File size: 13,877 Bytes
de4ade4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Copyright 2022 MosaicML LLM Foundry authors
# SPDX-License-Identifier: Apache-2.0

import logging
import os
import warnings
from typing import Any, Dict, List, Optional, Tuple, Union

import torch
from composer import algorithms
from composer.callbacks import (EarlyStopper, Generate, LRMonitor,
                                MemoryMonitor, OptimizerMonitor,
                                RuntimeEstimator, SpeedMonitor)
from composer.core import Algorithm, Callback, Evaluator
from composer.datasets.in_context_learning_evaluation import \
    get_icl_task_dataloader
from composer.loggers import (InMemoryLogger, LoggerDestination, MLFlowLogger,
                              TensorboardLogger, WandBLogger)
from composer.optim import DecoupledAdamW
from composer.optim.scheduler import (ComposerScheduler,
                                      ConstantWithWarmupScheduler,
                                      CosineAnnealingWithWarmupScheduler,
                                      LinearWithWarmupScheduler)
from composer.utils import dist
from omegaconf import DictConfig, ListConfig
from omegaconf import OmegaConf as om
from torch.optim.optimizer import Optimizer
from transformers import AutoTokenizer, PreTrainedTokenizerBase

from llmfoundry.callbacks import (EvalGauntlet, FDiffMetrics, GlobalLRScaling,
                                  HuggingFaceCheckpointer, LayerFreezing,
                                  MonolithicCheckpointSaver,
                                  ScheduledGarbageCollector)
from llmfoundry.optim import (DecoupledAdaLRLion, DecoupledClipLion,
                              DecoupledLionW, DecoupledLionW_8bit)
from llmfoundry.optim.scheduler import InverseSquareRootWithWarmupScheduler
from llmfoundry.tokenizers.tiktoken import TiktokenTokenizerWrapper

log = logging.getLogger(__name__)


def build_icl_data_and_gauntlet(
    icl_tasks_config: Union[str, ListConfig],
    eval_gauntlet_config: Optional[Union[str, DictConfig]],
    tokenizer: PreTrainedTokenizerBase,
    device_eval_batch_size: int,
    icl_seq_len: int,
    icl_subset_num_batches: Optional[int] = None
) -> Tuple[List[Evaluator], List[str], Optional[EvalGauntlet]]:
    icl_evaluators, logger_keys = build_icl_evaluators(
        icl_tasks_config,
        tokenizer,
        icl_seq_len,
        device_eval_batch_size,
        icl_subset_num_batches=icl_subset_num_batches)
    eval_gauntlet_cb = None
    if eval_gauntlet_config is not None:
        if isinstance(eval_gauntlet_config, str):
            with open(eval_gauntlet_config, 'r') as icl_f:
                eval_gauntlet_cfg = om.load(icl_f)
            eval_gauntlet = eval_gauntlet_cfg.eval_gauntlet
        elif isinstance(eval_gauntlet_config, DictConfig):  # pyright: ignore
            eval_gauntlet = eval_gauntlet_config
        else:
            raise ValueError(
                f'Got invalid type for eval_gauntlet_config: {type(eval_gauntlet_config)}'
            )
        eval_gauntlet.logger_keys = logger_keys
        eval_gauntlet.benchmark_sizes = {
            e.label: e.dataloader.num_samples for e in icl_evaluators
        }
        eval_gauntlet_cb = EvalGauntlet(**eval_gauntlet)
    return icl_evaluators, logger_keys, eval_gauntlet_cb


def build_callback(name: str, kwargs: Dict[str, Any]) -> Callback:
    if name == 'lr_monitor':
        return LRMonitor()
    elif name == 'memory_monitor':
        return MemoryMonitor()
    elif name == 'speed_monitor':
        return SpeedMonitor(window_size=kwargs.get('window_size', 1),
                            gpu_flops_available=kwargs.get(
                                'gpu_flops_available', None))
    elif name == 'fdiff':
        return FDiffMetrics(**kwargs)
    elif name == 'runtime_estimator':
        return RuntimeEstimator()
    elif name == 'optimizer_monitor':
        return OptimizerMonitor(log_optimizer_metrics=kwargs.get(
            'log_optimizer_metrics', True),)
    elif name == 'generate_callback':
        prompts = kwargs.pop('prompts')
        interval = kwargs.pop('interval', None)
        # Generate callback used to be batch_log_interval, so this is for backwards compatibility
        if interval is None:
            batch_log_interval: str = kwargs.pop('batch_log_interval', '')
            if batch_log_interval:
                interval = f'{batch_log_interval}ba'
                warnings.warn(
                    ('generate_callback.batch_log_interval is deprecated and will be removed in a future release.'
                     f'Please use interval: {interval}'),
                    DeprecationWarning,
                )
            else:
                raise KeyError(
                    '"interval" must be specified with generate callback')
        return Generate(prompts=list(prompts), interval=interval, **kwargs)
    elif name == 'global_lr_scaling':
        return GlobalLRScaling(**kwargs)
    elif name == 'layer_freezing':
        return LayerFreezing(**kwargs)
    elif name == 'mono_ckpt_saver':
        return MonolithicCheckpointSaver(**kwargs)
    elif name == 'scheduled_gc':
        return ScheduledGarbageCollector(**kwargs)
    elif name == 'early_stopper':
        return EarlyStopper(**kwargs)
    elif name == 'hf_checkpointer':
        return HuggingFaceCheckpointer(**kwargs)
    else:
        raise ValueError(f'Not sure how to build callback: {name}')


