def load_model(trainf, model_name, weights_file):
params = []
module = importlib.import_module('.' + modelname, 'models')
conf, ps, h = anssel_train.config(module.config, params)
s0, s1, y, vocab, gr = anssel_train.load_set(trainf)
model = anssel_train.build_model(glove, vocab, module.prep_model, conf)
model.load_weights(weights_file)
return model, vocab
def load_model(trainf, model_name, weights_file):
params = []
module = importlib.import_module('.'+modelname, 'models')
conf, ps, h = anssel_train.config(module.config, params)
s0, s1, y, vocab, gr = anssel_train.load_set(trainf)
model = anssel_train.build_model(glove, vocab, module.prep_model, conf)
model.load_weights(weights_file)
return model, vocab
last_is0 = ''
for is0, is1, iy in zip(s0, s1, y):
if hash(tuple(is0)) != last_is0:
last_is0 = hash(tuple(is0))
m = 0
n += 1
print('%d 0 %d 1 %f %s' % (n, m, iy, code), file=f)
m += 1
if __name__ == "__main__":
modelname, weightsfile, trainf, valf, trec_qrels_file, trec_top_file = sys.argv[1:7]
params = sys.argv[7:]
module = importlib.import_module('.'+modelname, 'models')
conf, ps, h = anssel_train.config(module.config, params)
print('GloVe')
glove = emb.GloVe(N=conf['embdim'])
print('Dataset')
s0, s1, y, vocab, gr = anssel_train.load_set(trainf)
s0t, s1t, yt, _, grt = anssel_train.load_set(valf, vocab)
print('Model')
model = anssel_train.build_model(glove, vocab, module.prep_model, conf)
print('Weights')
model.load_weights(weightsfile)
print('Prediction')
cnninit=['glorot_uniform', 'glorot_uniform', 'normal'],
cdim={
1: [0, 0, 1 / 2, 1, 2],
2: [0, 0, 1 / 2, 1, 2, 0],
3: [0, 0, 1 / 2, 1, 2, 0],
4: [0, 0, 1 / 2, 1, 2, 0],
5: [0, 0, 1 / 2, 1, 2]
},
project=[True, True, False],
pdim=[1, 2, 2.5, 3],
ptscorer=[B.mlp_ptscorer],
Ddim=[1, 2, 2.5, 3])
for ps, h, pardict in rs():
print(' ...... %s .................... %s' % (h, ps))
conf, ps, h = anssel_train.config(module.config, ps)
model = anssel_train.build_model(glove, vocab, module.prep_model, conf)
runid = '%s_%x' % (modelname, h)
model.fit(gr,
validation_data=grt,
callbacks=[
AnsSelCB(s0t, grt),
ModelCheckpoint('weights-' + runid + '.h5',
save_best_only=True,
monitor='mrr',
mode='max'),
EarlyStopping(monitor='mrr', mode='max', patience=1)
],
batch_size=160,
nb_epoch=16,
module = importlib.import_module('.'+modelname, 'models')
s0, s1, y, vocab, gr = anssel_train.load_set(trainf)
s0t, s1t, yt, _, grt = anssel_train.load_set(valf, vocab)
glove = emb.GloVe(300) # XXX hardcoded N
rs = RandomSearch(modelname+'_rlog.txt',
dropout=[1/2, 2/3, 3/4], inp_e_dropout=[1/2, 3/4, 4/5], l2reg=[1e-4, 1e-3, 1e-2],
cnnact=['tanh', 'tanh', 'relu'], cnninit=['glorot_uniform', 'glorot_uniform', 'normal'],
cdim={1: [0,0,1/2,1,2], 2: [0,0,1/2,1,2,0], 3: [0,0,1/2,1,2,0], 4: [0,0,1/2,1,2,0], 5: [0,0,1/2,1,2]},
project=[True, True, False], pdim=[1, 2, 2.5, 3],
ptscorer=[B.mlp_ptscorer], Ddim=[1, 2, 2.5, 3])
for ps, h, pardict in rs():
conf, ps, h = anssel_train.config(module.config, [])
for k, v in pardict.items():
conf[k] = v
ps, h = hash_params(conf)
runid = '%s_%x' % (modelname, h)
print(' ...... %x .................... %s' % (h, ps))
try:
model = anssel_train.build_model(glove, vocab, module.prep_model, conf)
model.fit(gr, validation_data=grt,
callbacks=[AnsSelCB(s0t, grt),
ModelCheckpoint('weights-'+runid+'.h5', save_best_only=True, monitor='mrr', mode='max'),
EarlyStopping(monitor='mrr', mode='max', patience=4)],
batch_size=160, nb_epoch=16, samples_per_epoch=int(len(s0)/4))
# mrr = max(hist.history['mrr'])
model.load_weights('weights-'+runid+'.h5')