NonTensorStack¶
- class tensordict.NonTensorStack(*args, **kwargs)¶
LazyStackedTensorDict 的一個薄封裝器,使對非張量資料進行堆疊更容易識別。
當在
stack()
清單上調用NonTensorData
或NonTensorStack
時,會傳回NonTensorStack
。範例
>>> from tensordict import NonTensorData >>> import torch >>> data = torch.stack([ ... torch.stack([NonTensorData(data=(i, j), batch_size=[]) for i in range(2)]) ... for j in range(3)]) >>> print(data) NonTensorStack( [[(0, 0), (1, 0)], [(0, 1), (1, 1)], [(0, 2), (1, ..., batch_size=torch.Size([3, 2]), device=None)
若要取得儲存在
NonTensorStack
中的值,請調用tolist
。- abs() T ¶
計算 TensorDict 中每個元素的絕對值。
- abs_() T ¶
就地計算 TensorDict 中每個元素的絕對值。
- acos() T ¶
計算 TensorDict 中每個元素的
acos()
值。
- acos_() T ¶
就地計算 TensorDict 中每個元素的
acos()
值。
- add(other: tensordict.base.TensorDictBase | torch.Tensor, *, alpha: Optional[float] = None, default: Optional[Union[str, Tensor]] = None) TensorDictBase ¶
將
other
乘以alpha
後,加到self
。\[\text{{out}}_i = \text{{input}}_i + \text{{alpha}} \times \text{{other}}_i\]- 參數:
other (TensorDictBase 或 torch.Tensor) – 要加到
self
的 tensor 或 TensorDict。- 關鍵字參數:
alpha (Number, optional) –
other
的乘數。default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- add_(other: tensordict.base.TensorDictBase | float, *, alpha: Optional[float] = None)¶
是
add()
的原地版本。注意
原地
add
不支援default
關鍵字參數。
- addcdiv(other1: tensordict.base.TensorDictBase | torch.Tensor, other2: tensordict.base.TensorDictBase | torch.Tensor, value: float | None = 1)¶
執行
other1
除以other2
的逐元素除法,將結果乘以純量value
,並將其加到self
。\[\text{out}_i = \text{input}_i + \text{value} \times \frac{\text{tensor1}_i}{\text{tensor2}_i}\]self
、other1
和other2
元素的形狀必須是可廣播的。對於 FloatTensor 或 DoubleTensor 類型的輸入,
value
必須是實數,否則必須是整數。- 參數:
other1 (TensorDict 或 Tensor) – 分子 tensordict(或 tensor)
tensor2 (TensorDict 或 Tensor) – 分母 tensordict(或 tensor)
- 關鍵字參數:
value (數值, 選填) – \(\text{tensor1} / \text{tensor2}\) 的乘數
- addcmul(other1, other2, *, value: float | None = 1)¶
執行
other1
與other2
的逐元素乘法,將結果乘以純量value
並將其加到self
。\[\text{out}_i = \text{input}_i + \text{value} \times \text{other1}_i \times \text{other2}_i\]self
、other1
和other2
的形狀必須是可廣播的 (broadcastable)。對於 FloatTensor 或 DoubleTensor 類型的輸入,
value
必須是實數,否則必須是整數。- 參數:
other1 (TensorDict 或 Tensor) – 要相乘的 tensordict 或 tensor
other2 (TensorDict 或 Tensor) – 要相乘的 tensordict 或 tensor
- 關鍵字參數:
value (數值, 選填) – \(other1 .* other2\) 的乘數
- all(dim: Optional[int] = None) bool | tensordict.base.TensorDictBase ¶
檢查 tensordict 中是否所有值都是 True/非空值。
- 參數:
dim (int, 選填) – 若為
None
,則返回一個布林值,指示是否所有 tensors 都返回 tensor.all() == True。如果是整數,則僅當此維度與 tensordict 形狀相容時,才對指定的維度呼叫 all。
- any(dim: Optional[int] = None) bool | tensordict.base.TensorDictBase ¶
檢查 tensordict 中是否有任何值為 True/非空值。
- 參數:
dim (int, 選填) – 若為
None
,則返回一個布林值,指示是否所有 tensors 都返回 tensor.any() == True。如果是整數,則僅當此維度與 tensordict 形狀相容時,才對指定的維度呼叫 all。
- append(tensordict: T) None ¶
將 TensorDict 附加到堆疊上。
類似於 list.append。附加的 TensorDict 必須具有相容的 batch_size 和 device。 append 操作是原地 (in-place) 執行,不返回任何值。
- 參數:
tensordict (TensorDictBase) – 要附加到堆疊上的 TensorDict。
- apply(fn: Callable, *others: T, batch_size: Optional[Sequence[int]] = None, device: torch.device | None = _NoDefault.ZERO, names: Optional[Sequence[str]] = _NoDefault.ZERO, inplace: bool = False, default: Any = _NoDefault.ZERO, filter_empty: Optional[bool] = None, propagate_lock: bool = False, call_on_nested: bool = False, out: Optional[TensorDictBase] = None, **constructor_kwargs) Optional[T] ¶
將一個可呼叫物件應用到 tensordict 中儲存的所有值,並將它們設定在一個新的 tensordict 中。
可呼叫物件的簽章必須是
Callable[Tuple[Tensor, ...], Optional[Union[Tensor, TensorDictBase]]]
。- 參數:
fn (Callable) – 要應用到 tensordict 中 tensor 的函數。
*others (TensorDictBase 實例, 選擇性) – 如果提供,這些 tensordict 實例應具有與 self 匹配的結構。
fn
參數應接收與 tensordict 數量一樣多的未命名輸入,包括 self。 如果其他 tensordict 缺少條目,則可以透過default
關鍵字參數傳遞預設值。
- 關鍵字參數:
batch_size (int 序列, 選擇性) – 如果提供,產生的 TensorDict 將具有所需的 batch_size。
batch_size
參數應與轉換後的 batch_size 相符。 這是一個僅限關鍵字的參數。device (torch.device, 選擇性) – 產生的裝置,如果有的話。
names (字串列表, 選擇性) – 新的維度名稱,以防 batch_size 被修改。
inplace (bool, 選擇性) – 如果為 True,則會進行原地修改。預設值為 False。 這是一個僅限關鍵字的參數。
default (Any, 選擇性) – 其他 tensordict 中遺失條目的預設值。 如果未提供,遺失的條目將引發 KeyError。
filter_empty (bool, 選擇性) – 如果為
True
,將過濾掉空的 tensordict。 這也具有較低的計算成本,因為不會建立和銷毀空的資料結構。 非 tensor 資料被視為葉節點,因此即使該函數未觸及,也會保留在 tensordict 中。 為了向後相容性,預設值為False
。propagate_lock (bool, optional) – 若為
True
,則鎖定的 tensordict 會產生另一個鎖定的 tensordict。預設為False
。call_on_nested (bool, optional) –
若為
True
,則該函式會被調用於第一層的 tensors 和 containers (TensorDict 或 tensorclass)。在這種情況下,func
負責將其呼叫傳播到巢狀層級。這允許在將呼叫傳播到巢狀 tensordicts 時進行細粒度的行為控制。若為False
,則該函式只會被調用於最底層的節點 (leaves),並且apply
會負責將該函式分派到所有節點。>>> td = TensorDict({"a": {"b": [0.0, 1.0]}, "c": [1.0, 2.0]}) >>> def mean_tensor_only(val): ... if is_tensor_collection(val): ... raise RuntimeError("Unexpected!") ... return val.mean() >>> td_mean = td.apply(mean_tensor_only) >>> def mean_any(val): ... if is_tensor_collection(val): ... # Recurse ... return val.apply(mean_any, call_on_nested=True) ... return val.mean() >>> td_mean = td.apply(mean_any, call_on_nested=True)
out (TensorDictBase, optional) –
用於寫入結果的 tensordict。這可以用於避免創建新的 tensordict。
>>> td = TensorDict({"a": 0}) >>> td.apply(lambda x: x+1, out=td) >>> assert (td==1).all()
警告
如果在 tensordict 上執行的操作需要訪問多個鍵才能進行單一計算,則提供一個等於
self
的out
參數可能會導致操作悄悄地提供錯誤的結果。例如>>> td = TensorDict({"a": 1, "b": 1}) >>> td.apply(lambda x: x+td["a"])["b"] # Right! tensor(2) >>> td.apply(lambda x: x+td["a"], out=td)["b"] # Wrong! tensor(3)
**constructor_kwargs – 要傳遞給 TensorDict 建構子的額外關鍵字參數。
- 返回值:
一個具有 transformed_in tensors 的新 tensordict。
範例
>>> td = TensorDict({ ... "a": -torch.ones(3), ... "b": {"c": torch.ones(3)}}, ... batch_size=[3]) >>> td_1 = td.apply(lambda x: x+1) >>> assert (td_1["a"] == 0).all() >>> assert (td_1["b", "c"] == 2).all() >>> td_2 = td.apply(lambda x, y: x+y, td) >>> assert (td_2["a"] == -2).all() >>> assert (td_2["b", "c"] == 2).all()
注意
如果函式返回
None
,則該條目將被忽略。這可以用於過濾 tensordict 中的資料。>>> td = TensorDict({"1": 1, "2": 2, "b": {"2": 2, "1": 1}}, []) >>> def filter(tensor): ... if tensor == 1: ... return tensor >>> td.apply(filter) TensorDict( fields={ 1: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: TensorDict( fields={ 1: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
注意
apply 方法將返回一個
TensorDict
實例,無論輸入類型如何。為了保持相同的類型,可以執行>>> out = td.clone(False).update(td.apply(...))
- apply_(fn: Callable, *others, **kwargs)¶
將一個可呼叫對象應用於 tensordict 中儲存的所有值,並將它們重新寫入到位。
- 參數:
fn (Callable) – 要應用到 tensordict 中 tensor 的函數。
*others (sequence of TensorDictBase, optional) – 要使用的其他 tensordicts (序列 of TensorDictBase,可選)。
Keyword Args: 請參閱
apply()
。- 返回值:
self 或應用該函式的 self 的副本
- asin() T ¶
計算 TensorDict 中每個元素的
asin()
值。
- asin_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
asin()
值。
- atan() T ¶
計算 TensorDict 中每個元素的
atan()
值。
- atan_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
atan()
值。
- auto_batch_size_(batch_dims: Optional[int] = None) T ¶
設定 tensordict 的最大批次大小,最多可設定 batch_dims。
- 參數:
batch_dims (int, optional) – 如果提供,批次大小最多為
batch_dims
長度。- 返回值:
self
範例
>>> from tensordict import TensorDict >>> import torch >>> td = TensorDict({"a": torch.randn(3, 4, 5), "b": {"c": torch.randn(3, 4, 6)}}, batch_size=[]) >>> td.auto_batch_size_() >>> print(td.batch_size) torch.Size([3, 4]) >>> td.auto_batch_size_(batch_dims=1) >>> print(td.batch_size) torch.Size([3])
- auto_device_() T ¶
如果裝置是唯一的,則自動設定裝置。
返回值:具有已編輯的
device
屬性的 self。
- property batch_size: Size¶
TensorDict 的形狀 (shape) 或批次大小 (batch_size)。
Tensordict 的形狀對應於其包含的張量的前
N
個共同維度,其中N
是一個任意數字。批次大小與代表張量語義相關形狀的「特徵大小」形成對比。例如,一批影片的形狀可能是[B, T, C, W, H]
,其中[B, T]
是批次大小(批次和時間維度),而[C, W, H]
是特徵維度(通道和空間維度)。TensorDict
的形狀由使用者在初始化時控制(也就是說,它不是從張量形狀推斷出來的)。如果新的大小與 TensorDict 的內容相容,則可以動態編輯
batch_size
。例如,始終允許將批次大小設定為空值。- 返回值:
描述 TensorDict 批次大小的
Size
物件。
範例
>>> data = TensorDict({ ... "key 0": torch.randn(3, 4), ... "key 1": torch.randn(3, 5), ... "nested": TensorDict({"key 0": torch.randn(3, 4)}, batch_size=[3, 4])}, ... batch_size=[3]) >>> data.batch_size = () # resets the batch-size to an empty value
- bfloat16()¶
將所有張量轉換為
torch.bfloat16
。
- bool()¶
將所有張量轉換為
torch.bool
。
- bytes(*, count_duplicates: bool = True) int ¶
計算包含張量的位元組數。
- 關鍵字參數:
count_duplicates (bool) – 是否將重複的張量視為獨立的張量來計數。如果
False
,則只會捨棄嚴格相同的張量(來自共同基礎張量的相同視圖但具有不同的 ID 將被計算兩次)。預設為 True(每個張量都假定為單一副本)。
- classmethod cat(input, dim=0, *, out=None)¶
將 tensordict 沿給定維度串聯成單個 tensordict。
此呼叫等效於呼叫
torch.cat()
,但與 torch.compile 相容。
- cat_from_tensordict(dim: int = 0, *, sorted: Optional[Union[bool, List[NestedKey]]] = None, out: Tensor = None) Tensor ¶
將 tensordict 的所有條目串聯成單個張量。
- 參數:
dim (int, optional) – 條目應沿其串聯的維度。
- 關鍵字參數:
sorted (bool 或 list of NestedKeys) – 如果
True
,則條目將按字母順序串聯。如果False
(預設),將使用 dict 順序。或者,可以提供金鑰名稱的清單,並且張量將相應地串聯。這會產生一些額外負擔,因為金鑰清單將針對 tensordict 中葉名稱的清單進行檢查。out (torch.Tensor, optional) – cat 操作的可選目標張量。
- cat_tensors(*keys: NestedKey, out_key: NestedKey, dim: int = 0, keep_entries: bool = False) T ¶
將多個條目串聯成一個新的條目,並可能移除原始值。
- 參數:
keys (NestedKey 序列) – 要串聯的條目。
- 關鍵字參數
out_key (NestedKey): 串聯後輸入的新鍵名。 keep_entries (bool, optional): 如果
False
,則keys
中的條目將被刪除。預設為
False
。- dim (int, optional): 串聯必須沿哪個維度發生。
預設為
0
。
回傳: self
範例
>>> td = TensorDict(a=torch.zeros(1), b=torch.ones(1)) >>> td.cat_tensors("a", "b", out_key="c") >>> assert "a" not in td >>> assert (td["c"] == torch.tensor([0, 1])).all()
- ceil() T ¶
計算 TensorDict 中每個元素的
ceil()
值。
- ceil_() T ¶
就地計算 TensorDict 中每個元素的
ceil()
值。
- chunk(chunks: int, dim: int = 0) tuple[tensordict.base.TensorDictBase, ...] ¶
如果可能,將 tensordict 分割成指定數量的區塊。
每個區塊都是輸入 tensordict 的視圖。
範例
>>> td = TensorDict({ ... 'x': torch.arange(24).reshape(3, 4, 2), ... }, batch_size=[3, 4]) >>> td0, td1 = td.chunk(dim=-1, chunks=2) >>> td0['x'] tensor([[[ 0, 1], [ 2, 3]], [[ 8, 9], [10, 11]], [[16, 17], [18, 19]]])
- clamp_max(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
如果
self
的元素大於other
,則將它們鉗制為other
的值。- 參數:
other (TensorDict 或 Tensor) – 另一個輸入的 tensordict 或 tensor。
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- clamp_max_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
clamp_max()
的原地 (in-place) 版本。注意
Inplace
clamp_max
不支援default
關鍵字參數。
- clamp_min(other: tensordict.base.TensorDictBase | torch.Tensor, default: Optional[Union[str, Tensor]] = None) T ¶
如果
self
的元素小於other
,則將它們限制為other
的值。- 參數:
other (TensorDict 或 Tensor) – 另一個輸入的 tensordict 或 tensor。
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- clamp_min_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
clamp_min()
的原地 (in-place) 版本。注意
Inplace
clamp_min
不支援default
關鍵字參數。
- clear() T ¶
清除 tensordict 的內容。
- clear_device_() T ¶
清除 tensordict 的裝置。
回傳: self
- clone(recurse: bool = True, **kwargs) T ¶
將 TensorDictBase 子類別實例複製到相同類型的新 TensorDictBase 子類別上。
若要從任何其他 TensorDictBase 子類型建立 TensorDict 實例,請改為呼叫
to_tensordict()
方法。- 參數:
recurse (bool, optional) – 如果
True
,則 TensorDict 中包含的每個張量也會被複製。 否則,只會複製 TensorDict 樹狀結構。 預設值為True
。
注意
與許多其他操作(逐點算術、形狀操作等)不同,
clone
不會繼承原始鎖定屬性。 這種設計選擇是為了可以建立一個可以修改的克隆,這是最常見的用法。
- complex128()¶
將所有張量轉換為
torch.complex128
。
- complex32()¶
將所有張量轉換為
torch.complex32
。
- complex64()¶
將所有張量轉換為
torch.complex64
。
- consolidate(filename: Optional[Union[Path, str]] = None, *, num_threads=0, device: Optional[device] = None, non_blocking: bool = False, inplace: bool = False, return_early: bool = False, use_buffer: bool = False, share_memory: bool = False, pin_memory: bool = False, metadata: bool = False) None ¶
為了快速序列化,此函式會將 tensordict 的內容合併到單一儲存空間中。
- 參數:
filename (Path, optional) – 可選的檔案路徑,用於作為 tensordict 儲存空間的記憶體對應張量。
- 關鍵字參數:
num_threads (integer, optional) – 用於填充儲存空間的執行緒數量。
device (torch.device, optional) – 儲存空間必須實例化的可選裝置。
non_blocking (bool, optional) – 傳遞給
copy_()
的non_blocking
引數。inplace (bool, optional) – 如果
True
,則產生的 tensordict 與具有更新值的self
相同。預設為False
。return_early (bool, optional) – 如果
True
且num_threads>0
,則此方法將傳回 tensordict 的 future。可以使用 future.result() 查詢產生的 tensordict。use_buffer (bool, optional) – 如果
True
且傳遞了檔案名稱,則會在共享記憶體中建立一個中間本機緩衝區,並將資料複製到儲存位置作為最後一個步驟。這可能比直接寫入遠端物理記憶體(例如 NFS)更快。預設為False
。share_memory (bool, optional) – 如果
True
,則儲存空間將放置在共享記憶體中。預設為False
。pin_memory (bool, optional) – 是否應將合併的資料放置在釘選的記憶體中。預設為
False
。metadata (bool, optional) – 如果
True
,則中繼資料將與通用儲存空間一起儲存。 如果提供了檔案名,則此設定無效。 當人們想要控制如何實現序列化時,儲存中繼資料可能很有用,因為如果中繼資料可用或不可用,TensorDict 會以不同的方式處理合併的 TD 的 pickling/unpickling。
注意
如果 tensordict 已經被合併,所有參數都會被忽略並返回
self
。呼叫contiguous()
重新合併。範例
>>> import pickle >>> import tempfile >>> import torch >>> import tqdm >>> from torch.utils.benchmark import Timer >>> from tensordict import TensorDict >>> data = TensorDict({"a": torch.zeros(()), "b": {"c": torch.zeros(())}}) >>> data_consolidated = data.consolidate() >>> # check that the data has a single data_ptr() >>> assert torch.tensor([ ... v.untyped_storage().data_ptr() for v in data_c.values(True, True) ... ]).unique().numel() == 1 >>> # Serializing the tensordict will be faster with data_consolidated >>> with open("data.pickle", "wb") as f: ... print("regular", Timer("pickle.dump(data, f)", globals=globals()).adaptive_autorange()) >>> with open("data_c.pickle", "wb") as f: ... print("consolidated", Timer("pickle.dump(data_consolidated, f)", globals=globals()).adaptive_autorange())
