def test_forward_one_batch(self):
comparator = DotComparator()
lhs_pos = torch.tensor(
[[[0.8931, 0.2241, 0.4241], [0.6557, 0.2492, 0.4157]]], requires_grad=True
)
rhs_pos = torch.tensor(
[[[0.9220, 0.2892, 0.7408], [0.1476, 0.6079, 0.1835]]], requires_grad=True
)
lhs_neg = torch.tensor(
[
[
[0.3836, 0.7648, 0.0965],
[0.8929, 0.8947, 0.4877],
[0.0000, 0.0000, 0.0000],
[0.7967, 0.6736, 0.2966],
]
],
requires_grad=True,
)
rhs_neg = torch.tensor(
[
[
[0.6116, 0.6010, 0.9500],
[0.0000, 0.0000, 0.0000],
[0.2360, 0.5923, 0.7536],
[0.1290, 0.3088, 0.2731],
]
],
requires_grad=True,
)
pos_scores, lhs_neg_scores, rhs_neg_scores = comparator(
comparator.prepare(lhs_pos),
comparator.prepare(rhs_pos),
comparator.prepare(lhs_neg),
comparator.prepare(rhs_neg),
)
self.assertTensorEqual(pos_scores, torch.tensor([[1.2024, 0.3246]]))
self.assertTensorEqual(
lhs_neg_scores,
torch.tensor(
[[[0.6463, 1.4433, 0.0000, 1.1491], [0.5392, 0.7652, 0.0000, 0.5815]]]
),
)
self.assertTensorEqual(
rhs_neg_scores,
torch.tensor(
[[[1.0838, 0.0000, 0.6631, 0.3002], [0.9457, 0.0000, 0.6156, 0.2751]]]
),
)
(pos_scores.sum() + lhs_neg_scores.sum() + rhs_neg_scores.sum()).backward()
self.assertTrue((lhs_pos.grad != 0).any())
self.assertTrue((rhs_pos.grad != 0).any())
self.assertTrue((lhs_neg.grad != 0).any())
self.assertTrue((rhs_neg.grad != 0).any())
print(embedding_user_0.shape)
print(embedding_all.shape)
from torchbiggraph.model import DotComparator
src_entity_offset = dictionary["entities"]["user_id"].index("0")
dest_1_entity_offset = dictionary["entities"]["user_id"].index("7")
dest_2_entity_offset = dictionary["entities"]["user_id"].index("135")
with h5py.File("model/demo/embeddings_user_id_0.v10.h5", "r") as hf:
src_embedding = hf["embeddings"][src_entity_offset, :]
dest_1_embedding = hf["embeddings"][dest_1_entity_offset, :]
dest_2_embedding = hf["embeddings"][dest_2_entity_offset, :]
dest_embeddings = hf["embeddings"][...]
import torch
comparator = DotComparator()
scores_1, _, _ = comparator(
comparator.prepare(torch.tensor(src_embedding.reshape([1,1,520]))),
comparator.prepare(torch.tensor(dest_1_embedding.reshape([1,1,520]))),
torch.empty(1, 0, 520), # Left-hand side negatives, not needed
torch.empty(1, 0, 520), # Right-hand side negatives, not needed
)
scores_2, _, _ = comparator(
comparator.prepare(torch.tensor(src_embedding.reshape([1,1,520]))),
comparator.prepare(torch.tensor(dest_2_embedding.reshape([1,1,520]))),
torch.empty(1, 0, 520), # Left-hand side negatives, not needed
torch.empty(1, 0, 520), # Right-hand side negatives, not needed
)
print("Now let's do some simple things within torch:")
from torchbiggraph.model import DotComparator
src_entity_offset = dictionary["entities"]["user_id"].index("0")
dest_1_entity_offset = dictionary["entities"]["user_id"].index("7")
dest_2_entity_offset = dictionary["entities"]["user_id"].index("1")
with h5py.File("model/example_2/embeddings_user_id_0.v10.h5", "r") as hf:
src_embedding = hf["embeddings"][src_entity_offset, :]
dest_1_embedding = hf["embeddings"][dest_1_entity_offset, :]
dest_2_embedding = hf["embeddings"][dest_2_entity_offset, :]
dest_embeddings = hf["embeddings"][...]
import torch
comparator = DotComparator()
scores_1, _, _ = comparator(
comparator.prepare(torch.tensor(src_embedding.reshape([1,1,10]))),
comparator.prepare(torch.tensor(dest_1_embedding.reshape([1,1,10]))),
torch.empty(1, 0, 10), # Left-hand side negatives, not needed
torch.empty(1, 0, 10), # Right-hand side negatives, not needed
)
scores_2, _, _ = comparator(
comparator.prepare(torch.tensor(src_embedding.reshape([1,1,10]))),
comparator.prepare(torch.tensor(dest_2_embedding.reshape([1,1,10]))),
torch.empty(1, 0, 10), # Left-hand side negatives, not needed
torch.empty(1, 0, 10), # Right-hand side negatives, not needed
)
