File size: 16,929 Bytes
ffead1e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
# coding=utf-8
# Copyright 2023 HuggingFace Inc..
#
# 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.


import logging
import os
import shutil
import subprocess
import sys
import tempfile
import unittest
from typing import List

from accelerate.utils import write_basic_config

from diffusers import DiffusionPipeline, UNet2DConditionModel


logging.basicConfig(level=logging.DEBUG)

logger = logging.getLogger()


# These utils relate to ensuring the right error message is received when running scripts
class SubprocessCallException(Exception):
    pass


def run_command(command: List[str], return_stdout=False):
    """
    Runs `command` with `subprocess.check_output` and will potentially return the `stdout`. Will also properly capture
    if an error occurred while running `command`
    """
    try:
        output = subprocess.check_output(command, stderr=subprocess.STDOUT)
        if return_stdout:
            if hasattr(output, "decode"):
                output = output.decode("utf-8")
            return output
    except subprocess.CalledProcessError as e:
        raise SubprocessCallException(
            f"Command `{' '.join(command)}` failed with the following error:\n\n{e.output.decode()}"
        ) from e


stream_handler = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)


class ExamplesTestsAccelerate(unittest.TestCase):
    @classmethod
    def setUpClass(cls):
        super().setUpClass()
        cls._tmpdir = tempfile.mkdtemp()
        cls.configPath = os.path.join(cls._tmpdir, "default_config.yml")

        write_basic_config(save_location=cls.configPath)
        cls._launch_args = ["accelerate", "launch", "--config_file", cls.configPath]

    @classmethod
    def tearDownClass(cls):
        super().tearDownClass()
        shutil.rmtree(cls._tmpdir)

    def test_train_unconditional(self):
        with tempfile.TemporaryDirectory() as tmpdir:
            test_args = f"""
                examples/unconditional_image_generation/train_unconditional.py
                --dataset_name hf-internal-testing/dummy_image_class_data
                --model_config_name_or_path diffusers/ddpm_dummy
                --resolution 64
                --output_dir {tmpdir}
                --train_batch_size 2
                --num_epochs 1
                --gradient_accumulation_steps 1
                --ddpm_num_inference_steps 2
                --learning_rate 1e-3
                --lr_warmup_steps 5
                """.split()

            run_command(self._launch_args + test_args, return_stdout=True)
            # save_pretrained smoke test
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "unet", "diffusion_pytorch_model.bin")))
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "scheduler", "scheduler_config.json")))

    def test_textual_inversion(self):
        with tempfile.TemporaryDirectory() as tmpdir:
            test_args = f"""
                examples/textual_inversion/textual_inversion.py
                --pretrained_model_name_or_path hf-internal-testing/tiny-stable-diffusion-pipe
                --train_data_dir docs/source/en/imgs
                --learnable_property object
                --placeholder_token <cat-toy>
                --initializer_token a
                --resolution 64
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 2
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                """.split()

            run_command(self._launch_args + test_args)
            # save_pretrained smoke test
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "learned_embeds.bin")))

    def test_dreambooth(self):
        with tempfile.TemporaryDirectory() as tmpdir:
            test_args = f"""
                examples/dreambooth/train_dreambooth.py
                --pretrained_model_name_or_path hf-internal-testing/tiny-stable-diffusion-pipe
                --instance_data_dir docs/source/en/imgs
                --instance_prompt photo
                --resolution 64
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 2
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                """.split()

            run_command(self._launch_args + test_args)
            # save_pretrained smoke test
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "unet", "diffusion_pytorch_model.bin")))
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "scheduler", "scheduler_config.json")))

    def test_dreambooth_checkpointing(self):
        instance_prompt = "photo"
        pretrained_model_name_or_path = "hf-internal-testing/tiny-stable-diffusion-pipe"

        with tempfile.TemporaryDirectory() as tmpdir:
            # Run training script with checkpointing
            # max_train_steps == 5, checkpointing_steps == 2
            # Should create checkpoints at steps 2, 4

            initial_run_args = f"""
                examples/dreambooth/train_dreambooth.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --instance_data_dir docs/source/en/imgs
                --instance_prompt {instance_prompt}
                --resolution 64
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 5
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --seed=0
                """.split()

            run_command(self._launch_args + initial_run_args)

            # check can run the original fully trained output pipeline
            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(instance_prompt, num_inference_steps=2)

            # check checkpoint directories exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))

            # check can run an intermediate checkpoint
            unet = UNet2DConditionModel.from_pretrained(tmpdir, subfolder="checkpoint-2/unet")
            pipe = DiffusionPipeline.from_pretrained(pretrained_model_name_or_path, unet=unet, safety_checker=None)
            pipe(instance_prompt, num_inference_steps=2)

            # Remove checkpoint 2 so that we can check only later checkpoints exist after resuming
            shutil.rmtree(os.path.join(tmpdir, "checkpoint-2"))

            # Run training script for 7 total steps resuming from checkpoint 4

            resume_run_args = f"""
                examples/dreambooth/train_dreambooth.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --instance_data_dir docs/source/en/imgs
                --instance_prompt {instance_prompt}
                --resolution 64
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 7
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --resume_from_checkpoint=checkpoint-4
                --seed=0
                """.split()

            run_command(self._launch_args + resume_run_args)