- contiguous() T ¶
返回一個新的 tensordict,其類型相同,具有連續的值(如果值已經是連續的,則返回 self)。
- copy()¶
返回 tensordict 的淺拷貝(即,複製結構但不複製資料)。
等同於 TensorDictBase.clone(recurse=False)
- copy_(tensordict: T, non_blocking: bool = False) T ¶
-
non-blocking 參數將被忽略,僅為了與
torch.Tensor.copy_()
相容而存在。
- copy_at_(tensordict: T, idx: Union[None, int, slice, str, Tensor, List[Any], Tuple[Any, ...]], non_blocking: bool = False) T ¶
- cos() T ¶
計算 TensorDict 中每個元素的
cos()
值。
- cos_() T ¶
就地計算 TensorDict 中每個元素的
cos()
值。
- cosh() T ¶
計算 TensorDict 中每個元素的
cosh()
值。
- cosh_() T ¶
就地計算 TensorDict 中每個元素的
cosh()
值。
- create_nested(key)¶
建立一個具有與當前 tensordict 相同形狀、設備和維度名稱的巢狀 tensordict。
如果該值已經存在,則此操作會覆蓋它。此操作在鎖定的 tensordict 中會被阻止。
範例
>>> data = TensorDict({}, [3, 4, 5]) >>> data.create_nested("root") >>> data.create_nested(("some", "nested", "value")) >>> print(data) TensorDict( fields={ root: TensorDict( fields={ }, batch_size=torch.Size([3, 4, 5]), device=None, is_shared=False), some: TensorDict( fields={ nested: TensorDict( fields={ value: TensorDict( fields={ }, batch_size=torch.Size([3, 4, 5]), device=None, is_shared=False)}, batch_size=torch.Size([3, 4, 5]), device=None, is_shared=False)}, batch_size=torch.Size([3, 4, 5]), device=None, is_shared=False)}, batch_size=torch.Size([3, 4, 5]), device=None, is_shared=False)
- cuda(device: Optional[int] = None, **kwargs) T ¶
將 tensordict 轉換到 cuda 裝置(如果尚未在其上)。
- 參數:
device (int, optional) – 如果提供,tensor 應轉換到的 cuda 裝置。
此函數還支援
to()
的所有關鍵字參數。
- property data¶
傳回一個包含葉張量的 .data 屬性的 tensordict。
- data_ptr(*, storage: bool = False)¶
傳回 tensordict 葉的 data_ptr。
這可以用來檢查兩個 tensordict 是否共享相同的
data_ptr()
。- 關鍵字參數:
storage (bool, optional) – 如果
True
, 將會呼叫 tensor.untyped_storage().data_ptr() 。預設為False
。
範例
>>> from tensordict import TensorDict >>> td = TensorDict(a=torch.randn(2), b=torch.randn(2), batch_size=[2]) >>> assert (td0.data_ptr() == td.data_ptr()).all()
注意
LazyStackedTensorDict
實例將顯示為巢狀 tensordict,以反映其葉的真實data_ptr()
>>> td0 = TensorDict(a=torch.randn(2), b=torch.randn(2), batch_size=[2]) >>> td1 = TensorDict(a=torch.randn(2), b=torch.randn(2), batch_size=[2]) >>> td = TensorDict.lazy_stack([td0, td1]) >>> td.data_ptr() TensorDict( fields={ 0: TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=cpu, is_shared=False), 1: TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=cpu, is_shared=False)}, batch_size=torch.Size([]), device=cpu, is_shared=False)
- densify(*, layout: layout = torch.strided)¶
嘗試用連續的張量(普通張量或巢狀張量)表示惰性堆疊。
- 關鍵字參數:
layout (torch.layout) – 巢狀張量的佈局(如果有的話)。預設為
strided
。
- detach() T ¶
分離 tensordict 中的張量。
- 返回值:
一個新的 tensordict,其中沒有張量需要梯度。
- detach_() T ¶
就地分離 tensordict 中的張量。
- 返回值:
self。
- property device: torch.device | None¶
TensorDict 的裝置。
如果 TensorDict 有指定的裝置,則其所有張量(包括巢狀張量)必須位於相同的裝置上。如果 TensorDict 裝置為
None
,則不同的值可以位於不同的裝置上。- 返回值:
torch.device 物件,指示張量所在的裝置,如果 TensorDict 沒有裝置,則為 None。
範例
>>> td = TensorDict({ ... "cpu": torch.randn(3, device='cpu'), ... "cuda": torch.randn(3, device='cuda'), ... }, batch_size=[], device=None) >>> td['cpu'].device device(type='cpu') >>> td['cuda'].device device(type='cuda') >>> td = TensorDict({ ... "x": torch.randn(3, device='cpu'), ... "y": torch.randn(3, device='cuda'), ... }, batch_size=[], device='cuda') >>> td['x'].device device(type='cuda') >>> td['y'].device device(type='cuda') >>> td = TensorDict({ ... "x": torch.randn(3, device='cpu'), ... "y": TensorDict({'z': torch.randn(3, device='cpu')}, batch_size=[], device=None), ... }, batch_size=[], device='cuda') >>> td['x'].device device(type='cuda') >>> td['y'].device # nested tensordicts are also mapped onto the appropriate device. device(type='cuda') >>> td['y', 'x'].device device(type='cuda')
- dim() int ¶
請參閱
batch_dims()
。
- div(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
將輸入
self
的每個元素除以other
中對應的元素。\[\text{out}_i = \frac{\text{input}_i}{\text{other}_i}\]支援廣播 (broadcasting)、類型提升 (type promotion) 以及整數、浮點數、tensordict 或 tensor 輸入。始終將整數類型提升為預設的純量類型。
- 參數:
other (TensorDict, Tensor 或 Number) – 除數。
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- div_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
div()
的原地 (in-place) 版本。注意
原地
div
不支援default
關鍵字引數。
- double()¶
將所有張量轉換為
torch.bool
。
- property dtype¶
如果 tensordict 中的值的 dtype 是唯一的,則傳回該 dtype。
- dumps(prefix: Optional[str] = None, copy_existing: bool = False, *, num_threads: int = 0, return_early: bool = False, share_non_tensor: bool = False) T ¶
將 tensordict 儲存到磁碟。
此函式是
memmap()
的代理。
- empty(recurse=False, *, batch_size=None, device=_NoDefault.ZERO, names=None) T ¶
傳回一個新的、空的 tensordict,具有相同的 device 和 batch size。
- 參數:
recurse (bool, optional) – 如果
True
,則將複製TensorDict
的整個結構,而不包含內容。 否則,只會複製根。 預設為False
。- 關鍵字參數:
batch_size (torch.Size, optional) – tensordict 的新 batch-size。
device (torch.device, optional) – 新的 device。
names (list of str, optional) – 維度名稱。
- entry_class(key: NestedKey) type ¶
傳回條目的類別,可能避免呼叫 isinstance(td.get(key), type)。
只要
get()
的執行成本很高,就應該優先使用此方法,而不是tensordict.get(key).shape
。
- erf() T ¶
計算 TensorDict 中每個元素的
erf()
值。
- erf_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
erf()
值。
- erfc() T ¶
計算 TensorDict 中每個元素的
erfc()
值。
- erfc_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
erfc()
值。
- exclude(*keys: NestedKey, inplace: bool = False) T ¶
排除 tensordict 的鍵,並返回一個不包含這些條目的新 tensordict。
數值不會被複製:對原始或新的 tensordict 的 tensor 進行就地修改,將導致兩個 tensordict 都發生變化。
- 參數:
- 返回值:
一個新的 tensordict(如果
inplace=True
,則為同一個)不包含被排除的條目。
範例
>>> from tensordict import TensorDict >>> td = TensorDict({"a": 0, "b": {"c": 1, "d": 2}}, []) >>> td.exclude("a", ("b", "c")) TensorDict( fields={ b: TensorDict( fields={ d: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> td.exclude("a", "b") TensorDict( fields={ }, batch_size=torch.Size([]), device=None, is_shared=False)
- exp() T ¶
計算 TensorDict 中每個元素的
exp()
值。
- exp_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
exp()
值。
- expand(*args: int, inplace: bool = False) T ¶
根據
expand()
函數擴展 tensordict 的每個 tensor,忽略特徵維度。支援可迭代物件以指定形狀。
範例
>>> td = TensorDict({ ... 'a': torch.zeros(3, 4, 5), ... 'b': torch.zeros(3, 4, 10)}, batch_size=[3, 4]) >>> td_expand = td.expand(10, 3, 4) >>> assert td_expand.shape == torch.Size([10, 3, 4]) >>> assert td_expand.get("a").shape == torch.Size([10, 3, 4, 5])
- expand_as(other: tensordict.base.TensorDictBase | torch.Tensor) TensorDictBase ¶
將 tensordict 的形狀廣播到 other 的形狀,並據此進行擴展。
如果輸入是 tensor 集合(tensordict 或 tensorclass),則葉節點將以一對一的方式擴展。
範例
>>> from tensordict import TensorDict >>> import torch >>> td0 = TensorDict({ ... "a": torch.ones(3, 1, 4), ... "b": {"c": torch.ones(3, 2, 1, 4)}}, ... batch_size=[3], ... ) >>> td1 = TensorDict({ ... "a": torch.zeros(2, 3, 5, 4), ... "b": {"c": torch.zeros(2, 3, 2, 6, 4)}}, ... batch_size=[2, 3], ... ) >>> expanded = td0.expand_as(td1) >>> assert (expanded==1).all() >>> print(expanded) TensorDict( fields={ a: Tensor(shape=torch.Size([2, 3, 5, 4]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([2, 3, 2, 6, 4]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([2, 3]), device=None, is_shared=False)}, batch_size=torch.Size([2, 3]), device=None, is_shared=False)
- expm1() T ¶
計算 TensorDict 中每個元素的
expm1()
值。
- expm1_() T ¶
就地 (in-place) 計算 TensorDict 中每個元素的
expm1()
值。
- filter_empty_()¶
就地 (in-place) 濾除所有空的 tensordict。
- filter_non_tensor_data() T ¶
濾除所有非 tensor 資料。
- flatten(start_dim=0, end_dim=- 1)¶
將 tensordict 的所有 tensors 平坦化 (flatten)。
範例
>>> td = TensorDict({ ... "a": torch.arange(60).view(3, 4, 5), ... "b": torch.arange(12).view(3, 4)}, batch_size=[3, 4]) >>> td_flat = td.flatten(0, 1) >>> td_flat.batch_size torch.Size([12]) >>> td_flat["a"] tensor([[ 0, 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]]) >>> td_flat["b"] tensor([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11])
- flatten_keys(separator: str = '.', inplace: bool = False, is_leaf: Optional[Callable[[Type], bool]] = None) T ¶
遞迴地將巢狀的 tensordict 轉換為平坦的 tensordict。
TensorDict 的類型將會遺失,且結果會是一個簡單的 TensorDict 實例。
- 參數:
範例
>>> data = TensorDict({"a": 1, ("b", "c"): 2, ("e", "f", "g"): 3}, batch_size=[]) >>> data.flatten_keys(separator=" - ") TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b - c: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), e - f - g: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
此方法和
unflatten_keys()
在處理 state-dicts 時特別有用,因為它們可以無縫地將平面字典轉換為模仿模型結構的資料結構。範例
>>> model = torch.nn.Sequential(torch.nn.Linear(3 ,4)) >>> ddp_model = torch.ao.quantization.QuantWrapper(model) >>> state_dict = TensorDict(ddp_model.state_dict(), batch_size=[]).unflatten_keys(".") >>> print(state_dict) TensorDict( fields={ module: TensorDict( fields={ 0: TensorDict( fields={ bias: Tensor(shape=torch.Size([4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Tensor(shape=torch.Size([4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> model_state_dict = state_dict.get("module") >>> print(model_state_dict) TensorDict( fields={ 0: TensorDict( fields={ bias: Tensor(shape=torch.Size([4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Tensor(shape=torch.Size([4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> model.load_state_dict(dict(model_state_dict.flatten_keys(".")))
- float()¶
將所有 tensors 轉換為
torch.float
。
- float16()¶
將所有 tensors 轉換為
torch.float16
。
- float32()¶
將所有 tensors 轉換為
torch.float32
。
- float64()¶
將所有 tensors 轉換為
torch.float64
。
- floor() T ¶
計算 TensorDict 中每個元素的
floor()
值。
- floor_() T ¶
原地 (in-place) 計算 TensorDict 中每個元素的
floor()
值。
- frac() T ¶
計算 TensorDict 中每個元素的
frac()
值。
- frac_() T ¶
原地 (in-place) 計算 TensorDict 中每個元素的
frac()
值。
- classmethod from_dict(input_dict, batch_size=None, device=None, batch_dims=None, stack_dim_name=None, stack_dim=0)¶
傳回一個從字典或另一個
TensorDict
建立的 TensorDict。如果未指定
batch_size
,則傳回可能的最大批次大小。此函數也適用於巢狀字典,或可用於確定巢狀 tensordict 的批次大小。
- 參數:
input_dict (dictionary, optional) – 用作資料來源的字典(相容巢狀鍵)。
batch_size (iterable of int, optional) – tensordict 的批次大小。
device (torch.device 或 相容類型, optional) – TensorDict 的裝置。
batch_dims (int, optional) –
batch_dims
(即,用於考量batch_size
的前導維度的數量)。與batch_size
互斥。請注意,這是 tensordict 的 __最大__ 批次維度數,允許較小的數字。names (list of str, optional) – tensordict 的維度名稱。
範例
>>> input_dict = {"a": torch.randn(3, 4), "b": torch.randn(3)} >>> print(TensorDict.from_dict(input_dict)) TensorDict( fields={ a: Tensor(shape=torch.Size([3, 4]), device=cpu, dtype=torch.float32, is_shared=False), b: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False) >>> # nested dict: the nested TensorDict can have a different batch-size >>> # as long as its leading dims match. >>> input_dict = {"a": torch.randn(3), "b": {"c": torch.randn(3, 4)}} >>> print(TensorDict.from_dict(input_dict)) TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([3, 4]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3, 4]), device=None, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False) >>> # we can also use this to work out the batch sie of a tensordict >>> input_td = TensorDict({"a": torch.randn(3), "b": {"c": torch.randn(3, 4)}}, []) >>> print(TensorDict.from_dict(input_td)) TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([3, 4]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3, 4]), device=None, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False)
- from_dict_instance(input_dict, batch_size=None, device=None, batch_dims=None, names=None)¶
from_dict()
的實例方法版本。與
from_dict()
不同,此方法會嘗試將 tensordict 類型保留在現有樹中(適用於任何現有葉節點)。範例
>>> from tensordict import TensorDict, tensorclass >>> import torch >>> >>> @tensorclass >>> class MyClass: ... x: torch.Tensor ... y: int >>> >>> td = TensorDict({"a": torch.randn(()), "b": MyClass(x=torch.zeros(()), y=1)}) >>> print(td.from_dict_instance(td.to_dict())) TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), b: MyClass( x=Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), y=Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> print(td.from_dict(td.to_dict())) TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ x: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), y: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
- classmethod from_h5(filename, mode='r')¶
從 h5 檔案建立一個 PersistentTensorDict。
此函數會自動確定每個巢狀 tensordict 的批次大小。
- classmethod from_module(module, as_module: bool = False, lock: bool = True, use_state_dict: bool = False)¶
將模組的參數和緩衝區複製到 tensordict 中。
- 參數:
module (nn.Module) – 要從中取得參數的模組。
as_module (bool, optional) – 如果為
True
,則會傳回TensorDictParams
實例,該實例可用於將參數儲存在torch.nn.Module
中。預設為False
。lock (bool, optional) – 如果為
True
,則產生的 tensordict 將會被鎖定。預設為True
。use_state_dict (bool, optional) –
如果為
True
,則會使用來自模組的 state-dict,並將其解扁平化為一個 TensorDict,其具有模型的樹狀結構。預設為False
。注意
當必須使用 state-dict hook 時,這特別有用。
範例
>>> from torch import nn >>> module = nn.TransformerDecoder( ... decoder_layer=nn.TransformerDecoderLayer(nhead=4, d_model=4), ... num_layers=1 ... ) >>> params = TensorDict.from_module(module) >>> print(params["layers", "0", "linear1"]) TensorDict( fields={ bias: Parameter(shape=torch.Size([2048]), device=cpu, dtype=torch.float32, is_shared=False), weight: Parameter(shape=torch.Size([2048, 4]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
