src/capi/impl/capi_parameters.cpp (14 lines): - line 246: // TODO: Make less ugly! - line 260: // TODO: Make less ugly! - line 274: // TODO: Make less ugly! - line 288: // TODO: Make less ugly! - line 302: // TODO: Make less ugly! - line 316: // TODO: Make less ugly! - line 330: // TODO: Make less ugly! - line 344: // TODO: Make less ugly! - line 358: // TODO: Make less ugly! - line 374: // TODO: Make less ugly! - line 388: // TODO: Make less ugly! - line 402: // TODO: Make less ugly! - line 415: // TODO: Make less ugly! - line 428: // TODO: Make less ugly! src/toolkits/object_detection/object_detector.cpp (9 lines): - line 85: // TODO: When we support alternative base models, we will have to generalize. - line 199: // TODO: What feedback can we give if the user requests a batch size that - line 686: // TODO: Should accept model_backend as an optional argument to avoid - line 922: // TODO: Move this into DarknetYOLOModelTrainer, since these heuristics are - line 942: // TODO: The original Python implementation also exposed "anchors", - line 1056: // TODO: Someday, this will all be an implementation detail of each - line 1059: // TODO: Remove this model-specific code once the inference path no longer - line 1088: // TODO: Once we adopt an asynchronous API, we can let this "double buffering" - line 1100: // TODO: Move this into object_detection::ModelTrainer once the src/toolkits/object_detection/od_darknet_yolo_model_trainer.cpp (8 lines): - line 65: // TODO: These should be exposed in a way that facilitates experimentation. - line 66: // TODO: A struct instead of a map would be nice, too. - line 72: // TODO: Have MPS path use these parameters, instead - line 113: // TODO: Someday, this will all be an implementation detail of each - line 138: // TODO: Also verify that activation shape here is [1024, 13, 13]? - line 271: // TODO: Recycle these allocations. - line 535: // TODO: Remove this method. It is only called by the base class implementation - line 554: // TODO: Dispatch augmentation to a separate thread/queue. src/toolkits/supervised_learning/supervised_learning.hpp (7 lines): - line 35: // TODO: List of todo's for this file - line 645: std::string output_type); // TODO: This should be const - line 654: // TODO: This function should be const. - line 661: // TODO: This function should be const - line 663: std::string output_type); // TODO: This should be const - line 670: // TODO: This function should be const - line 679: // TODO: This function should be const src/ml/neural_net/style_transfer/mps_style_transfer.m (6 lines): - line 285: // TODO: Change for testing - line 318: * TODO: write what needs to be done for batching - line 351: * TODO: Implement batching for multiple images - line 431: // TODO: add multiple batch sizes - line 444: // TODO: add multiple batch sizes - line 482: // TODO: make train asynchronous src/model_server/lib/extensions/model_base.hpp (5 lines): - line 41: // TODO: Clean up the relationship between the save/load interface defined here - line 43: // TODO: Remove the inheritance from ipc_object_base once the cppipc code has - line 47: // TODO: Remove this type alias once this class stops inheriting from - line 175: // TODO: Remove this vestigial macro invocation once the dependency on cppipc - line 255: // TODO: Remove this proxy subclass once the dependency on cppipc has been src/visualization/vega_renderer/TCVegaJSCanvas.m (5 lines): - line 281: // TODO - allow other font properties - line 360: // TODO apply font weights in non-macOS - line 377: // TODO should we be updating _font to reflect the system font we've fallen back to? - line 582: // TODO do we need to reset the CTM on each call here? (And reapply our CG->Canvas transformations?) - line 686: // TODO implement other alignments src/toolkits/object_detection/od_model_trainer.hpp (5 lines): - line 41: // TODO: Adopt NCHW. - line 59: // TODO: Migrate to neural_net::float_array_map - line 72: // TODO: Adopt EncodedBatch instead. - line 256: // TODO: This class should be responsible for producing the augmenter itself. - line 298: // TODO: Remove this method. Just let subclasses define the entire training src/ml/neural_net/mps_graph_layers.mm (4 lines): - line 54: // TODO: force has key - line 173: // TODO: force has key - line 227: // TODO: set node_fwd.flag to control training vs. inference mode once we have graphs for both - line 279: // TODO: Raise exception src/ml/neural_net/mps_graph_cnnmodule.mm (4 lines): - line 123: // TODO: The names should somehow be a parameter of this class. - line 184: // TODO: The names should somehow be a parameter of this class. - line 201: // TODO: The names should somehow be a parameter of this class. - line 247: // TODO: The names should somehow be a parameter of this class. src/toolkits/evaluation/metrics.cpp (4 lines): - line 137: TODO: Switch to gl_sarray and the new SDK implementation. - line 214: // TODO: Unify with the logic in supervised_learning_model_base::api_evaluate, - line 231: // TODO: Implement "report_by_class" using the standard evaluation framework. - line 260: // TODO: Should the metric implementations themselves (or src/ml/neural_net/tf_compute_context.cpp (4 lines): - line 274: // TODO: Codify priority levels? - line 291: // TODO: Returns 4GB as that makes sure default batch size is used. - line 359: // TODO: Remove resize_only by passing all the augmentation options - line 392: * TODO: Add proper arguments to create_drawing_classifier src/visualization/client/Turi Create Visualization/src/VegaContainer.swift (4 lines): - line 138: // TODO: Change if vega tooltips have log level options - line 267: // TODO: write function to check valid vega spec - line 293: // TODO: write function to check valid data spec - line 298: // TODO: write function to check valid image spec src/toolkits/supervised_learning/automatic_model_creation.cpp (4 lines): - line 204: // TODO: The original Python code path only runs the model selector on at most - line 210: // TODO: The original Python code path allowed users to specify no validation - line 249: // TODO: The original Python implementation only sampled the training data for - line 317: // TODO: Should this ultimately use std::random_device instead? src/ml/neural_net/mps_cnnmodule.mm (4 lines): - line 119: // TODO: The names should somehow be a parameter of this class. - line 165: // TODO: Images intended for clients to (optionally) read should be - line 245: // TODO: Recycle non-temporary MPSImage instances using image allocators - line 324: // TODO: Factor out more shared code from FillLossLabelsBatch and the below? src/toolkits/recsys/models/itemcf.cpp (4 lines): - line 788: // TODO: resolve these issues at the source level - line 835: // TODO - what if it has values > 32 bit? MLMultiArray only supports 32-bit ints. - line 836: // TODO - what if it's string? how to tell? can it be anything else? - line 854: // TODO BUG? Custom model doesn't seem to allow optional outputs src/toolkits/pattern_mining/fp_tree.cpp (3 lines): - line 430: // TODO Implement + Test - line 438: // TODO Test - line 451: // TODO Test src/toolkits/activity_classification/activity_classifier.hpp (3 lines): - line 23: // TODO: Move this model-spec generation code into a separate file, ideally - line 289: // TODO: Replace model_spec with a Checkpoint class that encapsulates - line 296: gl_sframe training_data_; // TODO: Avoid storing gl_sframe AND data_iterator. src/visualization/annotation/object_detection.cpp (3 lines): - line 346: // TODO: implement `cast_annotations` used with Image Saliency for OD - line 350: // TODO: implement `background_work` used with Image Saliency for OD - line 356: // TODO: implement `get_similar_items` used with Image Saliency for OD src/core/storage/sframe_interface/unity_sarray.cpp (3 lines): - line 398: // TODO not implemented yet. - line 1326: // TODO: This can be optimized