            # check can run new fully trained pipeline
            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(instance_prompt, num_inference_steps=2)

            # check old checkpoints do not exist
            self.assertFalse(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))

            # check new checkpoints exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-6")))

    def test_text_to_image(self):
        with tempfile.TemporaryDirectory() as tmpdir:
            test_args = f"""
                examples/text_to_image/train_text_to_image.py
                --pretrained_model_name_or_path hf-internal-testing/tiny-stable-diffusion-pipe
                --dataset_name hf-internal-testing/dummy_image_text_data
                --resolution 64
                --center_crop
                --random_flip
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 2
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                """.split()

            run_command(self._launch_args + test_args)
            # save_pretrained smoke test
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "unet", "diffusion_pytorch_model.bin")))
            self.assertTrue(os.path.isfile(os.path.join(tmpdir, "scheduler", "scheduler_config.json")))

    def test_text_to_image_checkpointing(self):
        pretrained_model_name_or_path = "hf-internal-testing/tiny-stable-diffusion-pipe"
        prompt = "a prompt"

        with tempfile.TemporaryDirectory() as tmpdir:
            # Run training script with checkpointing
            # max_train_steps == 5, checkpointing_steps == 2
            # Should create checkpoints at steps 2, 4

            initial_run_args = f"""
                examples/text_to_image/train_text_to_image.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --dataset_name hf-internal-testing/dummy_image_text_data
                --resolution 64
                --center_crop
                --random_flip
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 5
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --seed=0
                """.split()

            run_command(self._launch_args + initial_run_args)

            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # check checkpoint directories exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))

            # check can run an intermediate checkpoint
            unet = UNet2DConditionModel.from_pretrained(tmpdir, subfolder="checkpoint-2/unet")
            pipe = DiffusionPipeline.from_pretrained(pretrained_model_name_or_path, unet=unet, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # Remove checkpoint 2 so that we can check only later checkpoints exist after resuming
            shutil.rmtree(os.path.join(tmpdir, "checkpoint-2"))

            # Run training script for 7 total steps resuming from checkpoint 4

            resume_run_args = f"""
                examples/text_to_image/train_text_to_image.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --dataset_name hf-internal-testing/dummy_image_text_data
                --resolution 64
                --center_crop
                --random_flip
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 7
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --resume_from_checkpoint=checkpoint-4
                --seed=0
                """.split()

            run_command(self._launch_args + resume_run_args)

            # check can run new fully trained pipeline
            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # check old checkpoints do not exist
            self.assertFalse(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))

            # check new checkpoints exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-6")))

    def test_text_to_image_checkpointing_use_ema(self):
        pretrained_model_name_or_path = "hf-internal-testing/tiny-stable-diffusion-pipe"
        prompt = "a prompt"

        with tempfile.TemporaryDirectory() as tmpdir:
            # Run training script with checkpointing
            # max_train_steps == 5, checkpointing_steps == 2
            # Should create checkpoints at steps 2, 4

            initial_run_args = f"""
                examples/text_to_image/train_text_to_image.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --dataset_name hf-internal-testing/dummy_image_text_data
                --resolution 64
                --center_crop
                --random_flip
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 5
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --use_ema
                --seed=0
                """.split()

            run_command(self._launch_args + initial_run_args)

            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # check checkpoint directories exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))

            # check can run an intermediate checkpoint
            unet = UNet2DConditionModel.from_pretrained(tmpdir, subfolder="checkpoint-2/unet")
            pipe = DiffusionPipeline.from_pretrained(pretrained_model_name_or_path, unet=unet, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # Remove checkpoint 2 so that we can check only later checkpoints exist after resuming
            shutil.rmtree(os.path.join(tmpdir, "checkpoint-2"))

            # Run training script for 7 total steps resuming from checkpoint 4

            resume_run_args = f"""
                examples/text_to_image/train_text_to_image.py
                --pretrained_model_name_or_path {pretrained_model_name_or_path}
                --dataset_name hf-internal-testing/dummy_image_text_data
                --resolution 64
                --center_crop
                --random_flip
                --train_batch_size 1
                --gradient_accumulation_steps 1
                --max_train_steps 7
                --learning_rate 5.0e-04
                --scale_lr
                --lr_scheduler constant
                --lr_warmup_steps 0
                --output_dir {tmpdir}
                --checkpointing_steps=2
                --resume_from_checkpoint=checkpoint-4
                --use_ema
                --seed=0
                """.split()

            run_command(self._launch_args + resume_run_args)

            # check can run new fully trained pipeline
            pipe = DiffusionPipeline.from_pretrained(tmpdir, safety_checker=None)
            pipe(prompt, num_inference_steps=2)

            # check old checkpoints do not exist
            self.assertFalse(os.path.isdir(os.path.join(tmpdir, "checkpoint-2")))

            # check new checkpoints exist
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-4")))
            self.assertTrue(os.path.isdir(os.path.join(tmpdir, "checkpoint-6")))