- classmethod from_modules(*modules, as_module: bool = False, lock: bool = True, use_state_dict: bool = False, lazy_stack: bool = False, expand_identical: bool = False)¶
檢索多個模組的參數,用於 ensemble learning/vmap 期望應用的特徵。
- 參數:
modules (nn.Module 的序列) – 要從中取得參數的模組。如果模組結構不同,則需要 lazy stack (請參閱下面的
lazy_stack
參數)。- 關鍵字參數:
as_module (bool, optional) – 如果為
True
,將會回傳一個TensorDictParams
實例,可用於在torch.nn.Module
中儲存參數。預設為False
。lock (bool, optional) – 如果為
True
,產生的 tensordict 將被鎖定。預設為True
。use_state_dict (bool, optional) –
如果為
True
,則會使用來自模組的 state-dict,並將其解扁平化為一個 TensorDict,其具有模型的樹狀結構。預設為False
。注意
當必須使用 state-dict hook 時,這特別有用。
lazy_stack (bool, optional) –
參數應該以密集 (densely) 或延遲 (lazily) 堆疊。預設為
False
(密集堆疊)。注意
lazy_stack
和as_module
是互斥的功能。警告
lazy 和 non-lazy 輸出之間有一個關鍵差異,non-lazy 輸出會重新實例化具有所需 batch size 的參數,而
lazy_stack
只會將參數表示為延遲堆疊。這意味著當lazy_stack=True
時,原始參數可以安全地傳遞給最佳化器,而當設定為True
時,則需要傳遞新的參數。警告
雖然使用 lazy stack 來保留原始參數引用很吸引人,但請記住,每次調用
get()
時,lazy stack 都會執行堆疊。這將需要記憶體 (N 倍的參數大小,如果構建了圖形,則更多) 和時間來計算。這也意味著最佳化器將包含更多參數,並且像step()
或zero_grad()
之類的操作將需要更長的時間才能執行。通常,lazy_stack
應保留給極少數的使用案例。expand_identical (bool, optional) – 如果為
True
並且相同的參數 (相同的 identity) 正在堆疊到自身,則將回傳此參數的展開版本。當lazy_stack=True
時,此引數將被忽略。
範例
>>> from torch import nn >>> from tensordict import TensorDict >>> torch.manual_seed(0) >>> empty_module = nn.Linear(3, 4, device="meta") >>> n_models = 2 >>> modules = [nn.Linear(3, 4) for _ in range(n_models)] >>> params = TensorDict.from_modules(*modules) >>> print(params) TensorDict( fields={ bias: Parameter(shape=torch.Size([2, 4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Parameter(shape=torch.Size([2, 4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([2]), device=None, is_shared=False) >>> # example of batch execution >>> def exec_module(params, x): ... with params.to_module(empty_module): ... return empty_module(x) >>> x = torch.randn(3) >>> y = torch.vmap(exec_module, (0, None))(params, x) >>> assert y.shape == (n_models, 4) >>> # since lazy_stack = False, backprop leaves the original params untouched >>> y.sum().backward() >>> assert params["weight"].grad.norm() > 0 >>> assert modules[0].weight.grad is None
使用
lazy_stack=True
時,情況略有不同。>>> params = TensorDict.from_modules(*modules, lazy_stack=True) >>> print(params) LazyStackedTensorDict( fields={ bias: Tensor(shape=torch.Size([2, 4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Tensor(shape=torch.Size([2, 4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, exclusive_fields={ }, batch_size=torch.Size([2]), device=None, is_shared=False, stack_dim=0) >>> # example of batch execution >>> y = torch.vmap(exec_module, (0, None))(params, x) >>> assert y.shape == (n_models, 4) >>> y.sum().backward() >>> assert modules[0].weight.grad is not None
- classmethod from_namedtuple(named_tuple, *, auto_batch_size: bool = False)¶
遞迴地將 namedtuple 轉換為 TensorDict。
- 關鍵字參數:
auto_batch_size (bool, optional) – 如果為
True
,將自動計算 batch size。預設為False
。
範例
>>> from tensordict import TensorDict >>> import torch >>> data = TensorDict({ ... "a_tensor": torch.zeros((3)), ... "nested": {"a_tensor": torch.zeros((3)), "a_string": "zero!"}}, [3]) >>> nt = data.to_namedtuple() >>> print(nt) GenericDict(a_tensor=tensor([0., 0., 0.]), nested=GenericDict(a_tensor=tensor([0., 0., 0.]), a_string='zero!')) >>> TensorDict.from_namedtuple(nt, auto_batch_size=True) TensorDict( fields={ a_tensor: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False), nested: TensorDict( fields={ a_string: NonTensorData(data=zero!, batch_size=torch.Size([3]), device=None), a_tensor: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False)
- classmethod from_pytree(pytree, *, batch_size: Optional[Size] = None, auto_batch_size: bool = False, batch_dims: Optional[int] = None)¶
將 pytree 轉換為 TensorDict 實例。
此方法旨在盡可能保持 pytree 的巢狀結構。
新增額外的非張量鍵,以追蹤每個層級的身分,提供內建的 pytree 到 tensordict 的雙射轉換 API。
目前接受的類別包括列表、元組、具名元組和字典。
注意
對於字典,非 NestedKey 鍵會作為
NonTensorData
實例單獨註冊。注意
可轉換為張量的類型 (例如 int、float 或 np.ndarray) 將會轉換為 torch.Tensor 實例。請注意,此轉換是滿射的:將 tensordict 轉換回 pytree 無法恢復原始類型。
範例
>>> # Create a pytree with tensor leaves, and one "weird"-looking dict key >>> class WeirdLookingClass: ... pass ... >>> weird_key = WeirdLookingClass() >>> # Make a pytree with tuple, lists, dict and namedtuple >>> pytree = ( ... [torch.randint(10, (3,)), torch.zeros(2)], ... { ... "tensor": torch.randn( ... 2, ... ), ... "td": TensorDict({"one": 1}), ... weird_key: torch.randint(10, (2,)), ... "list": [1, 2, 3], ... }, ... {"named_tuple": TensorDict({"two": torch.ones(1) * 2}).to_namedtuple()}, ... ) >>> # Build a TensorDict from that pytree >>> td = TensorDict.from_pytree(pytree) >>> # Recover the pytree >>> pytree_recon = td.to_pytree() >>> # Check that the leaves match >>> def check(v1, v2): >>> assert (v1 == v2).all() >>> >>> torch.utils._pytree.tree_map(check, pytree, pytree_recon) >>> assert weird_key in pytree_recon[1]
- classmethod from_struct_array(struct_array: ndarray, device: Optional[device] = None) T ¶
將結構化的 numpy 陣列轉換為 TensorDict。
產生的 TensorDict 的內容將與 numpy 陣列共享相同的記憶體內容 (這是一個零複製操作)。就地變更結構化 numpy 陣列的值將會影響 TensorDict 的內容。
範例
>>> x = np.array( ... [("Rex", 9, 81.0), ("Fido", 3, 27.0)], ... dtype=[("name", "U10"), ("age", "i4"), ("weight", "f4")], ... ) >>> td = TensorDict.from_struct_array(x) >>> x_recon = td.to_struct_array() >>> assert (x_recon == x).all() >>> assert x_recon.shape == x.shape >>> # Try modifying x age field and check effect on td >>> x["age"] += 1 >>> assert (td["age"] == np.array([10, 4])).all()
- classmethod fromkeys(keys: List[NestedKey], value: Any = 0)¶
從鍵的清單和單一值建立 tensordict。
- 參數:
keys (NestedKey 的列表) – 一個可迭代物件,用於指定新字典的鍵。
value (相容的類型, 選用) – 所有鍵的值。預設為
0
。
- gather(dim: int, index: Tensor, out: Optional[T] = None) T ¶
沿著 dim 指定的軸收集值。
- 參數:
dim (int) – 沿著哪個維度收集元素
index (torch.Tensor) – 一個長張量,其維度數量與 tensordict 的維度數量相符,兩個張量只有一個維度不同 (收集維度)。其元素指的是沿所需維度收集的索引。
out (TensorDictBase, 選用) – 目的 tensordict。它必須具有與索引相同的形狀。
範例
>>> td = TensorDict( ... {"a": torch.randn(3, 4, 5), ... "b": TensorDict({"c": torch.zeros(3, 4, 5)}, [3, 4, 5])}, ... [3, 4]) >>> index = torch.randint(4, (3, 2)) >>> td_gather = td.gather(dim=1, index=index) >>> print(td_gather) TensorDict( fields={ a: Tensor(shape=torch.Size([3, 2, 5]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([3, 2, 5]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3, 2, 5]), device=None, is_shared=False)}, batch_size=torch.Size([3, 2]), device=None, is_shared=False)
Gather 會保留維度名稱。
範例
>>> td.names = ["a", "b"] >>> td_gather = td.gather(dim=1, index=index) >>> td_gather.names ["a", "b"]
- gather_and_stack(dst: int, group: 'dist.ProcessGroup' | None = None) T | None ¶
從各個 worker 收集 tensordict,並將它們堆疊到目標 worker 的 self 上。
- 參數:
dst (int) – 目標 worker 的 rank,
gather_and_stack()
將在此 worker 上被呼叫。group (torch.distributed.ProcessGroup, optional) – 如果設定,將使用指定的 process group 進行通訊。 否則,將使用預設的 process group。預設值為
None
。
範例
>>> from torch import multiprocessing as mp >>> from tensordict import TensorDict >>> import torch >>> >>> def client(): ... torch.distributed.init_process_group( ... "gloo", ... rank=1, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... # Create a single tensordict to be sent to server ... td = TensorDict( ... {("a", "b"): torch.randn(2), ... "c": torch.randn(2)}, [2] ... ) ... td.gather_and_stack(0) ... >>> def server(): ... torch.distributed.init_process_group( ... "gloo", ... rank=0, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... # Creates the destination tensordict on server. ... # The first dim must be equal to world_size-1 ... td = TensorDict( ... {("a", "b"): torch.zeros(2), ... "c": torch.zeros(2)}, [2] ... ).expand(1, 2).contiguous() ... td.gather_and_stack(0) ... assert td["a", "b"] != 0 ... print("yuppie") ... >>> if __name__ == "__main__": ... mp.set_start_method("spawn") ... ... main_worker = mp.Process(target=server) ... secondary_worker = mp.Process(target=client) ... ... main_worker.start() ... secondary_worker.start() ... ... main_worker.join() ... secondary_worker.join()
- get(key: NestedKey, *args, **kwargs) Tensor ¶
取得使用輸入鍵儲存的值。
- 參數:
key (str, str 的 tuple) – 要查詢的鍵。 如果是 str 的 tuple,則等同於鏈式呼叫 getattr。
default –
如果在 tensordict 中找不到鍵,則傳回預設值。
警告
目前,如果 tensordict 中不存在某個鍵且未傳遞預設值,則會引發 KeyError。 從 v0.7 開始,此行為將會變更為改為傳回 None 值。 若要採用新的行為,請設定環境變數 export TD_GET_DEFAULTS_TO_NONE=’1’ 或呼叫 :func`~tensordict.set_get_defaults_to_none`。
範例
>>> td = TensorDict({"x": 1}, batch_size=[]) >>> td.get("x") tensor(1) >>> set_get_defaults_to_none(False) # Current default behaviour >>> td.get("y") # Raises KeyError >>> set_get_defaults_to_none(True) >>> td.get("y") None
- get_at(key: NestedKey, index: Union[None, int, slice, str, Tensor, List[Any], Tuple[Any, ...]], default: Tensor = _NoDefault.ZERO) Tensor ¶
從索引 idx 的鍵 key 取得 tensordict 的值。
- 參數:
key (str, str 的 tuple) – 要檢索的鍵。
index (int, slice, torch.Tensor, iterable) – tensor 的索引。
default (torch.Tensor) – 如果 tensordict 中不存在鍵,則傳回預設值。
- 返回值:
已索引的 tensor。
範例
>>> td = TensorDict({"x": torch.arange(3)}, batch_size=[]) >>> td.get_at("x", index=1) tensor(1)
- get_item_shape(key)¶
取得 lazy stack 中項目的形狀。
異質維度會以 -1 傳回。
此實作效率不高,因為它會嘗試堆疊項目以計算其形狀,因此僅應用於列印。
- get_nestedtensor(key: NestedKey, default: Any = _NoDefault.ZERO, *, layout: Optional[layout] = None) Tensor ¶
當無法達成堆疊時,傳回巢狀 tensor。
- 參數:
key (NestedKey) – 要巢狀化的條目。
default (Any, optiona) –
如果鍵不在所有子 TensorDict 中,則傳回的預設值。
注意
如果預設值是一個張量,此方法會嘗試建構一個巢狀張量。否則,將傳回預設值。
- 關鍵字參數:
layout (torch.layout, optional) – 巢狀張量的版面配置。
範例
>>> td0 = TensorDict({"a": torch.zeros(4), "b": torch.zeros(4)}, []) >>> td1 = TensorDict({"a": torch.ones(5)}, []) >>> td = torch.stack([td0, td1], 0) >>> a = td.get_nestedtensor("a") >>> # using a tensor as default uses this default to build the nested tensor >>> b = td.get_nestedtensor("b", default=torch.ones(4)) >>> assert (a == b).all() >>> # using anything else as default returns the default >>> b2 = td.get_nestedtensor("b", None) >>> assert b2 is None
- get_non_tensor(key: NestedKey, default=_NoDefault.ZERO)¶
取得一個非張量值,如果存在,如果找不到非張量值,則傳回 default。
此方法對於張量/TensorDict 值具有強大的適應性,這意味著如果收集的值是一個常規張量,它也會被傳回(儘管此方法帶有一些額外負擔,不應在其自然範圍之外使用)。
有關如何在 TensorDict 中設定非張量值的更多資訊,請參閱
set_non_tensor()
。- 參數:
key (NestedKey) – NonTensorData 物件的位置。
default (Any, optional) – 如果找不到鍵,則傳回的值。
- 傳回:
tensordict.tensorclass.NonTensorData
的內容, 或者,如果
key
對應的項目不是tensordict.tensorclass.NonTensorData
,則傳回該項目(如果找不到該項目,則傳回default
)。
範例
>>> data = TensorDict({}, batch_size=[]) >>> data.set_non_tensor(("nested", "the string"), "a string!") >>> assert data.get_non_tensor(("nested", "the string")) == "a string!" >>> # regular `get` works but returns a NonTensorData object >>> data.get(("nested", "the string")) NonTensorData( data='a string!', batch_size=torch.Size([]), device=None, is_shared=False)
- property grad¶
傳回一個 TensorDict,其中包含葉張量的 .grad 屬性。
- half()¶
將所有張量轉換為
torch.half
。
- insert(index: int, tensordict: T) None ¶
將一個 TensorDict 插入堆疊中指定索引處。
類似於 list.insert。插入的 TensorDict 必須具有相容的 batch_size 和 device。插入是原地操作,不傳回任何值。
- 參數:
index (int) – 新 TensorDict 應該插入的索引。
tensordict (TensorDictBase) – 要插入堆疊的 TensorDict。
- int()¶
將所有張量轉換為
torch.int
。
- int16()¶
將所有張量轉換為
torch.int16
。
- int32()¶
將所有張量轉換為
torch.int32
。
- int64()¶
將所有張量轉換為
torch.int64
。
- int8()¶
將所有張量轉換為
torch.int8
。
- irecv(src: int, *, group: 'dist.ProcessGroup' | None = None, return_premature: bool = False, init_tag: int = 0, pseudo_rand: bool = False) tuple[int, list[torch.Future]] | list[torch.Future] | None ¶
非同步地接收 TensorDict 的內容並使用它更新內容。
請參考
isend()
方法中的範例以獲得更多背景資訊。- 參數:
src (int) – 來源工作節點的排名 (rank)。
- 關鍵字參數:
group (torch.distributed.ProcessGroup, optional) – 如果設定,將使用指定的 process group 進行通訊。 否則,將使用預設的 process group。預設值為
None
。return_premature (bool) – 如果
True
,則返回一個 futures 列表,用於等待 tensordict 更新完成。預設值為False
,即在呼叫中等待更新完成。init_tag (int) – 來源工作節點使用的
init_tag
。pseudo_rand (bool) – 如果為 True,標籤序列將是偽隨機的,允許從不同節點發送多個數據而不會重疊。請注意,生成這些偽隨機數的成本很高(1e-5 秒/數字),這意味著它可能會降低演算法的運行速度。此值必須與傳遞給
isend()
的值匹配。預設值為False
。
- 返回值:
- 如果
return_premature=True
,則返回一個用於等待的 futures 列表, 直到 tensordict 更新完成。
- 如果
- is_consolidated()¶
檢查 TensorDict 是否具有合併的儲存空間。
- is_memmap() bool ¶
檢查 tensordict 是否為記憶體映射 (memory-mapped)。
如果 TensorDict 實例是記憶體映射的,則會被鎖定(條目無法重新命名、移除或新增)。如果使用全部為記憶體映射的 tensors 建立
TensorDict
,這並不意味著is_memmap
會返回True
(因為新的 tensor 可能會或可能不會是記憶體映射的)。只有在呼叫 tensordict.memmap_() 時,tensordict 才會被視為記憶體映射的。對於 CUDA 設備上的 tensordicts,這始終為
True
。
檢查 tensordict 是否在共享記憶體中。
如果 TensorDict 實例位於共享記憶體中,則會被鎖定(條目無法重新命名、移除或新增)。如果使用全部位於共享記憶體中的 tensors 建立
TensorDict
,這並不意味著is_shared
會返回True
(因為新的 tensor 可能會或可能不會位於共享記憶體中)。只有在呼叫 tensordict.share_memory_() 或將 tensordict 放置在預設情況下內容共享的設備上(例如,"cuda"
),tensordict 才會被視為位於共享記憶體中。對於 CUDA 設備上的 tensordicts,這始終為
True
。
- isend(dst: int, *, group: 'dist.ProcessGroup' | None = None, init_tag: int = 0, pseudo_rand: bool = False) int ¶
非同步發送 tensordict 的內容。
- 參數:
dst (int) – 目的地工作節點的排名,內容應發送到該節點。
