print "# Computed counts for each hypothesis & nonterminal"
# print counts
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Build up the info about the data
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L = [] # each hypothesis's cumulative likelihood to each data point
NYes = []
NTrials = []
Output = []
GroupLength = []
for concept in concept2data.keys():
data = concept2data[concept]
# compute the likelihood for all the data here
for h in hypotheses:
h.stored_likelihood = [h.compute_single_likelihood(d) for d in data]
for di in xrange(25):
ll = [sum(h.stored_likelihood[:di])
for h in hypotheses] # each likelihood
out = [
map(lambda x: 1 * h(data[di].input, x), data[di].input)
for h in hypotheses
] # each response
Return standard sampling of hypotheses on this amount of data
"""
if LOTlib.SIG_INTERRUPTED:
return None, set()
myset = standard_sample(make_hypothesis,
lambda: concept2data[concept_key][:ndata],
N=options.TOP_COUNT,
steps=options.STEPS,
show=False, save_top=None)
return concept_key, set(myset.get_all())
if __name__ == "__main__":
from LOTlib.MPI.MPI_map import MPI_unorderedmap, is_master_process
from collections import defaultdict
hypotheses = defaultdict(set)
for key, s in MPI_unorderedmap(run, itertools.product(concept2data.keys(), xrange(35), xrange(options.CHAINS)) ):
hypotheses[key].update(s)
print "# Done %s found %s hypotheses" % (key, len(s))
# for h in s:
# print key, h
import pickle
with open(options.OUT_PATH, 'w') as f:
pickle.dump(hypotheses, f)
from Model.Grammar import grammar
trees = [h.value for h in hypotheses]
nt2counts, sig2idx, prior_offset = create_counts(grammar, trees, log=None)
print "# Computed counts for each hypothesis & nonterminal"
# print counts
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Build up the info about the data
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
concepts = concept2data.keys()
NYes = []
NNo = []
# Make NYes and Nno
for c in concepts:
data = concept2data[c]
for di, d in enumerate(data[:25]): # as far as we go into the trial
for ri in xrange(len(d.input)): # each individual response
k = tuple([c, di+1, ri+1])
# assert k in human_yes and k in human_no, "*** Hmmm key %s is not in the data" % k
if k in human_yes or k in human_no:
NYes.append( human_yes[k])
NNo.append( human_no[k])
from Model.Grammar import grammar
trees = [h.value for h in hypotheses]
nt2counts, sig2idx, prior_offset = create_counts(grammar, trees, log=None)
print "# Computed counts for each hypothesis & nonterminal"
# print counts
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Build up the info about the data
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
concepts = concept2data.keys()
NYes = []
NNo = []
# Make NYes and Nno
for c in concepts:
data = concept2data[c]
for di, d in enumerate(data[:25]): # as far as we go into the trial
for ri in xrange(len(d.input)): # each individual response
k = tuple([c, di + 1, ri + 1])
# assert k in human_yes and k in human_no, "*** Hmmm key %s is not in the data" % k
if k in human_yes or k in human_no:
NYes.append(human_yes[k])
NNo.append(human_no[k])
print "# Computed counts for each hypothesis & nonterminal"
# print counts
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Build up the info about the data
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L = [] # each hypothesis's cumulative likelihood to each data point
NYes = []
NTrials = []
Output = []
GroupLength = []
for concept in concept2data.keys():
data = concept2data[concept]
# compute the likelihood for all the data here
for h in hypotheses:
h.stored_likelihood = [h.compute_single_likelihood(d) for d in data]
for di in xrange(25):
ll = [ sum(h.stored_likelihood[:di]) for h in hypotheses ] # each likelihood
out = [ map(lambda x: 1*h(data[di].input, x), data[di].input) for h in hypotheses] # each response
# This group will be as long as this number of data points
GroupLength.append( len(data[di].input) )
L.append(ll)
myset = standard_sample(make_hypothesis,
lambda: concept2data[concept_key][:ndata],
N=options.TOP_COUNT,
steps=options.STEPS,
show=False,
save_top=None)
return concept_key, set(myset.get_all())
if __name__ == "__main__":
from LOTlib.MPI.MPI_map import MPI_unorderedmap, is_master_process
from collections import defaultdict
hypotheses = defaultdict(set)
for key, s in MPI_unorderedmap(
run,
itertools.product(concept2data.keys(), xrange(35),
xrange(options.CHAINS))):
hypotheses[key].update(s)
print "# Done %s found %s hypotheses" % (key, len(s))
# for h in s:
# print key, h
import pickle
with open(options.OUT_PATH, 'w') as f:
pickle.dump(hypotheses, f)
