with a cls token.

• optional float weight_decay = 16