- 關鍵字參數:
group (torch.distributed.ProcessGroup, optional) – 如果設定,將使用指定的 process group 進行通訊。 否則,將使用預設的 process group。預設值為
None
。init_tag (int) – 用於標記 tensors 的初始標籤。 請注意,這將增加 TensorDict 中包含的 tensors 的數量。
pseudo_rand (bool) – 如果為 True,標籤序列將會是虛擬隨機的,允許從不同節點傳送多個資料而不會重疊。請注意,產生這些虛擬隨機數的成本很高(每個數字 1e-5 秒),這意味著它可能會降低演算法的運行時間。預設值為
False
。
範例
>>> import torch >>> from tensordict import TensorDict >>> from torch import multiprocessing as mp >>> def client(): ... torch.distributed.init_process_group( ... "gloo", ... rank=1, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... ... td = TensorDict( ... { ... ("a", "b"): torch.randn(2), ... "c": torch.randn(2, 3), ... "_": torch.ones(2, 1, 5), ... }, ... [2], ... ) ... td.isend(0) ... >>> >>> def server(queue, return_premature=True): ... torch.distributed.init_process_group( ... "gloo", ... rank=0, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... td = TensorDict( ... { ... ("a", "b"): torch.zeros(2), ... "c": torch.zeros(2, 3), ... "_": torch.zeros(2, 1, 5), ... }, ... [2], ... ) ... out = td.irecv(1, return_premature=return_premature) ... if return_premature: ... for fut in out: ... fut.wait() ... assert (td != 0).all() ... queue.put("yuppie") ... >>> >>> if __name__ == "__main__": ... queue = mp.Queue(1) ... main_worker = mp.Process( ... target=server, ... args=(queue, ) ... ) ... secondary_worker = mp.Process(target=client) ... ... main_worker.start() ... secondary_worker.start() ... out = queue.get(timeout=10) ... assert out == "yuppie" ... main_worker.join() ... secondary_worker.join()
- isfinite() T ¶
傳回一個新的 tensordict,其中包含布林元素,表示每個元素是否為有限值。
當實數值不是 NaN、負無限大或無限大時,它們是有限的。當複數的實部和虛部都是有限值時,複數值才是有限的。
- isnan() T ¶
傳回一個新的 tensordict,其中包含布林元素,表示輸入的每個元素是否為 NaN。
當複數的實部和/或虛部為 NaN 時,複數值被視為 NaN。
- isneginf() T ¶
測試輸入的每個元素是否為負無限大。
- isposinf() T ¶
測試輸入的每個元素是否為負無限大。
- isreal() T ¶
傳回一個新的 tensordict,其中包含布林元素,表示輸入的每個元素是否為實數值。
- items(include_nested=False, leaves_only=False, is_leaf=None, *, sort: bool = False)¶
傳回 tensordict 的鍵值對產生器。
- 參數:
- 關鍵字參數:
sort (bool, optional) – 是否應對鍵進行排序。對於巢狀鍵,鍵會根據其聯結的名稱進行排序(即,
("a", "key")
將被視為"a.key"
用於排序)。請注意,在處理大型 tensordict 時,排序可能會產生顯著的開銷。預設值為False
。
- keys(include_nested: bool = False, leaves_only: bool = False, is_leaf: Optional[Callable[[Type], bool]] = None, *, sort: bool = False) _LazyStackedTensorDictKeysView ¶
傳回 tensordict 鍵的產生器。
警告
TensorDict 的
keys()
方法會傳回鍵的延遲視圖。如果查詢了keys
但未進行迭代,然後修改了 tensordict,則稍後迭代鍵將會傳回鍵的新配置。- 參數:
- 關鍵字參數:
sort (bool, optional) – 是否應該對鍵進行排序。對於巢狀鍵,鍵會根據其聯結的名稱進行排序(即,
("a", "key")
將被視為"a.key"
用於排序)。請注意,在處理大型 tensordict 時,排序可能會產生顯著的開銷。預設值為False
。
範例
>>> from tensordict import TensorDict >>> data = TensorDict({"0": 0, "1": {"2": 2}}, batch_size=[]) >>> data.keys() ['0', '1'] >>> list(data.keys(leaves_only=True)) ['0'] >>> list(data.keys(include_nested=True, leaves_only=True)) ['0', '1', ('1', '2')]
- classmethod lazy_stack(items: Sequence[TensorDictBase], dim: int = 0, *, device: Optional[Union[device, str, int]] = None, out: Optional[T] = None, stack_dim_name: Optional[str] = None) T ¶
將 tensordict 堆疊成 LazyStackedTensorDict。
- 參數:
items (TensorDictBase 實例的序列) – 要堆疊的 TensorDictBase 實例的序列。
dim (int, optional) – 執行 lazy stack 的維度。預設為 0。
- 關鍵字參數:
device (torch.device, optional) – 在無法從 tensordict 列表推斷的情況下,在 LazyStackedTensorDict 中設定的裝置 (例如,列表為空)。
out (TensorDictBase, optional) – 要在其中寫入資料的 LazyStackedTensorDict。
stack_dim_name (str, optional) – 堆疊維度的名稱。
strict_shape (bool, optional) – 如果
True
,則每個 tensordict 的形狀都必須匹配。預設為False
。
- lerp(end: tensordict.base.TensorDictBase | torch.Tensor, weight: tensordict.base.TensorDictBase | torch.Tensor | float)¶
根據純量或張量
weight
對兩個張量start
(由self
給出) 和end
執行線性內插。\[\text{out}_i = \text{start}_i + \text{weight}_i \times (\text{end}_i - \text{start}_i)\]start
和end
的形狀必須可廣播。如果weight
是一個張量,則weight
、start
和end
的形狀必須可廣播。- 參數:
end (TensorDict) – 具有結束點的 tensordict。
weight (TensorDict, tensor or float) – 內插公式的權重。
- lerp_(end: tensordict.base.TensorDictBase | float, weight: tensordict.base.TensorDictBase | float)¶
lerp()
的原地 (in-place) 版本。
- lgamma() T ¶
計算 TensorDict 中每個元素的
lgamma()
值。
- lgamma_() T ¶
原地 (in-place) 計算 TensorDict 中每個元素的
lgamma()
值。
- classmethod load(prefix: str | pathlib.Path, *args, **kwargs) T ¶
從磁碟載入一個 tensordict。
這個類別方法是
load_memmap()
的代理。
- load_(prefix: str | pathlib.Path, *args, **kwargs)¶
將磁碟中的 tensordict 載入到目前的 tensordict 中。
這個類別方法是
load_memmap_()
的代理。
- classmethod load_memmap(prefix: str | pathlib.Path, device: Optional[device] = None, non_blocking: bool = False, *, out: Optional[TensorDictBase] = None) T ¶
從磁碟載入記憶體對應的 tensordict。
- 參數:
prefix (str 或 資料夾路徑) – 儲存的 tensordict 應該從哪個資料夾擷取的路徑。
device (torch.device 或 等效物, 選擇性) – 如果提供,資料將非同步地轉換到該裝置。支援 “meta” 裝置,在這種情況下,資料不會被載入,而是創建一組空的“meta”張量。這對於了解總模型大小和結構而不實際打開任何檔案非常有用。
non_blocking (bool, 選擇性) – 如果
True
,則在將張量載入到裝置後不會呼叫同步。預設為False
。out (TensorDictBase, 選擇性) – 應將資料寫入的可選 tensordict。
範例
>>> from tensordict import TensorDict >>> td = TensorDict.fromkeys(["a", "b", "c", ("nested", "e")], 0) >>> td.memmap("./saved_td") >>> td_load = TensorDict.load_memmap("./saved_td") >>> assert (td == td_load).all()
此方法還允許載入巢狀 tensordicts。
範例
>>> nested = TensorDict.load_memmap("./saved_td/nested") >>> assert nested["e"] == 0
tensordict 也可以載入到“meta”裝置上,或者作為一個假的張量。
範例
>>> import tempfile >>> td = TensorDict({"a": torch.zeros(()), "b": {"c": torch.zeros(())}}) >>> with tempfile.TemporaryDirectory() as path: ... td.save(path) ... td_load = TensorDict.load_memmap(path, device="meta") ... print("meta:", td_load) ... from torch._subclasses import FakeTensorMode ... with FakeTensorMode(): ... td_load = TensorDict.load_memmap(path) ... print("fake:", td_load) meta: TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=meta, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([]), device=meta, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=meta, is_shared=False)}, batch_size=torch.Size([]), device=meta, is_shared=False) fake: TensorDict( fields={ a: FakeTensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), b: TensorDict( fields={ c: FakeTensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=cpu, is_shared=False)}, batch_size=torch.Size([]), device=cpu, is_shared=False)
- load_memmap_(prefix: str | pathlib.Path)¶
將記憶體對應 tensordict 的內容載入到呼叫
load_memmap_
的 tensordict 中。詳情請參閱
load_memmap()
。
- load_state_dict(state_dict: OrderedDict[str, Any], strict=True, assign=False, from_flatten=False) T ¶
載入一個 state_dict,格式如
state_dict()
,到 tensordict 中。- 參數:
state_dict (OrderedDict) – 要複製的 state_dict。
strict (bool, optional) – 是否嚴格要求
state_dict
中的鍵與此 tensordict 的torch.nn.Module.state_dict()
函數傳回的鍵匹配。預設值:True
assign (bool, optional) – 是否將 state dictionary 中的項目指定到 tensordict 中對應的鍵,而不是將它們原地複製到 tensordict 目前的 tensors 中。當
False
時,會保留目前模組中 tensors 的屬性,而當True
時,會保留 state dict 中 Tensors 的屬性。預設值:False
from_flatten (bool, optional) – 如果
True
,則假設輸入的 state_dict 是扁平化的。預設為False
。
範例
>>> data = TensorDict({"1": 1, "2": 2, "3": {"3": 3}}, []) >>> data_zeroed = TensorDict({"1": 0, "2": 0, "3": {"3": 0}}, []) >>> sd = data.state_dict() >>> data_zeroed.load_state_dict(sd) >>> print(data_zeroed["3", "3"]) tensor(3) >>> # with flattening >>> data_zeroed = TensorDict({"1": 0, "2": 0, "3": {"3": 0}}, []) >>> data_zeroed.load_state_dict(data.state_dict(flatten=True), from_flatten=True) >>> print(data_zeroed["3", "3"]) tensor(3)
- lock_() T ¶
鎖定 tensordict,使其無法進行非原地 (non in-place) 操作。
諸如
set()
、__setitem__()
、update()
、rename_key_()
或其他新增或移除條目的操作將被阻止。此方法可用作裝飾器 (decorator)。
範例
>>> from tensordict import TensorDict >>> td = TensorDict({"a": 1, "b": 2, "c": 3}, batch_size=[]) >>> with td.lock_(): ... assert td.is_locked ... try: ... td.set("d", 0) # error! ... except RuntimeError: ... print("td is locked!") ... try: ... del td["d"] ... except RuntimeError: ... print("td is locked!") ... try: ... td.rename_key_("a", "d") ... except RuntimeError: ... print("td is locked!") ... td.set("a", 0, inplace=True) # No storage is added, moved or removed ... td.set_("a", 0) # No storage is added, moved or removed ... td.update({"a": 0}, inplace=True) # No storage is added, moved or removed ... td.update_({"a": 0}) # No storage is added, moved or removed >>> assert not td.is_locked
- log() T ¶
計算 TensorDict 中每個元素的
log()
值。
- log10() T ¶
計算 TensorDict 中每個元素的
log10()
值。
- log10_() T ¶
計算 TensorDict 中每個元素的
log10()
值,並原地修改 (in-place)。
- log1p() T ¶
計算 TensorDict 中每個元素的
log1p()
值。
- log1p_() T ¶
計算 TensorDict 中每個元素的
log1p()
值,並原地修改 (in-place)。
- log2() T ¶
計算 TensorDict 中每個元素的
log2()
值。
- log2_() T ¶
計算 TensorDict 中每個元素的
log2()
值,並原地修改 (in-place)。
- log_() T ¶
計算 TensorDict 中每個元素的
log()
值,並原地修改 (in-place)。
- make_memmap(key: NestedKey, shape: torch.Size | torch.Tensor, *, dtype: Optional[dtype] = None) MemoryMappedTensor ¶
根據指定的形狀(shape)與資料類型(dtype),建立一個空的記憶體映射張量。
警告
此方法的設計本質上並非鎖定安全(lock-safe)。在多個節點上存在的記憶體映射 TensorDict 實例,需要使用
memmap_refresh_()
方法進行更新。寫入現有的條目將會導致錯誤。
- 參數:
key (NestedKey) – 要寫入的新條目的鍵。如果鍵已經存在於 tensordict 中,則會引發例外。
shape (torch.Size 或等效值, torch.Tensor 用於巢狀張量) – 要寫入的張量的形狀。
- 關鍵字參數:
dtype (torch.dtype, optional) – 新張量的資料類型。
- 返回值:
一個新的記憶體映射張量。
- make_memmap_from_storage(key: NestedKey, storage: UntypedStorage, shape: torch.Size | torch.Tensor, *, dtype: Optional[dtype] = None) MemoryMappedTensor ¶
根據儲存空間(storage)、形狀(shape)以及可能的資料類型(dtype),建立一個空的記憶體映射張量。
警告
此方法的設計本質上並非鎖定安全(lock-safe)。在多個節點上存在的記憶體映射 TensorDict 實例,需要使用
memmap_refresh_()
方法進行更新。注意
如果該儲存空間有相關聯的檔案名稱,則它必須與檔案的新檔案名稱相符。如果它沒有相關聯的檔案名稱,但 tensordict 有相關聯的路徑,則會導致例外情況。
- 參數:
key (NestedKey) – 要寫入的新條目的鍵。如果鍵已經存在於 tensordict 中,則會引發例外。
storage (torch.UntypedStorage) – 用於新的 MemoryMappedTensor 的儲存空間。 必須是物理記憶體儲存空間。
shape (torch.Size 或等效值, torch.Tensor 用於巢狀張量) – 要寫入的張量的形狀。
- 關鍵字參數:
dtype (torch.dtype, optional) – 新張量的資料類型。
- 返回值:
一個具有指定儲存空間的新記憶體映射張量。
- make_memmap_from_tensor(key: NestedKey, tensor: Tensor, *, copy_data: bool = True) MemoryMappedTensor ¶
根據現有的張量,建立一個空的記憶體映射張量。
警告
此方法的設計本質上並非鎖定安全(lock-safe)。在多個節點上存在的記憶體映射 TensorDict 實例,需要使用
memmap_refresh_()
方法進行更新。如果
copy_data
為True
,此方法始終會複製儲存空間的內容(即,儲存空間未共享)。- 參數:
key (NestedKey) – 要寫入的新條目的鍵。如果鍵已經存在於 tensordict 中,則會引發例外。
tensor (torch.Tensor) – 要在物理記憶體上複製的張量。
- 關鍵字參數:
copy_data (bool, optional) – 如果
False
,則新的張量將共享輸入的元數據,例如形狀和資料類型,但內容將為空。默認值為True
。- 返回值:
一個具有指定儲存空間的新記憶體映射張量。
- map(fn: Callable[[TensorDictBase], TensorDictBase | None], dim: int = 0, num_workers: int | None = None, *, out: TensorDictBase | None = None, chunksize: int | None = None, num_chunks: int | None = None, pool: mp.Pool | None = None, generator: torch.Generator | None = None, max_tasks_per_child: int | None = None, worker_threads: int = 1, index_with_generator: bool = False, pbar: bool = False, mp_start_method: str | None = None)¶
將函式映射到 tensordict 在一個維度上的分割。
此方法將會把一個函式應用到一個 tensordict 實例,方法是將其分塊成大小相等的 tensordict,然後將這些操作分派到所需的 worker 數量上。
函式簽章應該是
Callabe[[TensorDict], Union[TensorDict, Tensor]]
。輸出必須支援torch.cat()
操作。此函式必須是可序列化的。注意
當處理儲存在磁碟上的大型資料集(例如,記憶體映射的 tensordict)時,此方法特別有用,因為這些分塊將會是原始資料的零複製切片,可以幾乎零成本地傳遞到各個行程。 這允許以很小的成本處理非常大的資料集(例如,超過 Tb)。
- 參數:
- 關鍵字參數:
out (TensorDictBase, optional) – 輸出的可選容器。其沿著提供的
dim
的批次大小必須與self.ndim
相符。如果是共享或記憶體映射 (memmap) (is_shared()
或is_memmap()
回傳True
),它將在遠端程序中填充,避免資料向內傳輸。否則,來自self
切片的資料將被傳送到該程序,在目前的程序中收集並原地寫入out
。chunksize (int, optional) – 每個資料塊的大小。
chunksize
為 0 將沿著所需的維度解綁 TensorDict,並在應用該函式後重新堆疊它,而chunksize>0
將分割 TensorDict 並在產生的 TensorDict 列表上呼叫torch.cat()
。如果未提供,區塊的數量將等於工作者的數量。對於非常大的 TensorDict,如此大的區塊可能無法容納在記憶體中以進行操作,並且可能需要更多區塊才能使操作實際可行。此引數與num_chunks
互斥。num_chunks (int, optional) – 將 TensorDict 分割成的區塊數量。如果未提供,區塊的數量將等於工作者的數量。對於非常大的 TensorDict,如此大的區塊可能無法容納在記憶體中以進行操作,並且可能需要更多區塊才能使操作實際可行。此引數與
chunksize
互斥。pool (mp.Pool, optional) – 用於執行作業的多進程 Pool 實例。如果未提供,將在
map
方法中建立一個 Pool。generator (torch.Generator, optional) –
用於設定種子的產生器。將從其產生一個基礎種子,並且 Pool 的每個工作者將使用提供的種子遞增一個從
0
到num_workers
的唯一整數來設定種子。如果未提供產生器,將使用一個隨機整數作為種子。要使用未設定種子的工作者,應單獨建立一個 Pool 並直接傳遞給map()
。注意
當提供一個小值的種子時應謹慎,因為這可能會導致實驗之間的自相關,例如:如果要求 8 個工作者並且種子為 4,則工作者的種子範圍將從 4 到 11。如果種子為 5,則工作者的種子範圍將從 5 到 12。這兩個實驗將有 7 個種子的重疊,這可能會對結果產生意想不到的影響。
注意
為工作者設定種子的目的是在每個工作者上都有獨立的種子,而不是在 map 方法的呼叫之間產生可重現的結果。換句話說,兩個實驗可能並且很可能會傳回不同的結果,因為不可能知道哪個工作者會選擇哪個作業。但是,我們可以確保每個工作者都有不同的種子,並且每個工作者上的偽隨機操作將是不相關的。
max_tasks_per_child (int, optional) – 每個子程序選擇的最大作業數量。預設為
None
,即對作業數量沒有限制。worker_threads (int, optional) – 工作者的執行緒數量。預設為
1
。index_with_generator (bool, optional) – 如果
True
,TensorDict 的分割/分塊將在查詢期間完成,從而節省初始化時間。請注意,chunk()
和split()
比索引(在產生器中使用)效率高得多,因此在初始化時獲得處理時間可能會對總執行時間產生負面影響。預設為False
。pbar (bool, optional) – 如果
True
,將顯示進度條。需要有 tqdm 才能使用。預設為False
。mp_start_method (str, optional) – 多進程的啟動方法。如果未提供,將使用預設的啟動方法。可接受的字串為
"fork"
和"spawn"
。請記住,使用"fork"
啟動方法時,"cuda"
張量無法在進程之間共享。如果pool
傳遞給map
方法,則此方法無效。
範例
>>> import torch >>> from tensordict import TensorDict >>> >>> def process_data(data): ... data.set("y", data.get("x") + 1) ... return data >>> if __name__ == "__main__": ... data = TensorDict({"x": torch.zeros(1, 1_000_000)}, [1, 1_000_000]).memmap_() ... data = data.map(process_data, dim=1) ... print(data["y"][:, :10]) ... tensor([[1., 1., 1., 1., 1., 1., 1., 1., 1., 1.]])
