def test07_topk_pool(self):
set_log_level(2)
data_opt = self.get_ml100k_mm_opt()
opt = BPRMFOption().get_default_option()
opt.d = 20
opt.num_workers = 1
model = BPRMF(opt, data_opt=data_opt)
model.initialize()
model.train()
par = ParBPRMF(model)
pool = np.array([i for i in range(5)], dtype=np.int32)
model.build_userid_map()
all_keys = model._idmanager.userids[::][:10]
naive = model.topk_recommendation(all_keys, topk=10, pool=pool)
qkeys1, topks1, scores1 = par.topk_recommendation(all_keys, topk=10, pool=pool, repr=True)
for q, t in zip(qkeys1, topks1):
self.assertEqual(naive[q], t)
def test03_most_similar(self):
set_log_level(1)
data_opt = self.get_ml100k_mm_opt()
opt = BPRMFOption().get_default_option()
opt.d = 20
opt.num_workers = 1
bpr = BPRMF(opt, data_opt=data_opt)
bpr.initialize()
bpr.train()
bpr.build_itemid_map()
parbpr = ParBPRMF(bpr)
all_keys = bpr._idmanager.itemids[::]
start_t = time.time()
[bpr.most_similar(k, topk=10) for k in all_keys]
naive_elapsed = time.time() - start_t
parbpr.num_workers = 4
start_t = time.time()
parbpr.most_similar(all_keys, topk=10, repr=True)
parbpr_elapsed = time.time() - start_t
self.assertTrue(naive_elapsed > parbpr_elapsed * 3.0)