Coverage for /opt/cloud/slavespace/usr1/096471637100f3de0fcfc01072822a80/dttest/api/python/llm_datadist_v1/llm

1#!/usr/bin/env python3

2# -*- coding: utf-8 -*-

3# -----------------------------------------------------------------------------------------------------------

5# This program is free software, you can redistribute it and/or modify it under the terms and conditions of

6# CANN Open Software License Agreement Version 2.0 (the "License").

7# Please refer to the License for details. You may not use this file except in compliance with the License.

8# THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,

9# INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.

10# See LICENSE in the root of the software repository for the full text of the License.

11# -----------------------------------------------------------------------------------------------------------

13__all__ = [

14 "CacheDesc",

15 "CacheKey",

16 "CacheKeyByIdAndIndex",

17 "KvCache",

18 "BlocksCacheKey",

19 "Placement",

20 "CacheTask",

21 "TransferConfig",

22 "LayerSynchronizer",

23]

25from abc import ABC, abstractmethod

26from enum import Enum, IntEnum

27from typing import List, Optional, Tuple, Union

29from llm_datadist_v1 import llm_wrapper

31from .data_type import DataType

32from .status import LLMException, LLMStatusCode, raise_if_false, raise_if_true

33from .utils import log

34from .utils.utils import (

35 check_int32,

36 check_int64,

37 check_isinstance,

38 check_list_int64,

39 check_list_uint64,

40 check_type,

41 check_uint32,

42 check_uint64,

43)

45_INVALID_ID = 2**64 - 1

48class PushType(Enum):

49 NO_CACHE_KEY = 0

50 BLOCKS_CACHE_KEY = 1

51 CACHE_KEY_BY_ID = 2

54class RegisterMemStatus(Enum):

55 OK = 0

56 PREPARING = 1

57 FAILED = 2

60class Memtype(Enum):

61 MEM_TYPE_DEVICE = 0

62 MEM_TYPE_HOST = 1

65class MemInfo(object):

66 def __init__(self, mem_type: Memtype, addr: int, size: int):

67 """

68 初始化

69 Args:

70 mem_type: 内存类型

71 addr: 数据地址

72 size: 数据大小

73 """

74 check_isinstance("mem_type", mem_type, Memtype)

75 check_isinstance("addr", addr, int)

76 check_isinstance("size", size, int)

77 self._mem_type = mem_type

78 self._addr = addr

79 self._size = size

81 @property

82 def mem_type(self):

83 return self._mem_type

85 @property

86 def addr(self):

87 return self._addr

89 @property

90 def size(self):

91 return self._size

93 def __str__(self):

94 return f"MemInfo(mem_type={str(self.mem_type)}, addr={str(self.addr)}, size={str(self.size)})"

96 def __repr__(self):

97 return self.__str__()

100int_to_mem_status_dict = {

101 0: RegisterMemStatus.OK,

102 1: RegisterMemStatus.PREPARING,

103 2: RegisterMemStatus.FAILED,

104}

105

106

107class Placement(IntEnum):

108 HOST = 0

109 DEVICE = 1

110

111

112class CacheDesc(object):

113 """

114 Cache描述

115

116 Args:

117 num_tensors: Cache中tensor的个数, 操作Cache时, 所有tensor会做同样的操作

118 shape: tensor shape

119 data_type: data type

120 placement: 标识kv cache所在的设备

121 batch_dim_index (Optional): batch dim的index, 默认为0

122 seq_len_dim_index (Optional): seq_len dim的index, 高阶API pull_kv时需要该字段判断是否能够按实际大小拉取

123 kv_tensor_format (Optional): kv tensor的data format

124

125 Examples:

126 >>> from llm_datadist import CacheDesc

127 >>> tensor_num = 80

128 >>> tensor_shape = [4, 256]

129 >>> tensor_data_type = DataType.DT_FLOAT16

130 >>> cache_desc_1 = CacheDesc(tensor_num, tensor_shape, tensor_data_type)

131 >>> # 指定 batch_dim_index = 0, seq_len_dim_index = 1

132 >>> cache_desc_2 = CacheDesc(tensor_num, tensor_shape, tensor_data_type, Placement.DEVICE, 0, 1)

133 """