- map_iter(fn: Callable[[TensorDictBase], TensorDictBase | None], dim: int = 0, num_workers: int | None = None, *, shuffle: bool = False, chunksize: int | None = None, num_chunks: int | None = None, pool: mp.Pool | None = None, generator: torch.Generator | None = None, max_tasks_per_child: int | None = None, worker_threads: int = 1, index_with_generator: bool = True, pbar: bool = False, mp_start_method: str | None = None)¶
迭代地將函數映射到 tensordict 在一個維度上的分割。
這是
map()
的可迭代版本。此方法會將函數應用於 tensordict 實例,方式是將其分塊為大小相等的 tensordict,並將操作分派到所需數量的 workers 上執行。它會一次產生一個結果。
函數簽章應為
Callabe[[TensorDict], Union[TensorDict, Tensor]]
。該函數必須是可序列化的。注意
當處理儲存在磁碟上的大型資料集(例如,記憶體映射的 tensordict)時,此方法特別有用,因為這些分塊將會是原始資料的零複製切片,可以幾乎零成本地傳遞到各個行程。 這允許以很小的成本處理非常大的資料集(例如,超過 Tb)。
注意
此函數可用於表示資料集並從中載入資料,以類似 dataloader 的方式進行。
- 參數:
- 關鍵字參數:
shuffle (bool, optional) – 指示是否應全局地洗牌索引。如果為
True
,則每個批次將包含非連續的樣本。如果index_with_generator=False
且 `shuffle=True`,則會引發錯誤。預設值為False
。chunksize (int, optional) – 每個資料塊的大小。
chunksize
為 0 將沿著所需的維度解綁 TensorDict,並在應用該函式後重新堆疊它,而chunksize>0
將分割 TensorDict 並在產生的 TensorDict 列表上呼叫torch.cat()
。如果未提供,區塊的數量將等於工作者的數量。對於非常大的 TensorDict,如此大的區塊可能無法容納在記憶體中以進行操作,並且可能需要更多區塊才能使操作實際可行。此引數與num_chunks
互斥。num_chunks (int, optional) – 將 TensorDict 分割成的區塊數量。如果未提供,區塊的數量將等於工作者的數量。對於非常大的 TensorDict,如此大的區塊可能無法容納在記憶體中以進行操作,並且可能需要更多區塊才能使操作實際可行。此引數與
chunksize
互斥。pool (mp.Pool, optional) – 用於執行作業的多進程 Pool 實例。如果未提供,將在
map
方法中建立一個 Pool。generator (torch.Generator, optional) –
用於設定種子的產生器。將從其產生一個基礎種子,並且 Pool 的每個工作者將使用提供的種子遞增一個從
0
到num_workers
的唯一整數來設定種子。如果未提供產生器,將使用一個隨機整數作為種子。要使用未設定種子的工作者,應單獨建立一個 Pool 並直接傳遞給map()
。注意
當提供一個小值的種子時應謹慎,因為這可能會導致實驗之間的自相關,例如:如果要求 8 個工作者並且種子為 4,則工作者的種子範圍將從 4 到 11。如果種子為 5,則工作者的種子範圍將從 5 到 12。這兩個實驗將有 7 個種子的重疊,這可能會對結果產生意想不到的影響。
注意
為工作者設定種子的目的是在每個工作者上都有獨立的種子,而不是在 map 方法的呼叫之間產生可重現的結果。換句話說,兩個實驗可能並且很可能會傳回不同的結果,因為不可能知道哪個工作者會選擇哪個作業。但是,我們可以確保每個工作者都有不同的種子,並且每個工作者上的偽隨機操作將是不相關的。
max_tasks_per_child (int, optional) – 每個子程序選擇的最大作業數量。預設為
None
,即對作業數量沒有限制。worker_threads (int, optional) – 工作者的執行緒數量。預設為
1
。index_with_generator (bool, optional) –
如果
True
,則 tensordict 的分割 / 分塊將在查詢期間完成,從而節省初始化時間。請注意,chunk()
和split()
比索引(在生成器中使用)效率更高,因此在初始化時間節省的處理時間可能會對總運行時間產生負面影響。默認值為True
。注意
index_with_generator
的默認值對於map_iter
和map
是不同的,前者假設在記憶體中儲存 TensorDict 的分割版本的成本非常高昂。pbar (bool, optional) – 如果
True
,將顯示進度條。需要有 tqdm 才能使用。預設為False
。mp_start_method (str, optional) – 多進程的啟動方法。如果未提供,將使用預設的啟動方法。可接受的字串為
"fork"
和"spawn"
。請記住,使用"fork"
啟動方法時,"cuda"
張量無法在進程之間共享。如果pool
傳遞給map
方法,則此方法無效。
範例
>>> import torch >>> from tensordict import TensorDict >>> >>> def process_data(data): ... data.unlock_() ... data.set("y", data.get("x") + 1) ... return data >>> if __name__ == "__main__": ... data = TensorDict({"x": torch.zeros(1, 1_000_000)}, [1, 1_000_000]).memmap_() ... for sample in data.map_iter(process_data, dim=1, chunksize=5): ... print(sample["y"]) ... break ... tensor([[1., 1., 1., 1., 1.]])
- masked_fill(mask: Tensor, value: float | bool) T ¶
masked_fill 的異地版本。
- 參數:
mask (boolean torch.Tensor) – 要填充的值的遮罩。形狀必須與 tensordict 的批次大小相符。
value – 用於填充 tensors 的值。
- 返回值:
self
範例
>>> td = TensorDict(source={'a': torch.zeros(3, 4)}, ... batch_size=[3]) >>> mask = torch.tensor([True, False, False]) >>> td1 = td.masked_fill(mask, 1.0) >>> td1.get("a") tensor([[1., 1., 1., 1.], [0., 0., 0., 0.], [0., 0., 0., 0.]])
- masked_fill_(mask: Tensor, value: float | bool) T ¶
使用所需的值填充與遮罩相對應的值。
- 參數:
mask (boolean torch.Tensor) – 要填充的值的遮罩。形狀必須與 tensordict 的批次大小相符。
value – 用於填充 tensors 的值。
- 返回值:
self
範例
>>> td = TensorDict(source={'a': torch.zeros(3, 4)}, ... batch_size=[3]) >>> mask = torch.tensor([True, False, False]) >>> td.masked_fill_(mask, 1.0) >>> td.get("a") tensor([[1., 1., 1., 1.], [0., 0., 0., 0.], [0., 0., 0., 0.]])
- masked_select(mask: Tensor) T ¶
遮罩 TensorDict 的所有 tensors,並返回一個新的 TensorDict 實例,該實例具有指向遮罩值的相似鍵。
- 參數:
mask (torch.Tensor) – 用於 tensors 的布林遮罩。形狀必須與 TensorDict 的
batch_size
相符。
範例
>>> td = TensorDict(source={'a': torch.zeros(3, 4)}, ... batch_size=[3]) >>> mask = torch.tensor([True, False, False]) >>> td_mask = td.masked_select(mask) >>> td_mask.get("a") tensor([[0., 0., 0., 0.]])
- maximum(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
計算
self
和other
的元素級最大值。- 參數:
other (TensorDict 或 Tensor) – 另一個輸入的 tensordict 或 tensor。
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- maximum_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
maximum()
的原地版本。注意
原地
maximum
不支持default
關鍵字參數。
- classmethod maybe_dense_stack(items: Sequence[TensorDictBase], dim: int = 0, out: Optional[T] = None, strict: bool = False) T ¶
如果可能,密集堆疊 tensors 或 tensordicts,否則堆疊到 LazyStackedTensorDict 上。
範例
>>> td0 = TensorDict({"a": 0}, []) >>> td1 = TensorDict({"b": 0}, []) >>> LazyStackedTensorDict.maybe_dense_stack([td0, td0]) # returns a TensorDict with shape [2] >>> LazyStackedTensorDict.maybe_dense_stack([td0, td1]) # returns a LazyStackedTensorDict with shape [2] >>> LazyStackedTensorDict.maybe_dense_stack(list(torch.randn(2))) # returns a torch.Tensor with shape [2]
- maybe_to_stack()¶
非 tensors 的堆疊和非堆疊之間可互換的佔位符。
- mean(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, dtype: Optional[dtype] = None, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有元素的平均值。
- 參數:
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回 tensor 的所需資料類型。 如果指定,則在執行操作之前,輸入 tensor 會轉換為 dtype。 這對於防止資料類型溢位很有用。 預設值:
None
。reduce (bool, optional) – 如果
True
,則將在所有 TensorDict 值上進行縮減,並傳回單個縮減的 tensor。 預設為False
。
- memmap(prefix: Optional[str] = None, copy_existing: bool = False, *, num_threads: int = 0, return_early: bool = False, share_non_tensor: bool = False, existsok: bool = True) T ¶
將所有 tensors 寫入到新的 tensordict 中,對應的記憶體映射 Tensor。
- 參數:
- 關鍵字參數:
num_threads (int, optional) – 用於寫入 memmap tensors 的執行緒數量。預設為 0。
return_early (bool, optional) – 如果
True
且num_threads>0
,該方法將傳回 tensordict 的 future。share_non_tensor (bool, optional) – 如果
True
,非 tensor 資料將在進程和寫入操作之間共享 (例如,就地更新或設定),並且單一節點內任何 worker 上的操作將更新所有其他 worker 的值。如果非 tensor 葉子的數量很大 (例如,共享大量的非 tensor 資料),這可能會導致 OOM 或類似錯誤。預設為False
。existsok (bool, optional) – 如果
False
,如果 tensor 已經存在於相同的路徑中,則會引發例外。預設為True
。
然後 TensorDict 會被鎖定,這意味著任何非就地寫入操作都會拋出例外 (例如,重新命名、設定或移除條目)。一旦 tensordict 被解鎖,記憶體映射屬性將被設置為
False
,因為跨進程識別不再保證。- 返回值:
如果
return_early=False
,則傳回一個 tensors 儲存在磁碟上的新 tensordict,否則傳回一個TensorDictFuture
實例。
注意
以這種方式序列化對於深度巢狀的 tensordicts 來說可能很慢,因此不建議在訓練迴圈中呼叫此方法。
- memmap_(prefix: Optional[str] = None, copy_existing: bool = False, *, num_threads: int = 0, return_early: bool = False, share_non_tensor: bool = False, existsok: bool = True) T ¶
將所有 tensors 以原地(in-place)方式寫入對應的記憶體映射 Tensor。
- 參數:
- 關鍵字參數:
num_threads (int, optional) – 用於寫入 memmap tensors 的執行緒數量。預設為 0。
return_early (bool, optional) – 如果
True
且num_threads>0
,則此方法將傳回 tensordict 的 future。可以使用 future.result() 查詢產生的 tensordict。share_non_tensor (bool, optional) – 如果
True
,非 tensor 資料將在進程和寫入操作之間共享 (例如,就地更新或設定),並且單一節點內任何 worker 上的操作將更新所有其他 worker 的值。如果非 tensor 葉子的數量很大 (例如,共享大量的非 tensor 資料),這可能會導致 OOM 或類似錯誤。預設為False
。existsok (bool, optional) – 如果
False
,如果 tensor 已經存在於相同的路徑中,則會引發例外。預設為True
。
然後 TensorDict 會被鎖定,這意味著任何非就地寫入操作都會拋出例外 (例如,重新命名、設定或移除條目)。一旦 tensordict 被解鎖,記憶體映射屬性將被設置為
False
,因為跨進程識別不再保證。- 返回值:
如果
return_early=False
,則傳回 self,否則傳回TensorDictFuture
實例。
注意
以這種方式序列化對於深度巢狀的 tensordicts 來說可能很慢,因此不建議在訓練迴圈中呼叫此方法。
- memmap_like(prefix: Optional[str] = None, copy_existing: bool = False, *, existsok: bool = True, num_threads: int = 0, return_early: bool = False, share_non_tensor: bool = False) T ¶
建立一個內容為空的記憶體映射 tensordict,其形狀與原始 tensordict 相同。
- 參數:
- 關鍵字參數:
num_threads (int, optional) – 用於寫入 memmap tensors 的執行緒數量。預設為 0。
return_early (bool, optional) – 如果
True
且num_threads>0
,該方法將傳回 tensordict 的 future。share_non_tensor (bool, optional) – 如果
True
,非 tensor 資料將在進程和寫入操作之間共享 (例如,就地更新或設定),並且單一節點內任何 worker 上的操作將更新所有其他 worker 的值。如果非 tensor 葉子的數量很大 (例如,共享大量的非 tensor 資料),這可能會導致 OOM 或類似錯誤。預設為False
。existsok (bool, optional) – 如果
False
,如果 tensor 已經存在於相同的路徑中,則會引發例外。預設為True
。
然後 TensorDict 會被鎖定,這意味著任何非就地寫入操作都會拋出例外 (例如,重新命名、設定或移除條目)。一旦 tensordict 被解鎖,記憶體映射屬性將被設置為
False
,因為跨進程識別不再保證。- 返回值:
如果
return_early=False
,則傳回一個新的TensorDict
實例,其中資料儲存為記憶體映射 tensors;否則傳回TensorDictFuture
實例。
注意
這是將一組大型緩衝區寫入磁碟的建議方法,因為
memmap_()
將複製資訊,對於大型內容而言,這可能會很慢。範例
>>> td = TensorDict({ ... "a": torch.zeros((3, 64, 64), dtype=torch.uint8), ... "b": torch.zeros(1, dtype=torch.int64), ... }, batch_size=[]).expand(1_000_000) # expand does not allocate new memory >>> buffer = td.memmap_like("/path/to/dataset")
- memmap_refresh_()¶
如果記憶體映射 tensordict 具有
saved_path
,則重新整理其內容。如果未與其關聯任何路徑,此方法將引發例外。
- minimum(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
計算
self
和other
的元素級最小值。- 參數:
other (TensorDict 或 Tensor) – 另一個輸入的 tensordict 或 tensor。
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- minimum_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
minimum()
的原地版本。注意
原地
minimum
不支援default
關鍵字引數。
- mul(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
將
other
乘以self
。\[\text{{out}}_i = \text{{input}}_i \times \text{{other}}_i\]支援廣播、類型提升以及整數、浮點數和複數輸入。
- 參數:
other (TensorDict, Tensor 或 Number) – 要從
self
中減去的張量或數字。- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- mul_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
mul()
的原地 (in-place) 版本。注意
原地 (inplace)
mul
不支援default
關鍵字參數。
- named_apply(fn: Callable, *others: T, nested_keys: bool = False, batch_size: Optional[Sequence[int]] = None, device: torch.device | None = _NoDefault.ZERO, names: Optional[Sequence[str]] = _NoDefault.ZERO, inplace: bool = False, default: Any = _NoDefault.ZERO, filter_empty: Optional[bool] = None, propagate_lock: bool = False, call_on_nested: bool = False, out: Optional[TensorDictBase] = None, **constructor_kwargs) Optional[T] ¶
將一個鍵條件 (key-conditioned) 可呼叫物件應用到 tensordict 中儲存的所有值,並將它們設定在一個新的 atensordict 中。
可呼叫物件的簽名必須為
Callable[Tuple[str, Tensor, ...], Optional[Union[Tensor, TensorDictBase]]]
。- 參數:
fn (Callable) – 要應用到 tensordict 中的 (名稱, 張量) 對的函數。對於每個葉節點 (leaf),只會使用其葉節點名稱 (leaf name) (而不是完整的 NestedKey)。
*others (TensorDictBase 實例, 選擇性) – 如果提供,這些 tensordict 實例應具有與 self 匹配的結構。
fn
參數應接收與 tensordict 數量一樣多的未命名輸入,包括 self。 如果其他 tensordict 缺少條目,則可以透過default
關鍵字參數傳遞預設值。nested_keys (bool, optional) – 若為
True
,則會使用指向 leaf 節點的完整路徑。預設值為False
,也就是只有最後一個字串會傳遞給函式。batch_size (int 序列, 選擇性) – 如果提供,產生的 TensorDict 將具有所需的 batch_size。
batch_size
參數應與轉換後的 batch_size 相符。 這是一個僅限關鍵字的參數。device (torch.device, 選擇性) – 產生的裝置,如果有的話。
names (字串列表, 選擇性) – 新的維度名稱,以防 batch_size 被修改。
inplace (bool, optional) – 若為 True,則會進行原地 (in-place) 變更。預設值為 False。這是一個僅限關鍵字參數 (keyword only argument)。
default (Any, 選擇性) – 其他 tensordict 中遺失條目的預設值。 如果未提供,遺失的條目將引發 KeyError。
filter_empty (bool, optional) – 若為
True
,則會過濾掉空的 tensordicts。這也會降低運算成本,因為不會建立和銷毀空的資料結構。為了向後相容性,預設值為False
。propagate_lock (bool, optional) – 若為
True
,鎖定的 tensordict 會產生另一個鎖定的 tensordict。預設值為False
。call_on_nested (bool, optional) –
若為
True
,則該函式會被調用於第一層的 tensors 和 containers (TensorDict 或 tensorclass)。在這種情況下,func
負責將其呼叫傳播到巢狀層級。這允許在將呼叫傳播到巢狀 tensordicts 時進行細粒度的行為控制。若為False
,則該函式只會被調用於最底層的節點 (leaves),並且apply
會負責將該函式分派到所有節點。>>> td = TensorDict({"a": {"b": [0.0, 1.0]}, "c": [1.0, 2.0]}) >>> def mean_tensor_only(val): ... if is_tensor_collection(val): ... raise RuntimeError("Unexpected!") ... return val.mean() >>> td_mean = td.apply(mean_tensor_only) >>> def mean_any(val): ... if is_tensor_collection(val): ... # Recurse ... return val.apply(mean_any, call_on_nested=True) ... return val.mean() >>> td_mean = td.apply(mean_any, call_on_nested=True)
out (TensorDictBase, optional) –
用於寫入結果的 tensordict。這可以用於避免創建新的 tensordict。
>>> td = TensorDict({"a": 0}) >>> td.apply(lambda x: x+1, out=td) >>> assert (td==1).all()
警告
如果在 tensordict 上執行的操作需要訪問多個鍵才能進行單一計算,則提供一個等於
self
的out
參數可能會導致操作悄悄地提供錯誤的結果。例如>>> td = TensorDict({"a": 1, "b": 1}) >>> td.apply(lambda x: x+td["a"])["b"] # Right! tensor(2) >>> td.apply(lambda x: x+td["a"], out=td)["b"] # Wrong! tensor(3)
**constructor_kwargs – 要傳遞給 TensorDict 建構子的額外關鍵字參數。
- 返回值:
一個具有 transformed_in tensors 的新 tensordict。
範例
>>> td = TensorDict({ ... "a": -torch.ones(3), ... "nested": {"a": torch.ones(3), "b": torch.zeros(3)}}, ... batch_size=[3]) >>> def name_filter(name, tensor): ... if name == "a": ... return tensor >>> td.named_apply(name_filter) TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False), nested: TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False) >>> def name_filter(name, *tensors): ... if name == "a": ... r = 0 ... for tensor in tensors: ... r = r + tensor ... return tensor >>> out = td.named_apply(name_filter, td) >>> print(out) TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False), nested: TensorDict( fields={ a: Tensor(shape=torch.Size([3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False)}, batch_size=torch.Size([3]), device=None, is_shared=False) >>> print(out["a"]) tensor([-1., -1., -1.])