def build_logger(name: str, kwargs: Dict[str, Any]) -> LoggerDestination:
    if name == 'wandb':
        return WandBLogger(**kwargs)
    elif name == 'tensorboard':
        return TensorboardLogger(**kwargs)
    elif name == 'in_memory_logger':
        return InMemoryLogger(**kwargs)
    elif name == 'mlflow':
        return MLFlowLogger(**kwargs)
    elif name == 'inmemory':
        return InMemoryLogger(**kwargs)
    else:
        raise ValueError(f'Not sure how to build logger: {name}')


def build_algorithm(name: str, kwargs: Dict[str, Any]) -> Algorithm:
    if name == 'gradient_clipping':
        return algorithms.GradientClipping(**kwargs)
    elif name == 'alibi':
        return algorithms.Alibi(**kwargs)
    elif name == 'fused_layernorm':
        return algorithms.FusedLayerNorm(**kwargs)
    elif name == 'gated_linear_units':
        return algorithms.GatedLinearUnits(**kwargs)
    elif name == 'low_precision_layernorm':
        return algorithms.LowPrecisionLayerNorm(**kwargs)
    else:
        raise ValueError(f'Not sure how to build algorithm: {name}')


def build_optimizer(model: torch.nn.Module, name: str,
                    optimizer_config: Dict[str, Any]) -> Optimizer:
    if name == 'decoupled_adamw':
        return DecoupledAdamW(model.parameters(), **optimizer_config)
    elif name == 'decoupled_lionw':
        return DecoupledLionW(model.parameters(), **optimizer_config)
    elif name == 'clip_lion':
        return DecoupledClipLion(model.parameters(), **optimizer_config)
    elif name == 'adalr_lion':
        return DecoupledAdaLRLion(model.parameters(), **optimizer_config)
    elif name == 'decoupled_lionw_8b':
        return DecoupledLionW_8bit(model.parameters(), **optimizer_config)
    else:
        raise ValueError(f'Not sure how to build optimizer: {name}')


def build_scheduler(name: str,
                    scheduler_config: Dict[str, Any]) -> ComposerScheduler:
    if name == 'constant_with_warmup':
        return ConstantWithWarmupScheduler(**scheduler_config)
    elif name == 'cosine_with_warmup':
        return CosineAnnealingWithWarmupScheduler(**scheduler_config)
    elif name == 'inv_sqrt_with_warmup':
        return InverseSquareRootWithWarmupScheduler(**scheduler_config)
    elif name == 'linear_decay_with_warmup':
        return LinearWithWarmupScheduler(**scheduler_config)
    else:
        raise ValueError(f'Not sure how to build scheduler: {name}')


def build_tokenizer(
        tokenizer_name: str,
        tokenizer_kwargs: Dict[str, Any]) -> PreTrainedTokenizerBase:
    os.environ['TRANSFORMERS_NO_ADVISORY_WARNINGS'] = '1'
    os.environ['TOKENIZERS_PARALLELISM'] = 'false'

    if tokenizer_name.startswith('tiktoken'):
        tokenizer = TiktokenTokenizerWrapper(**tokenizer_kwargs)
    else:
        tokenizer = AutoTokenizer.from_pretrained(tokenizer_name,
                                                  **tokenizer_kwargs)

        # HuggingFace does not respect the model_max_length kwarg, and overrides it with
        # min(kwargs['model_max_length'], original_config['model_max_length']), so we
        # explicitly set it here
        tokenizer.model_max_length = tokenizer_kwargs.get(
            'model_max_length',
            int(1e30),
        )

    return tokenizer


def build_icl_evaluators(
    icl_tasks: Union[str, ListConfig],
    tokenizer: PreTrainedTokenizerBase,
    default_max_seq_len: int,
    default_batch_size: int,
    destination_dir: Optional[str] = None,
    icl_subset_num_batches: Optional[int] = None,
) -> Tuple[List[Evaluator], List[str]]:
    if destination_dir is None:
        destination_dir = os.getcwd()

