From 7604b28ff6e8293af383ae7328ea2285b3c9bba5 Mon Sep 17 00:00:00 2001 From: Alex Auvolat Date: Wed, 29 Apr 2015 16:01:13 -0400 Subject: Debug. --- hdist.py | 8 ++++---- model.py | 27 +++++++++++++++++++-------- transformers.py | 2 +- 3 files changed, 24 insertions(+), 13 deletions(-) diff --git a/hdist.py b/hdist.py index d5d7aec..3f62eb3 100644 --- a/hdist.py +++ b/hdist.py @@ -12,10 +12,10 @@ def hdist(a, b): rearth = const(6371) deg2rad = const(3.141592653589793 / 180) - lat1 = a[:, 1] * deg2rad - lon1 = a[:, 0] * deg2rad - lat2 = b[:, 1] * deg2rad - lon2 = b[:, 0] * deg2rad + lat1 = a[:, 0] * deg2rad + lon1 = a[:, 1] * deg2rad + lat2 = b[:, 0] * deg2rad + lon2 = b[:, 1] * deg2rad dlat = abs(lat1-lat2) dlon = abs(lon1-lon2) diff --git a/model.py b/model.py index 8e10e24..5caad8b 100644 --- a/model.py +++ b/model.py @@ -46,20 +46,30 @@ if __name__ == "__main__": def setup_stream(): # Load the training and test data - train = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5', which_set='train', subset=slice(0, config.train_size - config.n_valid), load_in_memory=True) + train = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5', + which_set='train', + subset=slice(0, config.train_size - config.n_valid), + load_in_memory=True) train = DataStream(train, iteration_scheme=SequentialExampleScheme(config.train_size - config.n_valid)) train = transformers.add_first_k(config.n_begin_end_pts, train) train = transformers.add_random_k(config.n_begin_end_pts, train) train = transformers.add_destination(train) - train = transformers.Select(train, ('origin_stand', 'origin_call', 'first_k_latitude', 'last_k_latitude', 'first_k_longitude', 'last_k_longitude', 'destination_latitude', 'destination_longitude')) + train = transformers.Select(train, ('origin_stand', 'origin_call', 'first_k_latitude', + 'last_k_latitude', 'first_k_longitude', 'last_k_longitude', + 'destination_latitude', 'destination_longitude')) train_stream = Batch(train, iteration_scheme=ConstantScheme(config.batch_size)) - valid = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5', which_set='train', subset=slice(config.train_size - config.n_valid, config.train_size), load_in_memory=True) + valid = H5PYDataset('/data/lisatmp3/simonet/taxi/data.hdf5', + which_set='train', + subset=slice(config.train_size - config.n_valid, config.train_size), + load_in_memory=True) valid = DataStream(valid, iteration_scheme=SequentialExampleScheme(config.n_valid)) valid = transformers.add_first_k(config.n_begin_end_pts, valid) valid = transformers.add_last_k(config.n_begin_end_pts, valid) valid = transformers.add_destination(valid) - valid = transformers.Select(valid, ('origin_stand', 'origin_call', 'first_k_latitude', 'last_k_latitude', 'first_k_longitude', 'last_k_longitude', 'destination_latitude', 'destination_longitude')) + valid = transformers.Select(valid, ('origin_stand', 'origin_call', 'first_k_latitude', + 'last_k_latitude', 'first_k_longitude', 'last_k_longitude', + 'destination_latitude', 'destination_longitude')) valid_stream = Batch(valid, iteration_scheme=ConstantScheme(1000)) return (train_stream, valid_stream) @@ -68,15 +78,15 @@ def main(): # The input and the targets x_firstk_latitude = (tensor.matrix('first_k_latitude') - data.porto_center[0]) / data.data_std[0] x_firstk_longitude = (tensor.matrix('first_k_longitude') - data.porto_center[1]) / data.data_std[1] - x_firstk = tensor.concatenate((x_firstk_latitude, x_firstk_longitude), axis=1) x_lastk_latitude = (tensor.matrix('last_k_latitude') - data.porto_center[0]) / data.data_std[0] x_lastk_longitude = (tensor.matrix('last_k_longitude') - data.porto_center[1]) / data.data_std[1] - x_lastk = tensor.concatenate((x_lastk_latitude, x_lastk_longitude), axis=1) x_client = tensor.lvector('origin_call') x_stand = tensor.lvector('origin_stand') - y = tensor.concatenate((tensor.vector('destination_latitude')[:, None], tensor.vector('destination_longitude')[:, None]), axis=1) + + y = tensor.concatenate((tensor.vector('destination_latitude')[:, None], + tensor.vector('destination_longitude')[:, None]), axis=1) # Define the model client_embed_table = LookupTable(length=config.n_clients+1, dim=config.dim_embed, name='client_lookup') @@ -87,7 +97,8 @@ def main(): # Create the Theano variables client_embed = client_embed_table.apply(x_client).flatten(ndim=2) stand_embed = stand_embed_table.apply(x_stand).flatten(ndim=2) - inputs = tensor.concatenate([x_firstk, x_lastk, client_embed, stand_embed], + inputs = tensor.concatenate([x_firstk_latitude, x_firstk_longitude, + x_lastk_latitude, x_lastk_longitude, client_embed, stand_embed], axis=1) # inputs = theano.printing.Print("inputs")(inputs) outputs = mlp.apply(inputs) diff --git a/transformers.py b/transformers.py index 13852ac..3473fd5 100644 --- a/transformers.py +++ b/transformers.py @@ -43,7 +43,7 @@ def add_random_k(k, stream): lon = at_least_k(k, x[id_latitude], True, True) loc = random.randrange(len(lat)-k+1) return (numpy.array(lat[loc:loc+k], dtype=theano.config.floatX), - numpy.array(lon[loc:loc+k], dtype=theano.config.floatX)), + numpy.array(lon[loc:loc+k], dtype=theano.config.floatX)) return Mapping(stream, random_k, ('last_k_latitude', 'last_k_longitude')) def add_last_k(k, stream): -- cgit v1.2.3