by precomputing this for all zones outsize - line 2596: // TODO: Performance, instead of performing a string comparison everywhere, src/toolkits/nearest_neighbors/nearest_neighbors.cpp (3 lines): - line 492: // - there should only be a single column if this is the case. TODO: Why? - line 538: // TODO: remove 'is_dense' once the ball tree is updated to use distance components - line 554: // TODO: Clean the choice of data_type up src/toolkits/activity_classification/activity_classifier.cpp (3 lines): - line 49: // TODO: Encapsulate these network architecture parameters better, to support - line 500: // TODO: This implementation would ideally just post-process the return value of - line 1026: // TODO: Expose seed parameter publicly. src/toolkits/text/topic_model.cpp (3 lines): - line 27: * TODO: - line 113: // TODO: Only sort the largest num_words - line 222: * TODO: Refactor this and perplexity to share code. src/python/turicreate/toolkits/_image_feature_extractor.py (3 lines): - line 108: # TODO: https://github.com/apple/turicreate/issues/2672 - line 127: # TODO: Recycle the ndarray instances we're allocating below with - line 143: # TODO: Converge to NCHW everywhere. src/ml/neural_net/mlc_compute_context.mm (3 lines): - line 37: // TODO: Codify priority levels - line 55: // TODO: Clean up logic surrounding batch size and memory budget. - line 58: // TODO: When MLFoundation supports training across multiple GPUs, ensure src/toolkits/object_detection/one_shot_object_detection/one_shot_object_detector.cpp (3 lines): - line 208: /* @TODO: Split all backgrounds into as many chunks as there are cores - line 277: // TODO: Automatically infer the image column name, or throw error if you - line 282: // TODO: Call object_detector::train from here. src/ml/neural_net/mlc_utils.mm (3 lines): - line 16: // TODO: If we standardize on NCHW in the toolkit code, then we can avoid the - line 46: std::vector hwc_data(chw_array.size()); // TODO: Avoid initializing - line 54: // TODO: Investigate the number of copies a tensor endures end-to-end. src/toolkits/object_detection/od_data_iterator.cpp (3 lines): - line 163: // TODO: Validate the dictionary keys in compute_properties. Let downstream - line 394: // TODO: This heavyweight shuffle operation introduces spikes into the - line 425: // TODO: Investigate parallelizing this file I/O. src/ml/neural_net/mlc_od_backend.mm (3 lines): - line 28: // TODO: Expose these from DarknetYOLOCheckpoint. - line 92: // TODO: Someday we will remove this brittle dependency on names by - line 301: // TODO: Eventually this should also support computing loss from labels, when src/model_server/extensions/json/decoder.cpp (2 lines): - line 57: // TODO distinguish list and vector - line 126: // TODO distinguish dict from SGraph src/toolkits/text/alias.hpp (2 lines): - line 17: TODO: - line 180: // TODO: convert interface above to use the extensions methods here src/ml/neural_net/mps_graph_networks.h (2 lines): - line 61: // -- TODO: Add object detection network - line 62: // -- TODO: Add activity classifier network src/deployment/TuriCreateObjC.h (2 lines): - line 10: // @TODO: think about side data later - line 11: // @TODO: think about rating data later src/toolkits/style_transfer/st_model_trainer.hpp (2 lines): - line 67: // TODO: Should these also be smoothed? - line 235: // TODO: Style transfer backends should support both training and inference. src/core/storage/sframe_data/join_impl.hpp (2 lines): - line 12: //TODO: What happens if a join key (or part of one) is NULL? - line 110: //TODO: Perhaps combine the sframe and the join positions into a struct? src/ml/neural_net/mps_compute_context.mm (2 lines): - line 45: // TODO: Codify priority levels? - line 195: /* TODO: const float_array_map& config if needed */ src/toolkits/nearest_neighbors/lsh_neighbors.cpp (2 lines): - line 30: // TODO: can it be set by the available RAM?? - line 120: // TODO: better format