注意
如果函式返回
None
,則該條目將被忽略。這可以用於過濾 tensordict 中的資料。>>> td = TensorDict({"1": 1, "2": 2, "b": {"2": 2, "1": 1}}, []) >>> def name_filter(name, tensor): ... if name == "1": ... return tensor >>> td.named_apply(name_filter) TensorDict( fields={ 1: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: TensorDict( fields={ 1: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
- property names¶
tensordict 的維度名稱。
可以使用
names
參數在建構時設定名稱。另請參閱
refine_names()
以取得有關如何在建構後設定名稱的詳細資訊。
- nanmean(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, dtype: Optional[dtype] = None, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有非 NaN 元素的平均值。
- 參數:
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回 tensor 的所需資料類型。 如果指定,則在執行操作之前,輸入 tensor 會轉換為 dtype。 這對於防止資料類型溢位很有用。 預設值:
None
。reduce (bool, optional) – 如果
True
,則將在所有 TensorDict 值上進行縮減,並傳回單個縮減的 tensor。 預設為False
。
- nansum(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, dtype: Optional[dtype] = None, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
返回輸入 tensordict 中所有非 NaN 元素的總和。
- 參數:
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回 tensor 的所需資料類型。 如果指定,則在執行操作之前,輸入 tensor 會轉換為 dtype。 這對於防止資料類型溢位很有用。 預設值:
None
。reduce (bool, optional) – 如果
True
,則將在所有 TensorDict 值上進行縮減,並傳回單個縮減的 tensor。 預設為False
。
- property ndim: int¶
請參閱
batch_dims()
。
- ndimension() int ¶
請參閱
batch_dims()
。
- neg() T ¶
計算 TensorDict 中每個元素的
neg()
值。
- neg_() T ¶
就地計算 TensorDict 中每個元素的
neg()
值。
- new_empty(*size: Size, dtype: Optional[dtype] = None, device: Union[device, str, int] = _NoDefault.ZERO, requires_grad: bool = False, layout: layout = torch.strided, pin_memory: Optional[bool] = None)¶
返回一個大小為
size
且帶有空 tensors 的 TensorDict。預設情況下,返回的 TensorDict 具有與此 tensordict 相同的
torch.dtype
和torch.device
。- 參數:
size (int...) – 定義輸出 tensor 形狀的整數的列表、tuple 或 torch.Size。
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回的 tensordict 所需的資料類型。預設值:如果
None
,則 torch.dtype 將保持不變。device (torch.device, optional) – 傳回的 tensordict 所需的裝置。預設值:如果
None
,則torch.device
將保持不變。requires_grad (bool, optional) – 如果 autograd 應該記錄傳回的 tensors 上的操作。預設值:
False
。layout (torch.layout, optional) – 傳回的 TensorDict 數值的所需布局。預設值:
torch.strided
。pin_memory (bool, optional) – 如果設定,傳回的 tensor 將配置在釘選記憶體中。僅適用於 CPU tensors。預設值:
False
。
- new_full(size: Size, fill_value, *, dtype: Optional[dtype] = None, device: Union[device, str, int] = _NoDefault.ZERO, requires_grad: bool = False, layout: layout = torch.strided, pin_memory: Optional[bool] = None)¶
傳回一個大小為
size
並以 1 填滿的 TensorDict。預設情況下,返回的 TensorDict 具有與此 tensordict 相同的
torch.dtype
和torch.device
。- 參數:
size (sequence of int) – 一個列表、元組或 torch.Size 的整數,用於定義輸出 tensor 的形狀。
fill_value (scalar) – 用於填滿輸出 tensor 的數字。
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回的 tensordict 所需的資料類型。預設值:如果
None
,則 torch.dtype 將保持不變。device (torch.device, optional) – 傳回的 tensordict 所需的裝置。預設值:如果
None
,則torch.device
將保持不變。requires_grad (bool, optional) – 如果 autograd 應該記錄傳回的 tensors 上的操作。預設值:
False
。layout (torch.layout, optional) – 傳回的 TensorDict 數值的所需布局。預設值:
torch.strided
。pin_memory (bool, optional) – 如果設定,傳回的 tensor 將配置在釘選記憶體中。僅適用於 CPU tensors。預設值:
False
。
- new_ones(*size: Size, dtype: Optional[dtype] = None, device: Union[device, str, int] = _NoDefault.ZERO, requires_grad: bool = False, layout: layout = torch.strided, pin_memory: Optional[bool] = None)¶
傳回一個大小為
size
並以 1 填滿的 TensorDict。預設情況下,返回的 TensorDict 具有與此 tensordict 相同的
torch.dtype
和torch.device
。- 參數:
size (int...) – 定義輸出 tensor 形狀的整數的列表、tuple 或 torch.Size。
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回的 tensordict 所需的資料類型。預設值:如果
None
,則 torch.dtype 將保持不變。device (torch.device, optional) – 傳回的 tensordict 所需的裝置。預設值:如果
None
,則torch.device
將保持不變。requires_grad (bool, optional) – 如果 autograd 應該記錄傳回的 tensors 上的操作。預設值:
False
。layout (torch.layout, optional) – 傳回的 TensorDict 數值的所需布局。預設值:
torch.strided
。pin_memory (bool, optional) – 如果設定,傳回的 tensor 將配置在釘選記憶體中。僅適用於 CPU tensors。預設值:
False
。
- new_tensor(data: torch.Tensor | tensordict.base.TensorDictBase, *, dtype: Optional[dtype] = None, device: Union[device, str, int] = _NoDefault.ZERO, requires_grad: bool = False, pin_memory: Optional[bool] = None)¶
傳回一個新的 TensorDict,其資料為 tensor
data
。預設情況下,傳回的 TensorDict 值具有與此 tensor 相同的
torch.dtype
和torch.device
。data
也可以是 tensor 集合(TensorDict
或tensorclass
),在這種情況下,new_tensor
方法會疊代self
和data
的 tensor 配對。- 參數:
data (torch.Tensor 或 TensorDictBase) – 要複製的資料。
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回的 tensordict 所需的資料類型。預設值:如果
None
,則 torch.dtype 將保持不變。device (torch.device, optional) – 傳回的 tensordict 所需的裝置。預設值:如果
None
,則torch.device
將保持不變。requires_grad (bool, optional) – 如果 autograd 應該記錄傳回的 tensors 上的操作。預設值:
False
。pin_memory (bool, optional) – 如果設定,傳回的 tensor 將配置在釘選記憶體中。僅適用於 CPU tensors。預設值:
False
。
- new_zeros(*size: Size, dtype: Optional[dtype] = None, device: Union[device, str, int] = _NoDefault.ZERO, requires_grad: bool = False, layout: layout = torch.strided, pin_memory: Optional[bool] = None)¶
傳回一個大小為
size
且填滿 0 的 TensorDict。預設情況下,返回的 TensorDict 具有與此 tensordict 相同的
torch.dtype
和torch.device
。- 參數:
size (int...) – 定義輸出 tensor 形狀的整數的列表、tuple 或 torch.Size。
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回的 tensordict 所需的資料類型。預設值:如果
None
,則 torch.dtype 將保持不變。device (torch.device, optional) – 傳回的 tensordict 所需的裝置。預設值:如果
None
,則torch.device
將保持不變。requires_grad (bool, optional) – 如果 autograd 應該記錄傳回的 tensors 上的操作。預設值:
False
。layout (torch.layout, optional) – 傳回的 TensorDict 數值的所需布局。預設值:
torch.strided
。pin_memory (bool, optional) – 如果設定,傳回的 tensor 將配置在釘選記憶體中。僅適用於 CPU tensors。預設值:
False
。
- norm(*, out=None, dtype: torch.dtype | None = None)¶
計算 tensordict 中每個 tensor 的範數 (norm)。
- 關鍵字參數:
out (TensorDict, optional) – 輸出 tensordict。
dtype (torch.dtype, optional) – 輸出 dtype (torch>=2.4)。
- numel() int ¶
批次 (batch) 中的元素總數。
下限為 1,因為具有空形狀的兩個 tensordict 的堆疊將有兩個元素,因此我們認為 tensordict 至少有 1 個元素大。
- numpy()¶
將 tensordict 轉換為 numpy 陣列的(可能為巢狀)字典。
非 Tensor 資料會以原樣暴露出來。
範例
>>> from tensordict import TensorDict >>> import torch >>> data = TensorDict({"a": {"b": torch.zeros(()), "c": "a string!"}}) >>> print(data) TensorDict( fields={ a: TensorDict( fields={ b: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.float32, is_shared=False), c: NonTensorData(data=a string!, batch_size=torch.Size([]), device=None)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> print(data.numpy()) {'a': {'b': array(0., dtype=float32), 'c': 'a string!'}}
- param_count(*, count_duplicates: bool = True) int ¶
計算參數的數量(可索引項目的總數),僅計算 Tensor。
- 關鍵字參數:
count_duplicates (bool) – 是否將重複的張量視為獨立的張量來計數。如果
False
,則只會捨棄嚴格相同的張量(來自共同基礎張量的相同視圖但具有不同的 ID 將被計算兩次)。預設為 True(每個張量都假定為單一副本)。
- permute(*args, **kwargs)¶
傳回一個 tensordict 的視圖,其批次維度根據 dims 排列。
- 參數:
*dims_list (int) – tensordict 批次維度的新排序方式。 或者,可以提供一個整數的可迭代物件。
dims (list of int) – 呼叫 permute(…) 的另一種方式。
- 返回值:
一個新的 tensordict,其批次維度按照所需的順序排列。
範例
>>> tensordict = TensorDict({"a": torch.randn(3, 4, 5)}, [3, 4]) >>> print(tensordict.permute([1, 0])) PermutedTensorDict( source=TensorDict( fields={ a: Tensor(torch.Size([3, 4, 5]), dtype=torch.float32)}, batch_size=torch.Size([3, 4]), device=cpu, is_shared=False), op=permute(dims=[1, 0])) >>> print(tensordict.permute(1, 0)) PermutedTensorDict( source=TensorDict( fields={ a: Tensor(torch.Size([3, 4, 5]), dtype=torch.float32)}, batch_size=torch.Size([3, 4]), device=cpu, is_shared=False), op=permute(dims=[1, 0])) >>> print(tensordict.permute(dims=[1, 0])) PermutedTensorDict( source=TensorDict( fields={ a: Tensor(torch.Size([3, 4, 5]), dtype=torch.float32)}, batch_size=torch.Size([3, 4]), device=cpu, is_shared=False), op=permute(dims=[1, 0]))
- pin_memory(num_threads: Optional[int] = None, inplace: bool = False) T ¶
在儲存的 tensors 上呼叫
pin_memory()
。- 參數:
num_threads (int 或 str) – 如果提供,則是要用於在 leaves 上呼叫
pin_memory
的執行緒數量。預設為None
,這會在ThreadPoolExecutor(max_workers=None)
中設定大量的執行緒。 若要在主執行緒上執行對pin_memory()
的所有呼叫,請傳遞num_threads=0
。inplace (bool, optional) – 如果
True
,則 tensordict 會就地修改。預設為False
。
- pin_memory_(num_threads: int | str = 0) T ¶
在儲存的 tensors 上呼叫
pin_memory()
,並傳回就地修改的 TensorDict。
- pop(key: NestedKey, default: Any = _NoDefault.ZERO) Tensor ¶
從 tensordict 移除並傳回一個值。
如果該值不存在且未提供預設值,則會擲回 KeyError。
- 參數:
key (str 或 nested key) – 要尋找的項目。
default (Any, optional) – 如果找不到 key,則傳回的值。
範例
>>> td = TensorDict({"1": 1}, []) >>> one = td.pop("1") >>> assert one == 1 >>> none = td.pop("1", default=None) >>> assert none is None
- pow(other: tensordict.base.TensorDictBase | torch.Tensor, *, default: Optional[Union[str, Tensor]] = None) T ¶
取
self
中每個元素與other
的次方,並傳回一個帶有結果的 tensor。other
可以是單個float
數字、Tensor 或TensorDict
。當
other
為張量時,input
和other
的形狀必須可廣播 (broadcastable)。- 參數:
other (float, tensor 或 tensordict) – 指數值
- 關鍵字參數:
default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- pow_(other: tensordict.base.TensorDictBase | torch.Tensor) T ¶
pow()
的原地 (in-place) 版本。注意
原地
pow
不支援default
關鍵字引數。
- prod(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, dtype: Optional[dtype] = None, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有元素的數值乘積。
- 參數:
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回 tensor 的所需資料類型。 如果指定,則在執行操作之前,輸入 tensor 會轉換為 dtype。 這對於防止資料類型溢位很有用。 預設值:
None
。reduce (bool, optional) – 如果
True
,則將在所有 TensorDict 值上進行縮減,並傳回單個縮減的 tensor。 預設為False
。
- qint32()¶
將所有張量轉換為
torch.qint32
。
- qint8()¶
將所有張量轉換為
torch.qint8
。
- quint4x2()¶
將所有張量轉換為
torch.quint4x2
。
- quint8()¶
將所有張量轉換為
torch.quint8
。
- reciprocal() T ¶
計算 TensorDict 中每個元素的
reciprocal()
值。
- reciprocal_() T ¶
原地計算 TensorDict 中每個元素的
reciprocal()
值。
- record_stream(stream: Stream)¶
將 tensordict 標記為已被此 stream 使用。
當 tensordict 被釋放時,請確保張量記憶體在解除分配時於 stream 上排隊的所有工作完成之前,不會被其他張量重複使用。
有關更多信息,請參閱
record_stream()
。
- recv(src: int, *, group: 'dist.ProcessGroup' | None = None, init_tag: int = 0, pseudo_rand: bool = False) int ¶
接收一個 tensordict 的內容並使用它來更新內容。
有關上下文,請查看 send 方法中的範例。
- 參數:
src (int) – 來源工作節點的排名 (rank)。
- 關鍵字參數:
- reduce(dst, op=None, async_op=False, return_premature=False, group=None)¶
跨所有機器縮減 tensordict。
只有具有
rank
dst 的進程才會收到最終結果。
- refine_names(*names) T ¶
根據名稱精簡 self 的維度名稱。
精簡是重新命名的一個特殊情況,它「提升」未命名的維度。 None dim 可以被精簡為任何名稱;一個命名的 dim 只能被精簡為相同的名稱。
因為命名的張量可以與未命名的張量共存,精簡名稱提供了一種很好的方式來編寫命名張量感知程式碼,該程式碼適用於命名和未命名的張量。
名稱可能包含最多一個省略符號 (…)。 省略符號會以貪婪的方式展開;它會就地展開,以使用 self.names 中相應索引的名稱來填滿與 self.dim() 相同長度的名稱。
返回值:具有根據輸入命名的維度的相同 tensordict。
範例
>>> td = TensorDict({}, batch_size=[3, 4, 5, 6]) >>> tdr = td.refine_names(None, None, None, "d") >>> assert tdr.names == [None, None, None, "d"] >>> tdr = td.refine_names("a", None, None, "d") >>> assert tdr.names == ["a", None, None, "d"]
- rename(*names, **rename_map)¶
返回一個具有重新命名維度的 tensordict 的副本。
範例
>>> td = TensorDict({}, batch_size=[1, 2, 3 ,4]) >>> td.names = list("abcd") >>> td_rename = td.rename(c="g") >>> assert td_rename.names == list("abgd")
- rename_(*names, **rename_map)¶
與
rename()
相同,但在原地執行重新命名。範例
>>> td = TensorDict({}, batch_size=[1, 2, 3 ,4]) >>> td.names = list("abcd") >>> assert td.rename_(c="g") >>> assert td.names == list("abgd")
- rename_key_(old_key: NestedKey, new_key: NestedKey, safe: bool = False) T ¶
使用新字串重新命名一個鍵,並傳回具有更新鍵名稱的相同 tensordict。
- 參數:
old_key (str 或 巢狀鍵) – 要重新命名的鍵。
new_key (str 或 巢狀鍵) – 條目的新名稱。
safe (bool, optional) – 如果
True
,當新鍵已存在於 TensorDict 中時,會拋出錯誤。
- 返回值:
self
- replace(*args, **kwargs)¶
建立 tensordict 的淺層副本,其中條目已被替換。
接受一個未命名的引數,該引數必須是
TensorDictBase
子類型的字典。 此外,可以使用具名的關鍵字引數更新第一層的條目。- 返回值:
如果輸入非空,則傳回具有更新條目的
self
副本。 如果提供空的字典或未提供字典,且 kwargs 為空,則傳回self
。
- requires_grad_(requires_grad=True) T ¶
變更是否應由 autograd 記錄此張量的運算:就地設定此張量的 requires_grad 屬性。
傳回此 tensordict。
- 參數:
requires_grad (bool, optional) – 是否應由 autograd 記錄此 tensordict 的運算。 預設為
True
。
- reshape(*args, **kwargs) T ¶
傳回具有所需形狀的連續、重塑張量。
- 參數:
*shape (int) – 結果 tensordict 的新形狀。
- 返回值:
具有重塑鍵的 TensorDict
範例
>>> td = TensorDict({ ... 'x': torch.arange(12).reshape(3, 4), ... }, batch_size=[3, 4]) >>> td = td.reshape(12) >>> print(td['x']) torch.Tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11])
- round() T ¶
計算 TensorDict 中每個元素的
round()
值。
- round_() T ¶
就地計算 TensorDict 中每個元素的
round()
值。
- save(prefix: Optional[str] = None, copy_existing: bool = False, *, num_threads: int = 0, return_early: bool = False, share_non_tensor: bool = False) T ¶
將 tensordict 儲存到磁碟。
此函式是
memmap()
的代理。
- property saved_path¶
傳回正在儲存記憶體對應儲存的 TensorDict 的路徑。
只要 is_memmap() 傳回
False
(例如,當 tensordict 被解鎖時),此引數就會消失。
- select(*keys: NestedKey, inplace: bool = False, strict: bool = True) T ¶
選取 tensordict 的鍵,並傳回僅包含選取鍵的新 tensordict。
數值不會被複製:對原始或新的 tensordict 的 tensor 進行就地修改,將導致兩個 tensordict 都發生變化。
- 參數:
- 返回值:
一個新的 tensordict(如果
inplace=True
則為同一個),僅包含選取的鍵。
注意
若要在 tensordict 中選取鍵,並回傳一個不包含這些鍵的版本,請參閱
split_keys()
方法。範例
>>> from tensordict import TensorDict >>> td = TensorDict({"a": 0, "b": {"c": 1, "d": 2}}, []) >>> td.select("a", ("b", "c")) TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> td.select("a", "b") TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), d: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> td.select("this key does not exist", strict=False) TensorDict( fields={ }, batch_size=torch.Size([]), device=None, is_shared=False)
- send(dst: int, *, group: 'dist.ProcessGroup' | None = None, init_tag: int = 0, pseudo_rand: bool = False) None ¶
將 tensordict 的內容傳送給遠端工作者。
- 參數:
dst (int) – 目的地工作節點的排名,內容應發送到該節點。
- 關鍵字參數:
group (torch.distributed.ProcessGroup, optional) – 如果設定,將使用指定的 process group 進行通訊。 否則,將使用預設的 process group。預設值為
None
。init_tag (int) – 用於標記 tensors 的初始標籤。 請注意,這將增加 TensorDict 中包含的 tensors 的數量。
pseudo_rand (bool) – 如果為 True,標籤序列將會是虛擬隨機的,允許從不同節點傳送多個資料而不會重疊。請注意,產生這些虛擬隨機數的成本很高(每個數字 1e-5 秒),這意味著它可能會降低演算法的運行時間。預設值為
False
。
範例
>>> from torch import multiprocessing as mp >>> from tensordict import TensorDict >>> import torch >>> >>> >>> def client(): ... torch.distributed.init_process_group( ... "gloo", ... rank=1, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... ... td = TensorDict( ... { ... ("a", "b"): torch.randn(2), ... "c": torch.randn(2, 3), ... "_": torch.ones(2, 1, 5), ... }, ... [2], ... ) ... td.send(0) ... >>> >>> def server(queue): ... torch.distributed.init_process_group( ... "gloo", ... rank=0, ... world_size=2, ... init_method=f"tcp://127.0.0.1:10003", ... ) ... td = TensorDict( ... { ... ("a", "b"): torch.zeros(2), ... "c": torch.zeros(2, 3), ... "_": torch.zeros(2, 1, 5), ... }, ... [2], ... ) ... td.recv(1) ... assert (td != 0).all() ... queue.put("yuppie") ... >>> >>> if __name__=="__main__": ... queue = mp.Queue(1) ... main_worker = mp.Process(target=server, args=(queue,)) ... secondary_worker = mp.Process(target=client) ... ... main_worker.start() ... secondary_worker.start() ... out = queue.get(timeout=10) ... assert out == "yuppie" ... main_worker.join() ... secondary_worker.join()
