File size: 2,399 Bytes
74e8f2f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 |
# Copyright 2024 Big Vision Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Tests for bert."""
import tempfile
from big_vision import input_pipeline
from big_vision.models.proj.flaxformer import bert
from big_vision.models.proj.flaxformer import bert_test_util
import big_vision.pp.builder as pp_builder
import big_vision.pp.ops_general # pylint: disable=unused-import
import big_vision.pp.proj.flaxformer.bert_ops # pylint: disable=unused-import
import flax
import jax
import jax.numpy as jnp
import tensorflow as tf
# BERT vocabulary for testing.
_BERT_VOCAB = [
"[PAD]",
"[UNK]",
"this",
"is",
"a",
"test",
"[CLS]",
"[SEP]",
]
_TOKEN_LEN = 16
class BertTest(tf.test.TestCase):
def test_load_apply(self):
inkey = "text"
vocab_path = f"{tempfile.mkdtemp()}/vocab.txt"
with open(vocab_path, "w") as f:
f.write("\n".join(_BERT_VOCAB))
ds2, _ = input_pipeline.make_for_inference(
tf.data.Dataset.from_tensor_slices(
{inkey: tf.ragged.constant([["this is a test"]])}),
num_ex_per_process=[1],
preprocess_fn=pp_builder.get_preprocess_fn(
f"bert_tokenize(inkey='{inkey}', vocab_path='{vocab_path}', "
f"max_len={_TOKEN_LEN})"
"|keep('labels')"),
batch_size=1,
)
text = jnp.array(next(iter(ds2))["labels"])
model = bert.Model(config="base")
variables = model.init(jax.random.PRNGKey(0), text)
params = bert.load(flax.core.unfreeze(variables)["params"],
bert_test_util.create_base_checkpoint())
x, out = model.apply({"params": params}, text)
self.assertAllEqual(jax.tree_map(jnp.shape, x), (1, 768))
self.assertAllEqual(
jax.tree_map(jnp.shape, out), {
"transformed": (1, 16, 768),
"pre_logits": (1, 768),
})
if __name__ == "__main__":
tf.test.main()
|