src/core/system/cppipc/client/console_cancel_handler.hpp (2 lines): - line 29: // TODO: Check that statement - line 30: // TODO: Get this to take an arbitrary function src/ml/sketches/countmin.hpp (2 lines): - line 86: size_t bin = hash64(seeds[j] ^ i) % num_bins; // TODO: bit mask - line 100: size_t bin = hash64(seeds[j] ^ i) % num_bins; // TODO: bit mask src/toolkits/recsys/models/factorization_models.hpp (2 lines): - line 103: // TODO: convert interface above to use the extensions methods here - line 121: // TODO: convert interface above to use the extensions methods here src/toolkits/recsys/user_item_graph.cpp (2 lines): - line 33: * TODO: make it a general function of setting up a bipartite graph - line 76: // TODO: find a better way of initializing vertex_data src/ml/neural_net/style_transfer/mps_style_transfer_backend.mm (2 lines): - line 60: // TODO: This copy is inefficient. This should be a wrapper around NSData: - line 79: // TODO: This copy is inefficient. We can construct an NSData that wraps a src/ml/sketches/countsketch.hpp (2 lines): - line 90: size_t bin = hash64(seeds[j] ^ i) % num_bins; // TODO: bit mask - line 110: size_t bin = hash64(seeds[j] ^ i) % num_bins; // TODO: bit mask src/ml/neural_net/mlc_od_backend.hpp (2 lines): - line 38: // TODO: No, really, replace this legacy interface with one that just accepts - line 57: // TODO: Replace these with an instance of TCModelTrainerBackendGraphs? src/python/turicreate/toolkits/style_transfer/_tf_model_architecture.py (2 lines): - line 139: # TODO: Refactor Instance Norm - line 601: # TODO: take care of batch size src/toolkits/style_transfer/style_transfer_model_definition.cpp (2 lines): - line 37: // TODO: refactor code to be more readable with loops - line 972: // TODO: refactor code to be more readable with loops src/toolkits/object_detection/od_data_iterator.hpp (2 lines): - line 89: * \TODO: This should be a flex_list to accomodate integer labels! - line 162: // TODO: gl_sframe_range::end() should be a const method. src/python/turicreate/toolkits/object_detector/_tf_model_architecture.py (2 lines): - line 129: # TODO: Evaluate method to update lr in set_learning_rate() - line 604: # TODO: Include self.labels: feed_dict['label'] to handle labels from validation set src/toolkits/object_detection/object_detector.hpp (2 lines): - line 155: // TODO: Remainder of interface: predict, etc. - line 261: gl_sframe training_data_; // TODO: Avoid storing gl_sframe AND data_iterator. src/toolkits/style_transfer/style_transfer.cpp (2 lines): - line 779: // TODO: Iterator needs to support resuming from an offset. - line 934: {"content_feature", content_feature}, // TODO: refactor to take content name and style name src/visualization/server/boxes_and_whiskers.cpp (2 lines): - line 36: // TODO: eventually we should probably have an "undefined" bin - line 44: // TODO - for now, skip values with nan/inf src/visualization/client/Turi Create Visualization/src/user_interface/src/elements/Annotate/index.js (2 lines): - line 19: /* TODO: Indicator to Show how many Images were Annotated */ - line 20: /* TODO: Render Labels on Images */ src/toolkits/feature_engineering/feature_binner.cpp (2 lines): - line 67: // TODO: This next section could benefit from an iterator over floats, - line 310: //TODO: Break this into its own function src/python/turicreate/visualization/_plot.py (2 lines): - line 198: # TODO: allow autodetection of light/dark mode. - line 334: # TODO: allow autodetection of light/dark mode. src/toolkits/drawing_classifier/dc_data_iterator.cpp (2 lines): - line 121: // TODO: If gl_sframe_range::end() were a const method, we wouldn't need to - line 166: // TODO: This heavyweight shuffle operation introduces spikes into the src/visualization/server/process_wrapper.cpp (2 lines): - line 93: // TODO - error handling? - line 103: // TODO - error handling? src/visualization/server/groupby.hpp (2 lines): - line 40: // TODO -- pick the