- set(key: NestedKey, item: Tensor, inplace: bool = False, *, non_blocking: bool = False, **kwargs: Any) T ¶
設定新的鍵值對。
- 參數:
key (str, str 的 tuple) – 要設定的鍵的名稱。
item (torch.Tensor 或等效物件, TensorDictBase 實例) – 要儲存在 tensordict 中的值。
inplace (bool, optional) – 如果
True
且鍵與 tensordict 中現有的鍵匹配,則會對該鍵值對進行原地更新。如果 inplace 為True
且找不到該條目,則會新增它。如需更嚴格的原地操作,請改用set_()
。預設為False
。
- 關鍵字參數:
non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。- 返回值:
self
範例
>>> td = TensorDict({}, batch_size[3, 4]) >>> td.set("x", torch.randn(3, 4)) >>> y = torch.randn(3, 4, 5) >>> td.set("y", y, inplace=True) # works, even if 'y' is not present yet >>> td.set("y", torch.zeros_like(y), inplace=True) >>> assert (y==0).all() # y values are overwritten >>> td.set("y", torch.ones(5), inplace=True) # raises an exception as shapes mismatch
- set_(key: NestedKey, item: Tensor, *, non_blocking: bool = False) T ¶
將值設定到現有的鍵,同時保留原始儲存空間。
- 參數:
key (str) – 值的名稱
item (torch.Tensor 或相容類型, TensorDictBase) – 要儲存在 tensordict 中的值
- 關鍵字參數:
non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。- 返回值:
self
範例
>>> td = TensorDict({}, batch_size[3, 4]) >>> x = torch.randn(3, 4) >>> td.set("x", x) >>> td.set_("x", torch.zeros_like(x)) >>> assert (x == 0).all()
- set_at_(key: NestedKey, value: Tensor, index: Union[None, int, slice, str, Tensor, List[Any], Tuple[Any, ...]], *, non_blocking: bool = False) T ¶
在
index
所指示的索引位置,就地 (in-place) 設定數值。- 參數:
key (str, str 的 tuple) – 要修改的鍵 (key)。
value (torch.Tensor) – 要在 index 索引位置設定的數值。
- 關鍵字參數:
non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。- 返回值:
self
範例
>>> td = TensorDict({}, batch_size[3, 4]) >>> x = torch.randn(3, 4) >>> td.set("x", x) >>> td.set_at_("x", value=torch.ones(1, 4), index=slice(1)) >>> assert (x[0] == 1).all()
- set_non_tensor(key: NestedKey, value: Any)¶
使用
tensordict.tensorclass.NonTensorData
在 tensordict 中註冊一個非 Tensor 的數值。可以使用
TensorDictBase.get_non_tensor()
或是直接使用 get 來檢索數值,其會回傳tensordict.tensorclass.NonTensorData
物件。回傳:self
範例
>>> data = TensorDict({}, batch_size=[]) >>> data.set_non_tensor(("nested", "the string"), "a string!") >>> assert data.get_non_tensor(("nested", "the string")) == "a string!" >>> # regular `get` works but returns a NonTensorData object >>> data.get(("nested", "the string")) NonTensorData( data='a string!', batch_size=torch.Size([]), device=None, is_shared=False)
- setdefault(key: NestedKey, default: Tensor, inplace: bool = False) Tensor ¶
如果 tensordict 中不存在
key
,則將key
條目插入,其數值為default
。如果 tensordict 中存在
key
,則回傳key
的數值,否則回傳default
。- 參數:
key (str 或 nested key) – 數值的名稱。
default (torch.Tensor 或 相容的類型, TensorDictBase) – 如果 key 尚未存在,則將此數值儲存在 tensordict 中。
- 返回值:
tensordict 中 key 的數值。如果先前未設定 key,則會是 default。
範例
>>> td = TensorDict({}, batch_size=[3, 4]) >>> val = td.setdefault("a", torch.zeros(3, 4)) >>> assert (val == 0).all() >>> val = td.setdefault("a", torch.ones(3, 4)) >>> assert (val == 0).all() # output is still 0
- property shape: Size¶
請參閱
batch_size
。
將 TensorDict 中的所有張量放置在共享記憶體中。
TensorDict 接著會被鎖定,這表示任何非原地 (in-place) 的寫入操作都會拋出例外(例如,重新命名、設定或移除一個條目)。相反地,一旦 tensordict 被解除鎖定,share_memory 屬性就會變為
False
,因為跨進程 (cross-process) 的識別性不再被保證。- 返回值:
self
- sigmoid() T ¶
計算 TensorDict 中每個元素的
sigmoid()
值。
- sigmoid_() T ¶
計算 TensorDict 中每個元素的
sigmoid()
值,原地 (in-place) 執行。
- sign() T ¶
計算 TensorDict 中每個元素的
sign()
值。
- sign_() T ¶
計算 TensorDict 中每個元素的
sign()
值,原地 (in-place) 執行。
- sin() T ¶
計算 TensorDict 中每個元素的
sin()
值。
- sin_() T ¶
計算 TensorDict 中每個元素的
sin()
值,原地 (in-place) 執行。
- sinh() T ¶
計算 TensorDict 中每個元素的
sinh()
值。
- sinh_() T ¶
計算 TensorDict 中每個元素的
sinh()
值,原地 (in-place) 執行。
- size(dim: Optional[int] = None) torch.Size | int ¶
傳回由
dim
指示的維度的大小。如果未指定
dim
,則傳回 TensorDict 的batch_size
屬性。
- property sorted_keys: list[tensordict._nestedkey.NestedKey]¶
傳回按字母順序排序的鍵。
不支援額外的參數。
如果 TensorDict 被鎖定,則鍵會被快取,直到 tensordict 被解除鎖定,以加快執行速度。
- split(split_size: int | list[int], dim: int = 0) list[tensordict.base.TensorDictBase] ¶
以給定的維度中使用指定的大小分割 TensorDict 中的每個張量,就像 torch.split 一樣。
傳回一個
TensorDict
實例的列表,其中包含分割區塊項目的檢視。- 參數:
split_size (int 或 List(int)) – 單個區塊的大小,或每個區塊的大小列表。
dim (int) – 要分割張量的維度。
- 返回值:
一個 TensorDict 的列表,在給定的維度中具有指定的大小。
範例
>>> td = TensorDict({ ... 'x': torch.arange(12).reshape(3, 4), ... }, batch_size=[3, 4]) >>> td0, td1 = td.split([1, 2], dim=0) >>> print(td0['x']) torch.Tensor([[0, 1, 2, 3]])
- split_keys(*key_sets, inplace=False, strict: bool = True, reproduce_struct: bool = False)¶
根據一組或多組鍵分割 tensordict 為子集。
該方法將返回
N+1
個 tensordict,其中N
是提供的參數數量。- 參數:
inplace (bool, optional) – 如果為
True
,則鍵會直接從self
中移除。預設為False
。strict (bool, optional) – 如果為
True
,當缺少鍵時,會引發例外。預設為True
。reproduce_struct (bool, optional) – 如果為
True
,所有返回的 tensordict 都具有與self
相同的樹狀結構,即使某些子 tensordict 不包含任何葉節點。
注意
None
非張量值將被忽略且不會返回。注意
該方法不會檢查所提供列表中的重複項。
範例
>>> td = TensorDict( ... a=0, ... b=0, ... c=0, ... d=0, ... ) >>> td_a, td_bc, td_d = td.split_keys(["a"], ["b", "c"]) >>> print(td_bc)
- sqrt()¶
計算
self
的元素平方根。
- squeeze(*args, **kwargs)¶
壓縮介於 -self.batch_dims+1 和 self.batch_dims-1 之間的所有張量維度,並將它們以新的 tensordict 形式返回。
- 參數:
dim (Optional[int]) – 要壓縮的維度。如果 dim 是
None
,則所有單例維度都將被壓縮。預設為None
。
範例
>>> td = TensorDict({ ... 'x': torch.arange(24).reshape(3, 1, 4, 2), ... }, batch_size=[3, 1, 4]) >>> td = td.squeeze() >>> td.shape torch.Size([3, 4]) >>> td.get("x").shape torch.Size([3, 4, 2])
此操作也可以用作上下文管理器。對原始 tensordict 的更改將以異地 (out-place) 方式進行,即原始張量的內容不會被更改。這也假設 tensordict 未被鎖定(否則,需要解鎖 tensordict)。此功能與隱式壓縮不相容。
>>> td = TensorDict({ ... 'x': torch.arange(24).reshape(3, 1, 4, 2), ... }, batch_size=[3, 1, 4]) >>> with td.squeeze(1) as tds: ... tds.set("y", torch.zeros(3, 4)) >>> assert td.get("y").shape == [3, 1, 4]
- classmethod stack(input, dim=0, *, out=None)¶
沿給定維度將 tensordict 堆疊成單個 tensordict。
此調用等效於調用
torch.stack()
,但與 torch.compile 相容。
- stack_from_tensordict(dim: int = 0, *, sorted: Optional[Union[bool, List[NestedKey]]] = None, out: Optional[Tensor] = None) Tensor ¶
將 tensordict 中的所有條目堆疊成單個張量。
- 參數:
dim ( int, optional) – 條目應該沿著堆疊的維度。
- 關鍵字參數:
sorted ( bool 或 NestedKeys 的列表) – 如果
True
,條目將按字母順序堆疊。 如果False
(預設),將使用 dict 順序。 或者,可以提供金鑰名稱的列表,並且張量將相應地堆疊。 這會產生一些額外開銷,因為金鑰列表將針對 tensordict 中的葉名稱列表進行檢查。out ( torch.Tensor, optional) – 用於堆疊操作的可選目標張量。
- stack_tensors(*keys: NestedKey, out_key: NestedKey, dim: int = 0, keep_entries: bool = False) T ¶
將條目堆疊到一個新條目中,並可能刪除原始值。
- 參數:
keys (NestedKey 的序列) – 要堆疊的條目。
- 關鍵字參數
out_key (NestedKey): 堆疊輸入的新金鑰名稱。 keep_entries (bool, optional): 如果
False
,keys
中的條目將被刪除。預設為
False
。- dim (int, optional): 必須發生堆疊的維度。
預設為
0
。
回傳: self
範例
>>> td = TensorDict(a=torch.zeros(()), b=torch.ones(())) >>> td.stack_tensors("a", "b", out_key="c") >>> assert "a" not in td >>> assert (td["c"] == torch.tensor([0, 1])).all()
- state_dict(destination=None, prefix='', keep_vars=False, flatten=False) OrderedDict[str, Any] ¶
從 tensordict 生成 state_dict。
除非將
flatten
設定為True
,否則 state-dict 的結構仍然是巢狀的。tensordict state-dict 包含重建 tensordict 所需的所有張量和元數據(目前不支援名稱)。
- 參數:
destination ( dict, optional) – 如果提供,tensordict 的狀態將更新到 dict 中,並傳回相同的物件。 否則,將建立並傳回一個
OrderedDict
。 預設值:None
。prefix ( str, optional) – 一個添加到張量名稱的前綴,用於組合 state_dict 中的金鑰。 預設值:
''
。keep_vars ( bool, optional) – 預設情況下,在 state dict 中傳回的
torch.Tensor
項目與 autograd 分離。 如果設定為True
,則不會執行分離。 預設值:False
。flatten ( bool, optional) – 是否應使用
"."
字元展平結構。 預設為False
。
範例
>>> data = TensorDict({"1": 1, "2": 2, "3": {"3": 3}}, []) >>> sd = data.state_dict() >>> print(sd) OrderedDict([('1', tensor(1)), ('2', tensor(2)), ('3', OrderedDict([('3', tensor(3)), ('__batch_size', torch.Size([])), ('__device', None)])), ('__batch_size', torch.Size([])), ('__device', None)]) >>> sd = data.state_dict(flatten=True) OrderedDict([('1', tensor(1)), ('2', tensor(2)), ('3.3', tensor(3)), ('__batch_size', torch.Size([])), ('__device', None)])
- std(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, correction: int = 1, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有元素的標準差值。
- 參數:
- 關鍵字參數:
- sub(other: tensordict.base.TensorDictBase | float, *, alpha: Optional[float] = None, default: Optional[Union[str, Tensor]] = None)¶
從
self
中減去按alpha
縮放的other
。\[\text{{out}}_i = \text{{input}}_i - \text{{alpha}} \times \text{{other}}_i\]支援廣播、類型提升以及整數、浮點數和複數輸入。
- 參數:
other (TensorDict, Tensor 或 Number) – 要從
self
中減去的張量或數字。- 關鍵字參數:
alpha (Number) –
other
的乘數。default (torch.Tensor 或 str, optional) – 用於獨佔條目的預設值。如果沒有提供,則兩個 tensordict 的鍵列表必須完全匹配。如果傳遞
default="intersection"
,則僅考慮相交的鍵集,並且將忽略其他鍵。在所有其他情況下,default
將用於運算兩側的所有缺失條目。
- sub_(other: tensordict.base.TensorDictBase | float, alpha: Optional[float] = None)¶
sub()
的原地版本。注意
原地
sub
不支援default
關鍵字引數。
- sum(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, dtype: Optional[dtype] = None, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有元素的總和值。
- 參數:
- 關鍵字參數:
dtype (torch.dtype, optional) – 傳回 tensor 的所需資料類型。 如果指定,則在執行操作之前,輸入 tensor 會轉換為 dtype。 這對於防止資料類型溢位很有用。 預設值:
None
。reduce (bool, optional) – 如果
True
,則將在所有 TensorDict 值上進行縮減,並傳回單個縮減的 tensor。 預設為False
。
- tan() T ¶
計算 TensorDict 中每個元素的
tan()
值。
- tan_() T ¶
就地計算 TensorDict 中每個元素的
tan()
值。
- tanh() T ¶
計算 TensorDict 中每個元素的
tanh()
值。
- tanh_() T ¶
就地計算 TensorDict 中每個元素的
tanh()
值。
- to(*args, **kwargs) T ¶
將 TensorDictBase 子類別對應到另一個裝置、dtype 或另一個 TensorDictBase 子類別(如果允許)。
不允許將 tensors 轉換為新的 dtype,因為 tensordicts 不一定包含單一 tensor dtype。
- 參數:
device (torch.device, optional) – tensordict 的目標裝置。
dtype (torch.dtype, optional) – tensordict 的目標浮點數或複數 dtype。
tensor (torch.Tensor, optional) – Tensor,其 dtype 和裝置是此 TensorDict 中所有 tensors 的目標 dtype 和裝置。
- 關鍵字參數:
non_blocking (bool, optional) – 操作是否應為 blocking。
memory_format (torch.memory_format, optional) – 此 tensordict 中 4D 參數和緩衝區的目標記憶體格式。
batch_size (torch.Size, optional) – 輸出 tensordict 的結果 batch-size。
other (TensorDictBase, optional) –
TensorDict 實例,其 dtype 和裝置是此 TensorDict 中所有 tensors 的目標 dtype 和裝置。
注意
由於
TensorDictBase
實例沒有 dtype,因此 dtype 是從範例 leaves 收集的。如果有多個 dtype,則不會進行 dtype 轉換。non_blocking_pin (bool, optional) –
如果
True
,則 tensors 會在傳送到裝置之前被釘選。這將以非同步方式完成,但可以透過num_threads
引數進行控制。注意
呼叫
tensordict.pin_memory().to("cuda")
通常比tensordict.to("cuda", non_blocking_pin=True)
慢得多,因為在第二種情況下,pin_memory 是以非同步方式呼叫的。如果 tensors 很大且數量眾多,則多執行緒pin_memory
通常是有益的:當要傳送的 tensors 太少時,產生執行緒和收集資料的 overhead 會超過多執行緒的優點,如果 tensors 很小,則迭代長列表的 overhead 也會非常大。num_threads ( int 或 None, optional) – 如果
non_blocking_pin=True
,則為pin_memory
使用的執行緒數量。預設情況下,將會產生max(1, torch.get_num_threads())
個執行緒。num_threads=0
將會取消所有對 pin_memory() 呼叫的多執行緒處理。
- 返回值:
如果裝置與 tensordict 裝置不同,以及/或者傳遞了 dtype,則會建立一個新的 tensordict 實例。否則,將會使用相同的 tensordict。只有
batch_size
的修改會在原地執行。
注意
如果 TensorDict 已合併,則產生的 TensorDict 也會被合併。每個新的張量都會是合併儲存體的視圖,並轉換為所需的裝置。
範例
>>> data = TensorDict({"a": 1.0}, [], device=None) >>> data_cuda = data.to("cuda:0") # casts to cuda >>> data_int = data.to(torch.int) # casts to int >>> data_cuda_int = data.to("cuda:0", torch.int) # multiple casting >>> data_cuda = data.to(torch.randn(3, device="cuda:0")) # using an example tensor >>> data_cuda = data.to(other=TensorDict({}, [], device="cuda:0")) # using a tensordict example
- to_dict(*, retain_none: bool = True) dict[str, Any] ¶
傳回一個字典,其鍵值對應於 tensordict 的鍵值對。
- 參數:
retain_none (bool) – 如果
True
,則來自 tensorclass 實例的None
值將會寫入字典中。否則,它們將會被捨棄。預設值:True
。
- to_h5(filename, **kwargs)¶
將 tensordict 轉換為具有 h5 後端的 PersistentTensorDict。
- 參數:
filename (str 或 path) – h5 檔案的路徑。
device (torch.device 或 compatible, optional) – 張量返回後預期的裝置。預設值為
None
(預設在 cpu 上)。**kwargs – 要傳遞給
h5py.File.create_dataset()
的 kwargs。
- 返回值:
連結到新建立檔案的
PersitentTensorDict
實例。
範例
>>> import tempfile >>> import timeit >>> >>> from tensordict import TensorDict, MemoryMappedTensor >>> td = TensorDict({ ... "a": MemoryMappedTensor.from_tensor(torch.zeros(()).expand(1_000_000)), ... "b": {"c": MemoryMappedTensor.from_tensor(torch.zeros(()).expand(1_000_000, 3))}, ... }, [1_000_000]) >>> >>> file = tempfile.NamedTemporaryFile() >>> td_h5 = td.to_h5(file.name, compression="gzip", compression_opts=9) >>> print(td_h5) PersistentTensorDict( fields={ a: Tensor(shape=torch.Size([1000000]), device=cpu, dtype=torch.float32, is_shared=False), b: PersistentTensorDict( fields={ c: Tensor(shape=torch.Size([1000000, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([1000000]), device=None, is_shared=False)}, batch_size=torch.Size([1000000]), device=None, is_shared=False)
- to_module(module: Module, *, inplace: bool | None = None, return_swap: bool = True, swap_dest=None, use_state_dict: bool = False, non_blocking: bool = False, memo=None)¶
將 TensorDictBase 實例的內容遞迴地寫入給定的 nn.Module 屬性中。
to_module
也可以用作上下文管理器,以臨時使用一組參數/緩衝區來填充模組(請參閱下面的範例)。- 參數:
module (nn.Module) – 要將參數寫入的模組。
- 關鍵字參數:
inplace (bool, optional) – 如果
True
,則模組中的參數或張量會就地更新。預設值為False
。return_swap (bool, optional) – 如果
True
,則將傳回舊的參數配置。預設值為False
。swap_dest (TensorDictBase, 可選) – 如果
return_swap
為True
,則為應寫入交換 (swap) 的 tensordict。use_state_dict (bool, 可選) – 如果
True
,將使用 state-dict API 來載入參數(包括 state-dict hooks)。預設值為False
。non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。
範例
>>> from torch import nn >>> module = nn.TransformerDecoder( ... decoder_layer=nn.TransformerDecoderLayer(nhead=4, d_model=4), ... num_layers=1) >>> params = TensorDict.from_module(module) >>> params.data.zero_() >>> params.to_module(module) >>> assert (module.layers[0].linear1.weight == 0).all()
將 tensordict 作為上下文管理器使用,對於進行函數式呼叫會很有用: .. rubric:: 範例
>>> from tensordict import from_module >>> module = nn.TransformerDecoder( ... decoder_layer=nn.TransformerDecoderLayer(nhead=4, d_model=4), ... num_layers=1) >>> params = TensorDict.from_module(module) >>> params = params.data * 0 # Use TensorDictParams to remake these tensors regular nn.Parameter instances >>> with params.to_module(module): ... # Call the module with zeroed params ... y = module(*inputs) >>> # The module is repopulated with its original params >>> assert (TensorDict.from_module(module) != 0).any()
- 返回值:
如果
return_swap
為True
,則為包含模組值的 tensordict,否則為None
。
- to_namedtuple(dest_cls: Optional[type] = None)¶
將 tensordict 轉換為 namedtuple。
- 參數:
dest_cls (Type, 可選) – 一個可選的 namedtuple 類別以供使用。
範例
>>> from tensordict import TensorDict >>> import torch >>> data = TensorDict({ ... "a_tensor": torch.zeros((3)), ... "nested": {"a_tensor": torch.zeros((3)), "a_string": "zero!"}}, [3]) >>> data.to_namedtuple() GenericDict(a_tensor=tensor([0., 0., 0.]), nested=GenericDict(a_tensor=tensor([0., 0., 0.]), a_string='zero!'))