134

135 def __init__(

136 self,

137 num_tensors: int,

138 shape: Union[Tuple[int], List[int]],

139 data_type: DataType,

140 placement: Placement = Placement.DEVICE,

141 batch_dim_index: int = 0,

142 seq_len_dim_index: int = -1,

143 kv_tensor_format: str = None,

144 ):

145 self._num_tensors = num_tensors

146 self._shape = shape

147 check_uint32("num_tensors", num_tensors)

148 check_isinstance("shape", shape, [tuple, list], int)

149 check_isinstance("data_type", data_type, DataType)

150 check_isinstance("placement", placement, Placement)

151 check_isinstance("batch_dim_index", batch_dim_index, int)

152 check_int32("seq_len_dim_index", seq_len_dim_index)

153 check_isinstance("kv_tensor_format", kv_tensor_format, str)

154 raise_if_false(

155 0 <= batch_dim_index < len(shape),

156 "batch_dim_index {0} out of range, [0, {1})",

157 batch_dim_index,

158 len(shape),

159 )

160 raise_if_false(

161 seq_len_dim_index == -1 or 0 <= seq_len_dim_index < len(shape),

162 "seq_len_dim_index {0} is invalid, should be -1 or in range [0, {1})",

163 seq_len_dim_index,

164 len(shape),

165 )

166 check_list_int64("shape", shape)

167 self._data_type = data_type

168 self._batch_dim = batch_dim_index

169 self._batch_size = shape[batch_dim_index]

170 self._seq_len_dim_index = seq_len_dim_index

171 self._size = -1

172 self._kv_tensor_format = kv_tensor_format

173 self._placement = placement

174 self._is_blocks = False

175

176 def __repr__(self):

177 return (

178 f"CacheDesc(num_tensors={self.num_tensors}, "

179 f"shape={self.shape}, "

180 f"data_type={self.data_type}, "

181 f"placement={self.placement}, "

182 f"batch_dim_index={self.batch_dim}, "

183 f"seq_len_dim_index={self._seq_len_dim_index},"

184 f"kv_tensor_format={self.kv_tensor_format})"

185 )

186

187 @property

188 def num_tensors(self) -> int:

189 return self._num_tensors

190

191 @property

192 def shape(self) -> List[int]:

193 return self._shape

194

195 def update_dim(self, dim_index: int, dim_value: int) -> None:

196 self._shape[dim_index] = dim_value

197 self._size = -1

198

199 @property

200 def data_type(self) -> DataType:

201 return self._data_type

202

203 @property

204 def batch_dim(self) -> int:

205 return self._batch_dim

206

207 @property

208 def batch_size(self) -> int:

209 return self._batch_size

210

211 @property

212 def seq_len_dim_index(self) -> int:

213 return self._seq_len_dim_index

214

215 @property

216 def size(self) -> int:

217 if self._size == -1:

218 self._size = llm_wrapper.calc_tensor_size(self.shape, self._data_type.value)

219 if self._size < 0:

220 raise LLMException(f"Failed to calc tensor size, shape = {self.shape}, data_type = {self.data_type}")

221 return self._size

222

223 @property

224 def kv_tensor_format(self) -> str:

225 return self._kv_tensor_format

226

227 @property

228 def placement(self) -> Placement:

229 return self._placement

230

231

232class BlocksCacheKey(object):

233 """

234 BlocksCacheKey, PROMPT allocate blocks cache时用于建立索引, DECODER pull_kv时作为索引传入

235

236 Args:

237 prompt_cluster_id or cluster_id: remote cluster id, 用于pull_kv时需要准确填写, allocate_cache时会忽略该字段

238 model_id (Optional): model id, 默认为0, 在投机等同时加载多个model的场景需要按需设置

239

240 Raises:

241 TypeError: if `prompt_cluster_id` or `cluster_id` is not int

242 TypeError: if `model_id` is not int

243

244 Example:

245 >>> from llm_datadist import BlocksCacheKey

246 >>> cluster_id = 1

247 >>> # model_id取默认值

248 >>> cache_key = BlocksCacheKey(cluster_id)

249 >>> # 指定model_id

250 >>> model_id = 1

251 >>> cache_key_1 = BlocksCacheKey(cluster_id, model_id)

252 """

253

254 def __init__(self, *args, **kwargs):

255 raise_if_false((len(args) + len(kwargs)) <= 2, "Param num is over limit")

256 if len(args) > 0:

257 kwargs["cluster_id"] = args[0]

258 if len(args) > 1:

259 kwargs["model_id"] = args[1]

260 raise_if_false(

261 "prompt_cluster_id" in kwargs or "cluster_id" in kwargs,

262 "Param prompt_cluster_id or cluster_id is required",

263 )

264 valid_keys = ["prompt_cluster_id", "cluster_id", "model_id"]

265 for k in kwargs.keys():

266 raise_if_false(k in valid_keys, "Unsupported param:{}", k)

267 self._cluster_id = kwargs["cluster_id"] if "cluster_id" in kwargs else kwargs["prompt_cluster_id"]

268 self._cluster_id_key = "cluster_id" if "cluster_id" in kwargs else "prompt_cluster_id"

269 self._model_id = kwargs["model_id"] if "model_id" in kwargs else 0

270 check_isinstance("cluster_id", self._cluster_id, int, extra_fmt="The first param ")

271 check_isinstance("model_id", self._model_id, int, extra_fmt="The second param ")

272

273 @property

274 def prompt_cluster_id(self) -> int:

275 return self._cluster_id

276

277 @property

278 def cluster_id(self) -> int:

279 return self._cluster_id

280

281 @property

282 def model_id(self) -> int:

283 return self._model_id

284

285 def __repr__(self):

286 return f"BlocksCacheKey({self._cluster_id_key}={self.cluster_id}, model_id={self.model_id})"

287

288

289class CacheKey(object):

290 """

291 CacheKey, REMOTE allocate_cache时用于建立索引, LOCAL pull_cache时作为索引传入

292

293 Args:

294 prompt_cluster_id or cluster_id: remote cluster id, 用于pull_kv时需要准确填写, allocate_cache时会忽略该字段

295 req_id: request id

296 model_id (Optional): model id, 默认为0, 在投机等同时加载多个model的场景需要按需设置

297 prefix_id (Optional): prefix id, 默认为2 ** 64 - 1

298

299 Raises:

300 TypeError: if `prompt_cluster_id` or `cluster_id` is not int

301 TypeError: if `req_id` is not int

302 TypeError: if `model_id` is not int

303 TypeError: if `prefix_id` is not int

304

305 Example:

306 >>> from llm_datadist import CacheKey

307 >>> cluster_id = 1

308 >>> request_id = 1

309 >>> # model_id取默认值

310 >>> cache_key = CacheKey(cluster_id, request_id)

311 >>> # 指定model_id

312 >>> cache_key_1 = CacheKey(cluster_id, request_id, 2)

313 """

314

315 def __init__(self, *args, **kwargs):

316 raise_if_false((len(args) + len(kwargs)) <= 4, "Param num is over limit")

317 if len(args) > 0:

318 kwargs["cluster_id"] = args[0]

319 if len(args) > 1:

320 kwargs["req_id"] = args[1]

321 if len(args) > 2:

322 kwargs["model_id"] = args[2]

323 if len(args) > 3:

324 kwargs["prefix_id"] = args[3]

325 raise_if_false(

326 "prompt_cluster_id" in kwargs or "cluster_id" in kwargs,

327 "Param prompt_cluster_id or cluster_id is required",

328 )

329 raise_if_false("req_id" in kwargs or "req_id" in kwargs, "Param req_id is required")

330 valid_keys = [

331 "prompt_cluster_id",

332 "cluster_id",

333 "req_id",

334 "model_id",

335 "prefix_id",

336 ]

337 for k in kwargs.keys():

338 raise_if_false(k in valid_keys, "Unsupported param:{}", k)

339 self._cluster_id = kwargs["cluster_id"] if "cluster_id" in kwargs else kwargs["prompt_cluster_id"]

340 self._req_id = kwargs["req_id"]

341 self._model_id = kwargs["model_id"] if "model_id" in kwargs else 0

342 self._prefix_id = kwargs["prefix_id"] if "prefix_id" in kwargs else _INVALID_ID

343 check_uint64("cluster_id", self._cluster_id)

344 check_uint64("req_id", self._req_id)

345 check_uint64("model_id", self._model_id)

346 check_uint64("prefix_id", self._prefix_id)

347

348 @property

349 def prompt_cluster_id(self) -> int:

350 return self._cluster_id

351

352 @property

353 def cluster_id(self) -> int:

354 return self._cluster_id

355

356 @property

357 def req_id(self) -> int:

358 return self._req_id

359

360 @property

361 def prefix_id(self) -> int:

362 return self._prefix_id

363

364 @property

365 def model_id(self) -> int:

366 return self._model_id

367

368 def __repr__(self):

369 return (

370 f"CacheKey(cluster_id={self.cluster_id}, "

371 f"req_id={self.req_id}, "

372 f"prefix_id={self.prefix_id}, "

373 f"model_id={self.model_id})"

374 )

375

376

377class CacheKeyByIdAndIndex(object):

378 """

379 索引Prompt中kv cache的单个batch index, 用于Decoder pull_cache时作为索引传入

380

381 Args:

382 cluster_id: kv所在的prompt的cluster_id

383 cache_id: kv cache的id

384 batch_index (Optional): batch index, 默认为0, 在投机等同时加载多个model的场景需要按需设置

385

386 Raises:

387 TypeError: if `cluster_id` is not int

388 TypeError: if `cache_id` is not int

389 TypeError: if `batch_index` is not int

390 """

391

392 def __init__(self, cluster_id: int, cache_id: int, batch_index=0):

393 check_uint64("cluster_id", cluster_id)

394 check_int64("cache_id", cache_id)

395 check_uint32("batch_index", batch_index)

396 self._prompt_cluster_id = cluster_id

397 self._prompt_cache_id = cache_id

398 self._prompt_batch_index = batch_index

399 self._req_id = _INVALID_ID

400

401 @property

402 def cluster_id(self) -> int:

403 return self._prompt_cluster_id

404

405 @property

406 def cache_id(self) -> int:

407 return self._prompt_cache_id

408

409 @property

410 def batch_index(self) -> int:

411 return self._prompt_batch_index

412

413 def __repr__(self):

414 return (

415 f"CacheKeyByIdAndIndex(cluster_id={self.cluster_id}, "

416 f"cache_id={self.cache_id}, "

417 f"batch_index={self.batch_index})"

418 )

419

420

421class KvCache(object):

422 """

423 Kv Cache, 由KvCacheManager.allocate_cache创建，用户不应直接构造

424 """

425

426 def __init__(

427 self,

428 cache_id: int,

429 cache_desc: CacheDesc,

430 per_device_tensor_addrs: List[List[int]],

431 kv_cache_manager,

432 ):

433 self._cache_id = cache_id

434 check_int64("cache_id", cache_id)

435 self._cache_desc = cache_desc

436 self._per_device_tensor_addrs = per_device_tensor_addrs

437 self._valid = True

438 self._kv_cache_manager = kv_cache_manager

439

440 def __del__(self):

441 # 保底释放kv cache, 但更推荐主动通过调用kv_cache_manager.deallocate_cache来释放kv cache，而不应该遗留到此处自动释放

442 if self._valid and self._kv_cache_manager is not None and self._kv_cache_manager.is_initialized():

443 log.info("auto deallocate kv cache: %s", self.cache_id)

444 self._kv_cache_manager.deallocate_cache(self)

445

446 @property

447 def cache_id(self) -> int:

448 return self._cache_id

449

450 @property

451 def cache_desc(self) -> CacheDesc:

452 return self._cache_desc

453

454 @property

455 def per_device_tensor_addrs(self) -> List[List[int]]:

456 return self._per_device_tensor_addrs

457

458 def __str__(self):

459 return f"KvCache(cache_id = {self._cache_id}, num_devices = {len(self._per_device_tensor_addrs)})"

460

461 def __repr__(self):

462 return self.__str__()

463

464 @classmethod

465 def create_cpu_cache(cls, cache_desc: CacheDesc, addrs: Union[List[int], List[List[int]]]):

466 check_isinstance("cache_desc", cache_desc, CacheDesc)

467 raise_if_false(cache_desc.placement == Placement.HOST, "cache_desc placement must be HOST")

468 check_isinstance("addrs", addrs, list)

469 raise_if_false(len(addrs) > 0, "addrs length should be bigger than zero.")