limited set of categories intelligently (n most popular - line 66: // TODO this is bad -- we need a non-const Aggregation in order to call src/toolkits/nearest_neighbors/brute_force_neighbors.cpp (2 lines): - line 517: // TODO: think about templating to make this simpler. - line 623: x[i] = DenseVector(1); // TODO: Why doesn't 0 work? src/ml/neural_net/mps_utils.mm (2 lines): - line 261: // TODO: raise exception - line 277: // TODO: raise exception src/core/storage/sframe_data/join_impl.cpp (2 lines): - line 621: // TODO: Add estimated disk and memory size to SFrames. - line 642: //TODO: for now src/toolkits/drawing_classifier/drawing_classifier.hpp (2 lines): - line 65: // TODO: Expose via forthcoming C-API checkpointing mechanism? - line 305: gl_sframe training_data_; // TODO: Avoid storing gl_sframe AND data_iterator. src/toolkits/pattern_mining/fp_growth.cpp (1 line): - line 245: // TODO: Streamline this to prevent all the copies from the various types. src/capi/impl/capi_sframe.cpp (1 line): - line 498: // TODO: add options. src/visualization/client/Turi Create Visualization/src/user_interface/src/elements/Annotate/InfiniteScroll/index.js (1 line): - line 120: // TODO: filter annotation array use `x` as index into the array src/core/storage/fileio/fileio_constants.cpp (1 line): - line 132: // TODO: Where is the right place for this? Probably not here... src/ml/neural_net/model_spec.hpp (1 line): - line 408: // TODO: Support additional layers (and further parameterize the above) as src/toolkits/style_transfer/style_transfer_data_iterator.hpp (1 line): - line 72: // TODO: gl_sframe_range::end() should be a const method. src/toolkits/supervised_learning/xgboost.cpp (1 line): - line 1256: * TODO: This code is *very* similar to the code in supervised_learning.cpp. src/toolkits/supervised_learning/linear_regression.cpp (1 line): - line 33: // TODO: List of todo's for this file src/toolkits/style_transfer/st_model_trainer.cpp (1 line): - line 105: // TODO: Ideally this would not require copying. But we should move away src/core/storage/sgraph_data/sgraph_engine.hpp (1 line): - line 198: // TODO: revisit the code when we actually have vertex groups src/core/storage/query_engine/operators/lambda_transform.hpp (1 line): - line 70: // TODO exception handling src/visualization/annotation/annotation_base.hpp (1 line): - line 108: // TODO: Potentially plumb `::google::protobuf::MessageLite` to variant src/core/storage/query_engine/operators/reduce.hpp (1 line): - line 67: // TODO make add_element take a sframe_row::row_reference instead src/toolkits/object_detection/one_shot_object_detection/util/mapping_function.hpp (1 line): - line 31: // TODO: Figure out the right failure behavior for when denominator is 0. src/visualization/server/item_frequency.cpp (1 line): - line 16: // TODO - do we need to do anything here? perhaps not. src/ml/optimization/constraints-inl.hpp (1 line): - line 21: // TODO: List of todo's for this file src/ml/neural_net/mps_layer_instance_norm_data_loader.mm (1 line): - line 210: // TODO: refactor for multiple indicies cmake/MakeLibrary.cmake (1 line): - line 66: # TODO we can link the requirements from here when target_link_libraries src/toolkits/util/float_array_serialization.hpp (1 line): - line 17: // TODO: Consider converting float_array values to flex_nd_vec, but we must src/toolkits/recsys/recsys_model_base.cpp (1 line): - line 1282: // TODO: Redo this function!!! There are tons of in-memory sections. src/toolkits/image_deep_feature_extractor/mlmodel_image_feature_extractor.mm (1 line): - line 368: // TODO: Just save this value before writing it to disk. cmake/SetupCompiler.cmake (1 line): - line 105: # TODO: replace all these with in-build macros based on standard macros in Availability.h src/ml/optimization/utils.cpp (1 line): - line 13: // TODO: List of todo's for this file src/ml/optimization/gradient_descent-inl.hpp (1 line): - line 19: // TODO: List