- to_padded_tensor(padding=0.0, mask_key: Optional[NestedKey] = None)¶
將所有巢狀 tensors 轉換為已填充 (padded) 的版本,並相應地調整批次大小。
- 參數:
padding (float) – tensordict 中 tensors 的填充值。預設值為
0.0
。mask_key (NestedKey, 可選) – 如果提供,則為將寫入有效值遮罩 (mask) 的 key。如果異質維度 (heterogeneous dimension) 不是 tensordict 批次大小的一部分,將導致錯誤。預設值為
None
。
- to_pytree()¶
將 tensordict 轉換為 PyTree。
如果 tensordict 不是從 pytree 建立的,此方法只會返回
self
,而不會進行修改。請參閱
from_pytree()
以取得更多資訊和範例。
- to_struct_array()¶
將 tensordict 轉換為 numpy 結構化陣列。
在
from_struct_array()
-to_struct_array()
迴圈中,輸入和輸出陣列的內容應相符。但是,to_struct_array 不會保留原始陣列的記憶體內容。請參閱
from_struct_array()
以取得更多資訊。
- to_tensordict(*, retain_none: Optional[bool] = None)¶
從 TensorDictBase 返回一個規則的 TensorDict 實例。
- 參數:
retain_none (bool) –
如果
True
,tensorclass 實例中的None
值將寫入 tensordict。否則,它們將被捨棄。預設值:True
。注意
從 v0.8 開始,預設值將切換為
False
。- 返回值:
包含相同值的新 TensorDict 物件。
- tolist()¶
將
tensordict.tensorclass.NonTensorStack
的內容提取到巢狀列表 (nested list) 中。範例
>>> from tensordict import NonTensorData >>> import torch >>> data = torch.stack([ ... torch.stack([NonTensorData(data=(i, j), batch_size=[]) for i in range(2)]) ... for j in range(3)]) >>> data.tolist() [[(0, 0), (1, 0)], [(0, 1), (1, 1)], [(0, 2), (1, 2)]]
- transpose(dim0, dim1)¶
返回一個 tensordict,它是輸入的轉置版本。給定的維度
dim0
和dim1
會被交換。轉置 tensordict 的原地 (in-place) 或異地 (out-place) 修改也會影響原始 tensordict,因為記憶體是共享的,並且操作會映射回原始 tensordict。
範例
>>> tensordict = TensorDict({"a": torch.randn(3, 4, 5)}, [3, 4]) >>> tensordict_transpose = tensordict.transpose(0, 1) >>> print(tensordict_transpose.shape) torch.Size([4, 3]) >>> tensordict_transpose.set("b",, torch.randn(4, 3)) >>> print(tensordict.get("b").shape) torch.Size([3, 4])
- trunc() T ¶
計算 TensorDict 每個元素的
trunc()
值。
- trunc_() T ¶
原地計算 TensorDict 每個元素的
trunc()
值。
- uint16()¶
將所有 tensors 轉換為
torch.uint16
。
- uint32()¶
將所有 tensors 轉換為
torch.uint32
。
- uint64()¶
將所有 tensors 轉換為
torch.uint64
。
- uint8()¶
將所有 tensors 轉換為
torch.uint8
。
- unbind(dim: int) tuple[T, ...] ¶
沿指定的維度返回一個索引的 tensordicts 元組。
範例
>>> td = TensorDict({ ... 'x': torch.arange(12).reshape(3, 4), ... }, batch_size=[3, 4]) >>> td0, td1, td2 = td.unbind(0) >>> td0['x'] tensor([0, 1, 2, 3]) >>> td1['x'] tensor([4, 5, 6, 7])
- unflatten(dim, unflattened_size)¶
展開 tensordict dim,將其擴展到期望的形狀。
- 參數:
dim (int) – 指定要展開的輸入 tensor 的維度。
unflattened_size (shape) – 是 tensordict 展開維度的新形狀。
範例
>>> td = TensorDict({ ... "a": torch.arange(60).view(3, 4, 5), ... "b": torch.arange(12).view(3, 4)}, ... batch_size=[3, 4]) >>> td_flat = td.flatten(0, 1) >>> td_unflat = td_flat.unflatten(0, [3, 4]) >>> assert (td == td_unflat).all()
- unflatten_keys(separator: str = '.', inplace: bool = False) T ¶
遞迴地將平面 tensordict 轉換為巢狀 tensordict。
TensorDict 類型將會遺失,結果將會是一個簡單的 TensorDict 實例。 巢狀 tensordict 的 metadata 將從 root 推斷:資料樹中的所有實例將共享相同的 batch-size、維度名稱和設備。
- 參數:
範例
>>> data = TensorDict({"a": 1, "b - c": 2, "e - f - g": 3}, batch_size=[]) >>> data.unflatten_keys(separator=" - ") TensorDict( fields={ a: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False), b: TensorDict( fields={ c: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False), e: TensorDict( fields={ f: TensorDict( fields={ g: Tensor(shape=torch.Size([]), device=cpu, dtype=torch.int64, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)
此方法和
unflatten_keys()
在處理 state-dicts 時特別有用,因為它們可以無縫地將平面字典轉換為模仿模型結構的資料結構。範例
>>> model = torch.nn.Sequential(torch.nn.Linear(3 ,4)) >>> ddp_model = torch.ao.quantization.QuantWrapper(model) >>> state_dict = TensorDict(ddp_model.state_dict(), batch_size=[]).unflatten_keys(".") >>> print(state_dict) TensorDict( fields={ module: TensorDict( fields={ 0: TensorDict( fields={ bias: Tensor(shape=torch.Size([4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Tensor(shape=torch.Size([4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> model_state_dict = state_dict.get("module") >>> print(model_state_dict) TensorDict( fields={ 0: TensorDict( fields={ bias: Tensor(shape=torch.Size([4]), device=cpu, dtype=torch.float32, is_shared=False), weight: Tensor(shape=torch.Size([4, 3]), device=cpu, dtype=torch.float32, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False)}, batch_size=torch.Size([]), device=None, is_shared=False) >>> model.load_state_dict(dict(model_state_dict.flatten_keys(".")))
- unsqueeze(*args, **kwargs)¶
將所有 tensors 在 -td.batch_dims 和 td.batch_dims 之間的維度上進行 unsqueeze 操作,並將其返回到一個新的 tensordict 中。
- 參數:
dim (int) – 沿著該維度進行 unsqueeze 操作
範例
>>> td = TensorDict({ ... 'x': torch.arange(24).reshape(3, 4, 2), ... }, batch_size=[3, 4]) >>> td = td.unsqueeze(-2) >>> td.shape torch.Size([3, 1, 4]) >>> td.get("x").shape torch.Size([3, 1, 4, 2])
此操作也可以用作上下文管理器。 對原始 tensordict 的更改將會異地發生,即原始 tensors 的內容不會被更改。 這也假設 tensordict 沒有被鎖定(否則,需要解鎖 tensordict)。
>>> td = TensorDict({ ... 'x': torch.arange(24).reshape(3, 4, 2), ... }, batch_size=[3, 4]) >>> with td.unsqueeze(-2) as tds: ... tds.set("y", torch.zeros(3, 1, 4)) >>> assert td.get("y").shape == [3, 4]
- update(input_dict_or_td: Union[dict[str, torch.Tensor], T], clone: bool = False, inplace: bool = False, *, non_blocking: bool = False, keys_to_update: Optional[Sequence[NestedKey]] = None, is_leaf: Optional[Callable[[Type], bool]] = None) T ¶
使用字典或另一個 TensorDict 的值更新 TensorDict。
- 參數:
input_dict_or_td (TensorDictBase 或 dict) – 要寫入 self 的輸入資料。
clone (bool, optional) – 是否應在設定之前複製輸入(tensor)字典中的 tensors。預設值為
False
。inplace (bool, optional) – 如果
True
且鍵與 tensordict 中現有的鍵匹配,則更新將針對該鍵值對就地發生。如果找不到該條目,將會新增它。預設值為False
。
- 關鍵字參數:
keys_to_update (NestedKeys 的序列, optional) – 如果提供,則只會更新
key_to_update
中的鍵列表。 這是為了避免呼叫data_dest.update(data_src.select(*keys_to_update))
。non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。is_leaf (Callable[[Type], bool], optional) – 一個可呼叫物件,指示物件類型是否應被視為葉節點並交換,還是張量集合。
- 返回值:
self
範例
>>> td = TensorDict({}, batch_size=[3]) >>> a = torch.randn(3) >>> b = torch.randn(3, 4) >>> other_td = TensorDict({"a": a, "b": b}, batch_size=[]) >>> td.update(other_td, inplace=True) # writes "a" and "b" even though they can't be found >>> assert td['a'] is other_td['a'] >>> other_td = other_td.clone().zero_() >>> td.update(other_td) >>> assert td['a'] is not other_td['a']
- update_(input_dict_or_td: Union[dict[str, torch.Tensor], T], clone: bool = False, *, non_blocking: bool = False, keys_to_update: Optional[Sequence[NestedKey]] = None) T ¶
使用字典或另一個 TensorDict 中的值,就地更新 TensorDict。
與
update()
不同,如果self
不認識此鍵,此函數會拋出錯誤。- 參數:
input_dict_or_td (TensorDictBase 或 dict) – 要寫入 self 的輸入資料。
clone (bool, optional) – 是否應在設定之前複製輸入(tensor)字典中的 tensors。預設值為
False
。
- 關鍵字參數:
keys_to_update (NestedKeys 的序列, 選用) – 如果提供,則只會更新
key_to_update
中的鍵列表。 目的是為了避免呼叫data_dest.update_(data_src.select(*keys_to_update))
。non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。
- 返回值:
self
範例
>>> a = torch.randn(3) >>> b = torch.randn(3, 4) >>> td = TensorDict({"a": a, "b": b}, batch_size=[3]) >>> other_td = TensorDict({"a": a*0, "b": b*0}, batch_size=[]) >>> td.update_(other_td) >>> assert td['a'] is not other_td['a'] >>> assert (td['a'] == other_td['a']).all() >>> assert (td['a'] == 0).all()
- update_at_(input_dict_or_td: dict[str, torch.Tensor] | tensordict.base.TensorDictBase, index: Union[None, int, slice, str, Tensor, List[Any], Tuple[Any, ...]], clone: bool = False, *, non_blocking: bool = False) T ¶
使用字典或另一個 TensorDict 中的值,在指定索引處就地更新 TensorDict。
與 TensorDict.update 不同,如果 TensorDict 不認識此鍵,此函數會拋出錯誤。
- 參數:
input_dict_or_td (TensorDictBase 或 dict) – 要寫入 self 的輸入資料。
idx (int, torch.Tensor, 可迭代物件, slice) – 應該進行更新的 tensordict 的索引。
clone (bool, 選用) – 輸入的張量 (tensor) 字典中的張量是否應該在設定前被複製。預設值為 False。
- 關鍵字參數:
keys_to_update (NestedKeys 的序列, 選用) – 如果提供,則只會更新
key_to_update
中的鍵列表。non_blocking (bool, optional) – 如果
True
且此複製發生在不同裝置之間,則此複製可能會相對於主機非同步發生。
- 返回值:
self
範例
>>> td = TensorDict({ ... 'a': torch.zeros(3, 4, 5), ... 'b': torch.zeros(3, 4, 10)}, batch_size=[3, 4]) >>> td.update_at_( ... TensorDict({ ... 'a': torch.ones(1, 4, 5), ... 'b': torch.ones(1, 4, 10)}, batch_size=[1, 4]), ... slice(1, 2)) TensorDict( fields={ a: Tensor(torch.Size([3, 4, 5]), dtype=torch.float32), b: Tensor(torch.Size([3, 4, 10]), dtype=torch.float32)}, batch_size=torch.Size([3, 4]), device=None, is_shared=False) >>> assert (td[1] == 1).all()
- valid_keys(include_nested: bool = False, leaves_only: bool = False, is_leaf: Optional[Callable[[Type], bool]] = None, *, sort: bool = False) _LazyStackedTensorDictKeysView ¶
傳回 tensordict 鍵的產生器。
警告
TensorDict 的
keys()
方法會傳回鍵的延遲視圖。如果查詢了keys
但未進行迭代,然後修改了 tensordict,則稍後迭代鍵將會傳回鍵的新配置。- 參數:
- 關鍵字參數:
sort (bool, optional) – 是否應該對鍵進行排序。對於巢狀鍵,鍵會根據其聯結的名稱進行排序(即,
("a", "key")
將被視為"a.key"
用於排序)。請注意,在處理大型 tensordict 時,排序可能會產生顯著的開銷。預設值為False
。
範例
>>> from tensordict import TensorDict >>> data = TensorDict({"0": 0, "1": {"2": 2}}, batch_size=[]) >>> data.keys() ['0', '1'] >>> list(data.keys(leaves_only=True)) ['0'] >>> list(data.keys(include_nested=True, leaves_only=True)) ['0', '1', ('1', '2')]
- values(include_nested=False, leaves_only=False, is_leaf=None, *, sort: bool = False)¶
傳回一個產生器,表示 tensordict 的值。
- 參數:
- 關鍵字參數:
sort (bool, optional) – 是否應對鍵進行排序。對於巢狀鍵,鍵會根據其聯結的名稱進行排序(即,
("a", "key")
將被視為"a.key"
用於排序)。請注意,在處理大型 tensordict 時,排序可能會產生顯著的開銷。預設值為False
。
- var(dim: Union[int, Tuple[int]] = _NoDefault.ZERO, keepdim: bool = _NoDefault.ZERO, *, correction: int = 1, reduce: Optional[bool] = None) tensordict.base.TensorDictBase | torch.Tensor ¶
傳回輸入 tensordict 中所有元素的變異數值。
- 參數:
- 關鍵字參數:
- view(*shape: int, size: list | tuple | torch.Size | None = None, batch_size: torch.Size | None = None)¶
傳回一個 tensordict,其具有根據與 tensordict batch_size 相容的新形狀檢視的 tensors。
或者,可以提供 dtype 作為第一個未命名的引數。在這種情況下,所有 tensors 都會以相應的 dtype 進行檢視。請注意,這假設新形狀會與提供的 dtype 相容。有關 dtype 檢視的更多資訊,請參閱
view()
。- 參數:
*shape (int) – 結果 tensordict 的新形狀。
dtype (torch.dtype) – 或者,用於表示 tensor 內容的 dtype。
size – 可迭代物件
- 關鍵字參數:
batch_size (torch.Size, optional) – 如果提供了 dtype,則可以使用此關鍵字引數重設 batch-size。如果使用形狀呼叫
view
,則這無效。- 返回值:
具有所需 batch_size 的新 tensordict。
範例
>>> td = TensorDict(source={'a': torch.zeros(3,4,5), ... 'b': torch.zeros(3,4,10,1)}, batch_size=torch.Size([3, 4])) >>> td_view = td.view(12) >>> print(td_view.get("a").shape) # torch.Size([12, 5]) >>> print(td_view.get("b").shape) # torch.Size([12, 10, 1]) >>> td_view = td.view(-1, 4, 3) >>> print(td_view.get("a").shape) # torch.Size([1, 4, 3, 5]) >>> print(td_view.get("b").shape) # torch.Size([1, 4, 3, 10, 1])
- where(condition, other, *, out=None, pad=None)¶
根據條件,傳回從 self 或 other 選取的元素的
TensorDict
。- 參數:
condition (BoolTensor) – 當
True
(非零) 時,產生self
,否則產生other
。other (TensorDictBase 或 Scalar) – 值 (如果
other
是一個純量) 或在 condition 為False
的索引處選取的值。
- 關鍵字參數:
out (TensorDictBase, optional) – 輸出的
TensorDictBase
實例。pad (純量, optional) – 如果提供,來源或目標 tensordict 中遺失的鍵將被寫成 torch.where(mask, self, pad) 或 torch.where(mask, pad, other)。預設為
None
,即不容許遺失的鍵。
- zero_() T ¶
就地將 tensordict 中的所有 tensors 歸零。