470 last_type = type(addrs[0])

471 for addr in addrs:

472 check_isinstance("the internal element of addrs", addr, [list, int])

473 if check_type(addr, list):

474 check_isinstance("the internal element of addrs", addr, list, int)

475 raise_if_false(

476 cache_desc.num_tensors == len(addr),

477 f"cache_desc num_tensors:{cache_desc.num_tensors} "

478 f"should be equal to size of the internal element of addrs:{len(addr)}",

479 )

480 raise_if_false(

481 check_type(addr, last_type),

482 "the type of the internal element of addrs should be consistent.",

483 )

484 last_type = type(addr)

485 if last_type == int:

486 raise_if_false(

487 cache_desc.num_tensors == len(addrs),

488 f"cache_desc num_tensors:{cache_desc.num_tensors} should be equal to size of addrs:{len(addrs)}",

489 )

490 return cls(-1, cache_desc, [addrs] if last_type == int else addrs, None)

491

492

493class LayerSynchronizer(ABC):

494 @abstractmethod

495 def synchronize_layer(self, layer_index: int, timeout_in_millis: Optional[int]) -> bool:

496 """

497 阻塞等待指定层计算完成

498

499 Args:

500 layer_index(int): 要同步的layer的index

501 timeout_in_millis(Optional[int]): 超时时间, 不配置则不超时

502

503 Returns:

504 True: 同步成功, False: 同步失败

505 """

506

507

508def check_layer_range(name, layer_range: Optional[range], allow_none=True):

509 if allow_none and layer_range is None:

510 return

511 check_isinstance(name, layer_range, range, allow_none=allow_none)

512 raise_if_false(layer_range.step == 1, f"check {name}.step == 1 failed")

513 raise_if_false(

514 0 <= layer_range.start < layer_range.stop,

515 "check 0 <= range.start < range.stop failed, {0} = {1}",

516 name,

517 layer_range,

518 )

519

520

521class TransferWithCacheKeyConfig:

522 def __init__(

523 self,

524 cache_key: Union[BlocksCacheKey, CacheKeyByIdAndIndex],

525 src_layer_range: range = None,

526 dst_layer_range: range = None,

527 src_batch_index: int = 0,

528 ):

529 check_isinstance(

530 "cache_key",

531 cache_key,

532 [BlocksCacheKey, CacheKeyByIdAndIndex],

533 allow_none=False,

534 )

535 self._cache_key = cache_key

536 check_layer_range("src_layer_range", src_layer_range, allow_none=False)

537 self._src_layer_range = src_layer_range

538 check_layer_range("dst_layer_range", dst_layer_range, allow_none=False)

539 self._dst_layer_range = dst_layer_range

540 raise_if_false(

541 (src_layer_range.stop - src_layer_range.start) == (dst_layer_range.stop - src_layer_range.start),

542 "src_layer_range size shoulde be equal to dst_layer_range size",

543 )

544 check_uint32("src_batch_index", src_batch_index)

545

546 raise_if_true(

547 check_type(cache_key, BlocksCacheKey) and src_batch_index != 0,

548 "src_batch_index shoulde be 0 when cache_key is BlocksCacheKey.",

549 )

550 self._src_batch_index = src_batch_index

551 self.dst_cluster_id = cache_key.cluster_id

552

553 def __repr__(self) -> str:

554 return (

555 f"TransferWithCacheKeyConfig(cache_key={self._cache_key},"

556 f" src_layer_range={self._src_layer_range},"

557 f" dst_layer_range={self._dst_layer_range},"

558 f" src_batch_index={self._src_batch_index})"

559 )

560

561 @property

562 def cache_key(self):

563 return self._cache_key

564

565 @cache_key.setter

566 def cache_key(self, cache_key: Union[BlocksCacheKey, CacheKeyByIdAndIndex]) -> None:

567 check_isinstance("cache_key", cache_key, [BlocksCacheKey, CacheKeyByIdAndIndex])

568 self._cache_key = cache_key

569

570 @property

571 def src_layer_range(self) -> Optional[range]:

572 return self._src_layer_range

573

574 @src_layer_range.setter

575 def src_layer_range(self, src_layer_range: Optional[range]) -> None:

576 check_layer_range("src_layer_range", src_layer_range)

577 self._src_layer_range = src_layer_range

578

579 @property

580 def dst_layer_range(self) -> Optional[range]:

581 return self._dst_layer_range

582

583 @dst_layer_range.setter

584 def dst_layer_range(self, dst_layer_range: Optional[range]) -> None:

585 check_layer_range("dst_layer_range", dst_layer_range)

586 self._dst_layer_range = dst_layer_range

587

588 @property

589 def src_batch_index(self) -> int:

590 return self._src_batch_index

591

592 @src_batch_index.setter

593 def src_batch_index(self, src_batch_index: int) -> None:

594 raise_if_true(

595 check_type(self.cache_key, BlocksCacheKey) and src_batch_index != 0,

596 "src_batch_index shoulde be 0 when cache_key is BlocksCacheKey.",

597 )

598 self._check_src_batch_index(src_batch_index)

599 self._src_batch_index = src_batch_index

600

601 @staticmethod

602 def _check_src_batch_index(src_batch_index: int):

603 check_uint32("src_batch_index", src_batch_index)

604

605

606class TransferConfig:

607 def __init__(

608 self,

609 dst_cluster_id: int,

610 dst_addrs: List[int],

611 src_layer_range: Optional[range] = None,

612 src_batch_index: int = 0,

613 ):

614 self._check_dst_cluster_id(dst_cluster_id)

615 self._check_dst_addrs(dst_addrs)

616 check_layer_range("src_layer_range", src_layer_range)

617 self._check_src_batch_index(src_batch_index)

618 self._dst_cluster_id = dst_cluster_id

619 self._dst_addrs = dst_addrs

620 self._src_layer_range = src_layer_range

621 self._src_batch_index = src_batch_index

622 self.dst_layer_range = None

623

624 def __repr__(self) -> str:

625 return (

626 f"TransferConfig(src_cluster_id={self._dst_cluster_id},"

627 f" dst_addrs={self._dst_addrs},"

628 f" src_layer_range={self._src_layer_range},"

629 f" src_batch_index={self._src_batch_index})"

630 )

631

632 def __str__(self):

633 return self.__repr__()

634

635 @property

636 def dst_cluster_id(self) -> int:

637 return self._dst_cluster_id

638

639 @property

640 def dst_addrs(self) -> List[int]:

641 return self._dst_addrs

642

643 @property

644 def src_layer_range(self) -> Optional[range]:

645 return self._src_layer_range

646

647 @property

648 def src_batch_index(self) -> int:

649 return self._src_batch_index

650

651 @dst_cluster_id.setter

652 def dst_cluster_id(self, cluster_id: int) -> None:

653 self._check_dst_cluster_id(cluster_id)

654 self._dst_cluster_id = cluster_id

655

656 @dst_addrs.setter

657 def dst_addrs(self, dst_addrs: List[int]) -> None:

658 self._check_dst_addrs(dst_addrs)

659 self._dst_addrs = dst_addrs

660

661 @src_layer_range.setter

662 def src_layer_range(self, layer_range: Optional[range]) -> None:

663 check_layer_range("src_layer_range", layer_range)

664 self._src_layer_range = layer_range

665

666 @src_batch_index.setter

667 def src_batch_index(self, batch_index: int) -> None:

668 self._check_src_batch_index(batch_index)

669 self._src_batch_index = batch_index

670

671 @staticmethod

672 def _check_dst_cluster_id(dst_cluster_id: int):

673 check_uint64("dst_cluster_id", dst_cluster_id)

674

675 @staticmethod

676 def _check_dst_addrs(dst_addrs: List[int]):

677 check_isinstance("dst_addrs", dst_addrs, [list, tuple], int, allow_none=False)

678 check_list_uint64("dst_addrs", dst_addrs)

679

680 @staticmethod

681 def _check_src_batch_index(src_batch_index: int):

682 check_uint32("src_batch_index", src_batch_index)

683

684

685class CacheTask:

686 def __init__(self, transfer_async_task):

687 self._transfer_async_task = transfer_async_task

688

689 def synchronize(self, timeout_in_millis: Optional[int] = None) -> LLMStatusCode:

690 timeout = None

691 if timeout_in_millis is not None:

692 check_uint32("timeout_in_millis", timeout_in_millis)

693 timeout = timeout_in_millis / 1000

694 return self._transfer_async_task.get(timeout)

695

696 def get_results(self, timeout_in_millis: Optional[int] = None) -> List[LLMStatusCode]:

697 timeout = None

698 if timeout_in_millis is not None:

699 check_uint32("timeout_in_millis", timeout_in_millis)

700 timeout = timeout_in_millis / 1000

701 return self._transfer_async_task.get_results(timeout)

Coverage for /opt/cloud/slavespace/usr1/096471637100f3de0fcfc01072822a80/dttest/api/python/llm_datadist_v1/llm_types.py: 97%

358 statements