microsoft/onnxruntime-extensions
Publicmirrored fromhttps://github.com/microsoft/onnxruntime-extensionsAvailable
test/test_gpt2tok.py
113lines · modecode
| 1 | import unittest |
| 2 | import numpy as np |
| 3 | import onnxruntime as _ort |
| 4 | |
| 5 | from pathlib import Path |
| 6 | from onnx import helper, onnx_pb as onnx_proto |
| 7 | from transformers import GPT2Tokenizer |
| 8 | from onnxruntime_customops import ( |
| 9 | onnx_op, |
| 10 | enable_custom_op, |
| 11 | PyCustomOpDef, |
| 12 | expand_onnx_inputs, |
| 13 | get_library_path as _get_library_path) |
| 14 | |
| 15 | |
| 16 | def _get_file_content(path): |
| 17 | with open(path, "rb") as file: |
| 18 | return file.read() |
| 19 | |
| 20 | |
| 21 | def _get_test_data_file(*sub_dirs): |
| 22 | test_dir = Path(__file__).parent |
| 23 | return str(test_dir.joinpath(*sub_dirs)) |
| 24 | |
| 25 | |
| 26 | def _create_test_model(): |
| 27 | nodes = [helper.make_node( |
| 28 | 'Identity', ['input_1'], ['output_1'])] |
| 29 | |
| 30 | input1 = helper.make_tensor_value_info( |
| 31 | 'input_1', onnx_proto.TensorProto.INT64, [None, None]) |
| 32 | output1 = helper.make_tensor_value_info( |
| 33 | 'output_1', onnx_proto.TensorProto.INT64, [None, None]) |
| 34 | |
| 35 | graph = helper.make_graph(nodes, 'test0', [input1], [output1]) |
| 36 | model = helper.make_model(graph, opset_imports=[helper.make_operatorsetid('', 12)]) |
| 37 | return model |
| 38 | |
| 39 | |
| 40 | def _bind_tokenizer(model, **kwargs): |
| 41 | vocab_file = kwargs["vocab_file"] |
| 42 | merges_file = kwargs["merges_file"] |
| 43 | |
| 44 | return expand_onnx_inputs( |
| 45 | model, 'input_1', |
| 46 | [helper.make_node( |
| 47 | 'GPT2Tokenizer', ['string_input'], ['input_1'], vocab=_get_file_content(vocab_file), |
| 48 | merges=_get_file_content(merges_file), name='bpetok', |
| 49 | domain='ai.onnx.contrib')], |
| 50 | [helper.make_tensor_value_info('string_input', onnx_proto.TensorProto.STRING, [None])], |
| 51 | ) |
| 52 | |
| 53 | |
| 54 | class TestGPT2Tokenizer(unittest.TestCase): |
| 55 | @classmethod |
| 56 | def setUpClass(cls): |
| 57 | tokjson = _get_test_data_file('data', 'gpt2.vocab') |
| 58 | merges = tokjson.replace('.vocab', '.merges.txt') |
| 59 | cls.tokenizer = GPT2Tokenizer(tokjson, merges) |
| 60 | |
| 61 | model = _create_test_model() |
| 62 | cls.binded_model = _bind_tokenizer(model, vocab_file=tokjson, merges_file=merges) |
| 63 | |
| 64 | @onnx_op(op_type="GPT2Tokenizer", |
| 65 | inputs=[PyCustomOpDef.dt_string], |
| 66 | outputs=[PyCustomOpDef.dt_int64]) |
| 67 | def bpe_toenizer(s): |
| 68 | # The user custom op implementation here. |
| 69 | TestGPT2Tokenizer.pyop_invoked = True |
| 70 | return np.array( |
| 71 | [TestGPT2Tokenizer.tokenizer.encode(st_) for st_ in s]) |
| 72 | |
| 73 | def _run_tokenizer(self, pyop_flag, test_sentence): |
| 74 | enable_custom_op(pyop_flag) |
| 75 | |
| 76 | so = _ort.SessionOptions() |
| 77 | so.register_custom_ops_library(_get_library_path()) |
| 78 | sess = _ort.InferenceSession(self.binded_model.SerializeToString(), so) |
| 79 | input_text = np.array([test_sentence]) |
| 80 | txtout = sess.run(None, {'string_input': input_text}) |
| 81 | |
| 82 | np.testing.assert_array_equal(txtout[0], np.array([self.tokenizer.encode(test_sentence)])) |
| 83 | del sess |
| 84 | del so |
| 85 | |
| 86 | def test_tokenizer(self): |
| 87 | TestGPT2Tokenizer.pyop_invoked = False |
| 88 | |
| 89 | self._run_tokenizer(False, "I can feel the magic, can you?") |
| 90 | self.assertFalse(TestGPT2Tokenizer.pyop_invoked) |
| 91 | |
| 92 | self._run_tokenizer(False, "Hey Cortana") |
| 93 | self._run_tokenizer(False, "你好123。david") |
| 94 | self._run_tokenizer(False, "women'thinsulate 3 button leather car co") |
| 95 | self._run_tokenizer(False, "#$%^&()!@?><L:{}\\[];',./`ǠǡǢǣǤǥǦǧǨ") |
| 96 | self._run_tokenizer(False, "ڠڡڢڣڤڥڦڧڨکڪګڬڭڮگ") |
| 97 | self._run_tokenizer(False, "⛀⛁⛂⛃⛄⛅⛆⛇⛈⛉⛊⛋⛌⛍⛎⛏") |
| 98 | |
| 99 | def test_tokenizer_pyop(self): |
| 100 | TestGPT2Tokenizer.pyop_invoked = False |
| 101 | self._run_tokenizer(True, "I can feel the magic, can you?") |
| 102 | self.assertTrue(TestGPT2Tokenizer.pyop_invoked) |
| 103 | |
| 104 | self._run_tokenizer(True, "Hey Cortana") |
| 105 | self._run_tokenizer(True, "你好123。david") |
| 106 | self._run_tokenizer(True, "women'thinsulate 3 button leather car co") |
| 107 | self._run_tokenizer(True, "#$%^&()!@?><L:{}\\[];',./`ǠǡǢǣǤǥǦǧǨ") |
| 108 | self._run_tokenizer(True, "ڠڡڢڣڤڥڦڧڨکڪګڬڭڮگ") |
| 109 | self._run_tokenizer(True, "⛀⛁⛂⛃⛄⛅⛆⛇⛈⛉⛊⛋⛌⛍⛎⛏") |
| 110 | |
| 111 | |
| 112 | if __name__ == "__main__": |
| 113 | unittest.main() |
| 114 | |