    evaluators = []
    logger_keys = []

    icl_tasks_list = None
    if isinstance(icl_tasks, str):
        log.info(f'Extracting ICL task config from path: {icl_tasks}')
        with open(icl_tasks, 'r') as icl_f:
            icl_task_cfg = om.load(icl_f)
        icl_tasks_list = icl_task_cfg.icl_tasks
    else:
        icl_tasks_list = icl_tasks

    def _validate_cfg(icl_cfg: DictConfig):
        assert 'label' in icl_cfg
        assert 'dataset_uri' in icl_cfg and icl_cfg.dataset_uri is not None
        assert 'icl_task_type' in icl_cfg
        assert 'num_fewshot' in icl_cfg

        if 'metric_names' not in icl_cfg:
            if icl_cfg.icl_task_type == 'language_modeling':
                icl_cfg.metric_names = ['InContextLearningLMAccuracy']
            elif icl_cfg.icl_task_type == 'multiple_choice':
                icl_cfg.metric_names = [
                    'InContextLearningMultipleChoiceAccuracy'
                ]
            elif icl_cfg.icl_task_type == 'schema':
                icl_cfg.metric_names = [
                    'InContextLearningMultipleChoiceAccuracy'
                ]
            elif icl_cfg.icl_task_type == 'question_answering':
                icl_cfg.metric_names = ['InContextLearningQAAccuracy']
            elif icl_cfg.icl_task_type == 'code_evaluation':
                icl_cfg.metric_names = ['InContextLearningCodeEvalAccuracy']
            else:
                raise ValueError(
                    f'No metric_names defined, unable to build default metrics for icl_task_type={icl_cfg.icl_task_type}.'
                )

        if 'prompt_string' not in icl_cfg:
            icl_cfg.prompt_string = ''
        if 'example_delimiter' not in icl_cfg:
            icl_cfg.example_delimiter = '\n'
        if 'continuation_delimiter' not in icl_cfg:
            icl_cfg.continuation_delimiter = ' '
        if 'max_seq_len' not in icl_cfg:
            icl_cfg.max_seq_len = default_max_seq_len
        if 'batch_size' not in icl_cfg:
            icl_cfg.batch_size = default_batch_size
        if 'pass_at_k' not in icl_cfg:
            icl_cfg.pass_at_k = 1
        if 'num_beams' not in icl_cfg:
            icl_cfg.num_beams = 20

    for icl_cfg in icl_tasks_list:
        assert isinstance(icl_cfg, DictConfig)
        _validate_cfg(icl_cfg)
        for num_fewshot in list(icl_cfg.num_fewshot):
            if tokenizer.pad_token_id is None:
                # Current workaround to support GPT2 tokenizer with `pad_token_id = None`
                pad_tok_id = tokenizer.eos_token_id
            else:
                pad_tok_id = tokenizer.pad_token_id
            label = f'{icl_cfg.label}/{num_fewshot}-shot'
            metric_names = list(icl_cfg.metric_names)
            # TODO: fix Composer bug when copying local paths and destination exists
            destination_path = f'{destination_dir}/{icl_cfg.label}-{num_fewshot}.jsonl'
            if dist.get_local_rank() == 0 and os.path.exists(destination_path):
                os.remove(destination_path)
            dist.barrier()

            dataloaders = get_icl_task_dataloader(
                icl_cfg.icl_task_type,
                icl_cfg.dataset_uri,
                tokenizer,
                batch_size=icl_cfg.batch_size,
                max_seq_len=icl_cfg.max_seq_len,
                pad_tok_id=pad_tok_id,
                num_fewshot=num_fewshot,
                prompt_string=icl_cfg.prompt_string,
                example_delimiter=icl_cfg.example_delimiter,
                continuation_delimiter=icl_cfg.continuation_delimiter,
                destination_path=destination_path,
                pass_at_k=icl_cfg.pass_at_k,
                generations_per_sample=icl_cfg.num_beams,
                has_categories=icl_cfg.get('has_categories', False),
            )
            if hasattr(
                    icl_cfg,
                    'has_categories') and icl_cfg.has_categories and isinstance(
                        dataloaders, dict):
                for category in dataloaders.keys():
                    logger_keys.extend([
                        f'metrics/{label}/{category}/{m}' for m in metric_names
                    ])
                    evaluators.append(
                        Evaluator(label=f'{label}/{category}',
                                  dataloader=dataloaders[category],
                                  metric_names=metric_names),)
            else:
                logger_keys.extend(
                    [f'metrics/{label}/{m}' for m in metric_names])
                evaluators.append(
                    Evaluator(label=label,
                              dataloader=dataloaders,
                              metric_names=metric_names,
                              subset_num_batches=icl_subset_num_batches))

    return evaluators, logger_keys