of todo's for this file src/toolkits/pattern_mining/rule_mining.cpp (1 line): - line 16: * TODO Refactor this into subclasses + subfunctions src/ml/neural_net/mps_graph_networks.mm (1 line): - line 128: // TODO: In C++17, we can use std::map::merge to move the table entries src/model_server/extensions/grouped_sframe.hpp (1 line): - line 150: //TODO: This is what will run out of memory first when scaling up src/toolkits/supervised_learning/linear_regression_opt_interface.cpp (1 line): - line 27: // TODO: List of todo's for this file src/visualization/client/Turi Create Visualization/src/user_interface/public/index.js (1 line): - line 5: // TODO: refactor to share code with the non-browser case, src/core/system/platform/process/process_win.cpp (1 line): - line 160: //TODO: Figure out how to get error string on Windows src/visualization/server/histogram.hpp (1 line): - line 252: // TODO -- what should we assert here instead, to make sure we have enough src/model_server/lib/extensions/model_base.cpp (1 line): - line 156: // TODO: fix all warnings and make these errors. src/core/data/flexible_type/json_util.hpp (1 line): - line 14: * TODO: Fill in details src/core/storage/fileio/s3_api.cpp (1 line): - line 153: // TODO: add proxy support src/core/system/startup_teardown/startup_teardown.cpp (1 line): - line 158: // TODO: MANY MANY HEURISTICS src/toolkits/text/cgs.hpp (1 line): - line 104: // TODO: convert interface above to use the extensions methods here src/toolkits/text/alias.cpp (1 line): - line 285: // TODO: This hack should not be required. src/python/turicreate/toolkits/sound_classifier/sound_classifier.py (1 line): - line 453: # TODO: early stopping src/core/system/platform/process/process_util_win.cpp (1 line): - line 13: //TODO: This has a race condition that I'm ignoring as I don't think it will src/ml/neural_net/float_array.hpp (1 line): - line 159: // TODO: Operations such as reshape, slice, etc.? src/model_server/extensions/timeseries/timeseries.cpp (1 line): - line 349: // TODO: Temporary row_number column to do a stable sort, better solution! src/visualization/vega_renderer/TCVegaCSSCursorMap.m (1 line): - line 40: // TODO: below are not available from NSCursor, images are needed src/capi/impl/capi_wrapper_structs.hpp (1 line): - line 22: // TODO: memory pool implementations, etc. src/visualization/client/src/pipe.cpp (1 line): - line 10: // TODO send kill signal to threads src/ml/neural_net/mps_compute_context.hpp (1 line): - line 51: /* TODO: const float_array_map& config if needed */ src/deployment/AudioPreprocessing.mm (1 line): - line 14: // TODO: Make some of these parameters to the model src/ml/neural_net/model_backend.hpp (1 line): - line 37: // TODO: Accessors describing name inputs and expected shapes. src/ml/optimization/constraint_interface.hpp (1 line): - line 19: // TODO: List of todo's for this file src/visualization/server/show.cpp (1 line): - line 64: // TODO -- actually show this with the axes the user asked for src/toolkits/util/precision_recall.cpp (1 line): - line 47: // TODO: optimize this, as some of the range is already sorted src/ml/optimization/accelerated_gradient-inl.hpp (1 line): - line 21: // TODO: List of todo's for this file src/toolkits/ml_data_2/standardization-inl.hpp (1 line): - line 23: // TODO: List of todo's for this file src/toolkits/supervised_learning/linear_regression_opt_interface.hpp (1 line): - line 22: // TODO: List of todo's for this file src/toolkits/nearest_neighbors/lsh_neighbors.hpp (1 line): - line 112: // TODO: convert interface above to use the extensions methods here src/python/turicreate/_cython/cy_server.pyx (1 line): - line 111: # Now, set up the environment. TODO: move these in to actual src/toolkits/pattern_mining/rule_mining.hpp (1 line): - line 157: // Helper class for extracting rules // TODO REFACTOR src/ml/neural_net/TCMPSImageAugmenting.m (1 line): - line 54: // TODO: Investigate higher-quality downsampling? The original Python toolkit src/toolkits/drawing_classifier/dc_data_iterator.hpp (1 line): - line 25: * \TODO Factor out the shared structure for data iterators src/python/turicreate/toolkits/_pre_trained_models.py (1 line): - line 291: # @TODO: Think about how to bypass the md5 checksum if the user wants to src/visualization/server/server.cpp (1 line): - line 597: const static size_t NUM_THREADS = 6; // handle requests on 6 threads - TODO optimize this number src/visualization/vega_renderer/TCVegaHTMLElement.m (1 line): - line 12: // TODO: modifications to the cursor attribute should actually be applied src/ml/neural_net/mps_od_backend.mm (1 line): - line 50: // TODO: Remove this eager construction once we stop putting weak pointers in src/core/storage/sframe_data/sframe_rows.hpp (1 line): - line 45: * \TODO: We *could* templatize this around the column type, allowing this to src/python/turicreate/toolkits/recommender/item_content_recommender.py (1 line): - line 242: # TODO: automatically tune this. src/core/system/startup_teardown/startup_teardown.hpp (1 line): - line 52: * TODO This can be more intelligent as required. For now, it is kinda dumb. src/ml/neural_net/combine_iterator.hpp (1 line): - line 121: // TODO: Use an Optional type instead. src/ml/neural_net/mps_networks.mm (1 line): - line 126: // TODO: In C++17, we can use std::map::merge to move the table entries src/visualization/server/table.cpp (1 line): - line 181: // TODO - not sure what to do here. src/visualization/server/vega_spec.cpp (1 line): - line 60: // TODO: optimize this. src/ml/neural_net/mps_graph_cnnmodule.h (1 line): - line 45: // TODO: Parameters such as graph mode should be more explicit than just a src/toolkits/supervised_learning/supervised_learning.cpp (1 line): - line 33: // TODO: List of todo's for this file src/ml/optimization/regularizers-inl.hpp (1 line): - line 20: // TODO: List of todo's for this file src/python/turicreate/meta/asttools/visitors/copy_tree.py (1 line): - line 17: # FIXME: add tests src/toolkits/object_detection/one_shot_object_detection/one_shot_object_detector.hpp (1 line): - line 27: // TODO: augment -> train src/visualization/server/dark_mode.cpp (1 line): - line 26: // TODO - what should this do on other platforms? src/core/storage/sframe_interface/unity_sarray.hpp (1 line): - line 162: * TODO: will slowly move away all users of this function to get_lazy_sarray src/toolkits/drawing_classifier/data_preparation.cpp (1 line): - line 338: // @TODO: When we make a dylib, we'll need some sophisticated src/ml/neural_net/compute_context.hpp (1 line): - line 174: /* TODO: const float_array_map& config if needed */ src/toolkits/activity_classification/ac_data_iterator.cpp (1 line): - line 681: // TODO: If gl_sframe_range::end() were a const method, we wouldn't need to src/toolkits/recsys/models/popularity.hpp (1 line): - line 47: #pragma clang diagnostic ignored "-Woverloaded-virtual" // TODO: fix this issue src/ml/neural_net/mlc_dc_backend.mm (1 line): - line 82: // TODO: Someday we will remove this brittle dependency on names by src/toolkits/supervised_learning/supervised_learning_utils-inl.hpp (1 line): - line 316: // TODO: Check that k exists in eval_map. src/python/turicreate/toolkits/drawing_classifier/drawing_classifier.py (1 line): - line 166: # @TODO: Should be able to automatically choose number of iterations src/core/storage/query_engine/algorithm/ec_sort.cpp (1 line): - line 96: // TODO: yes 20 is a magic number. src/core/system/platform/network/net_util.cpp (1 line): - line 302: // TODO: This is broken on Windows, but is only used by RPC, so putting it off src/ml/optimization/regularizer_interface.hpp (1 line): - line 19: // TODO: List of todo's for this file src/toolkits/supervised_learning/standardization-inl.hpp (1 line): - line 23: // TODO: List of todo's for this file src/visualization/client/Turi Create Visualization/src/user_interface/src/elements/Annotate/ErrorBar/index.js (1 line): - line 6: /* TODO: Design Better Error Message Compornent */ src/ml/neural_net/mps_layers.h (1 line): - line 463: // TODO: add free all matrices and array. Mostly copy_weight_matrices_ src/toolkits/drawing_classifier/drawing_classifier.cpp (1 line): - line 776: /** TODO: Figure out a better solution to having `num_samples` be a src/ml/optimization/optimization_interface.hpp (1 line): - line 15: // TODO: List of todo's for this file src/ml/neural_net/model_spec.cpp (1 line): - line 526: // TODO: Parse from an input stream, without loading entire file into memory src/ml/optimization/newton_method-inl.hpp (1 line): - line 20: // TODO: List of todo's for this file src/ml/neural_net/combine_map.hpp (1 line): - line 98: // TODO: Define Optional. (Or require C++17 for std::optional?) src/core/system/cppipc/client/console_cancel_handler_unix.cpp (1 line): - line 36: //TODO: Make specific errors available src/core/system/lambda/python_callbacks.cpp (1 line): - line 34: // TODO: fill this out. This should get much more sophisticated. src/core/logging/backtrace.cpp (1 line): - line 53: //TODO: Implement this functionality in Windows. src/toolkits/nearest_neighbors/ball_tree_neighbors.hpp (1 line): - line 156: // TODO: convert interface above to use the extensions methods here src/visualization/client/Turi Create Visualization/src/user_interface/build/index.js (1 line): - line 5: // TODO: refactor to share code with the non-browser case, src/python/turicreate/util/_cloudpickle/_cloudpickle_fast.py (1 line): - line 604: # TODO: decorrelate reducer_override (which is tied to CPython's src/core/logging/logger.hpp (1 line): - line 604: // TODO: previously, we had a check for if (endltype(f) == endltype(std::endl)) src/toolkits/nearest_neighbors/brute_force_neighbors.hpp (1 line): - line 155: // TODO: convert interface above to use the extensions methods here src/toolkits/pattern_mining/fp_growth.hpp (1 line): - line 33: * TODO: Add some comments hare about what the base class is supposed to be. src/core/storage/query_engine/planning/optimizations/optimization_transforms.cpp (1 line): - line 109: // TODO: Merge all logical_filter transforms that have identical masks. src/toolkits/feature_engineering/transform_utils.hpp (1 line): - line 722: * TODO: this should src/visualization/server/categorical_heatmap.cpp (1 line): - line 24: // TODO - do we need to do anything here? perhaps not. src/ml/neural_net/image_augmentation.hpp (1 line): - line 190: // TODO: The semantics below are adopted from Core Image. src/toolkits/evaluation/evaluation_interface-inl.hpp (1 line): - line 15: #pragma clang diagnostic ignored "-Woverloaded-virtual" // TODO: fix these issues below src/toolkits/clustering/kmeans.hpp (1 line): - line 414: // TODO: convert interface above to use the extensions methods here src/core/system/cppipc/client/console_cancel_handler_win.cpp (1 line): - line 32: //TODO: Why don't I have to take the address of cancel_handler? src/ml/optimization/line_search-inl.hpp (1 line): - line 18: // TODO: List of todo's for this file src/toolkits/factorization/als.hpp (1 line): - line 20: // TODO: List of todo's for this file src/ml/neural_net/tf_compute_context.hpp (1 line): - line 47: /* TODO: const float_array_map& config if needed */ src/model_server/extensions/grouped_sframe.cpp (1 line): - line 79: //TODO: This is a HUGE hack. From the Python side, a list of ints is turned src/toolkits/object_detection/od_yolo.cpp (1 line): - line 56: // TODO: Add a mutable float_array interface so we can validate size. src/core/storage/query_engine/algorithm/sort.cpp (1 line): - line 350: // TODO: Estimate the size of the sframe so that we could decide number of src/ml/ml_data/ml_data_column_modes.cpp (1 line): - line 90: << std::endl; // TODO: finish src/toolkits/style_transfer/style_transfer.hpp (1 line): - line 28: // TODO: Move these helper functions to st_model_trainer.cpp