All pull requests

Pull requests

• [BugFix] Keep writes to an unpickled consolidated tensordict in its next pickle

  Test
  #2516·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make NonTensorData.empty(names=...) apply the names

  bugtensorclassTest
  #2515·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Keep the dim names in LazyStackedTensorDict.empty()

  Test
  #2514·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make to(dtype_or_device, batch_size=...) cast and move UnbatchedTensor entries

  bugTest
  #2513·peterdsharpe opened Oct 13, 2026Collaborator·

  ·

• [BugFix] Make to(attrs, inplace=True, num_threads=...) work on nested tensordicts

  bugTest
  #2512·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Return values and indices from max and min with a dim and reduce=True

  bugTest
  #2511·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Let add, add_, sub and sub_ take alpha with a number as other

  Test
  #2510·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make TypedTensorDict reshape, masked_select and repeat_interleave with inplace=True return self

  CLA SignedTest
  #2509·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make in-place shape ops on an h5 PersistentTensorDict raise before they change the file

  bugCLA SignedpersistentTest
  #2508·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Give reshape, flatten and unflatten with inplace=True the right dim names

  bugCLA SignedTest
  #2507·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Give reshape and view on a lazy stack with a zero-size dim the shape torch gives

  bugCLA SignedTest
  #2506·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make pad_sequence pad tensordicts that hold only non-tensor data

  bugCLA SignedfunctionalTest
  #2505·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Say why reduce=True fails on a tensordict without tensor leaves

  CLA SignedTest
  #2504·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make store update() work when a sibling leaf differs in shape across elements

  CLA SignedstoreTest
  #2489·peterdsharpe opened Oct 13, 2026Collaborator·

  ·

• [Test] Make test_shared_copy pass on main again

  CLA SignedTest
  #2488·peterdsharpe opened Oct 13, 2026Collaborator·

  ·

• [BugFix] Drop the cached sub-tensordicts that PersistentTensorDict.del_ deletes

  CLA SignedpersistentTest
  #2486·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Raise the ValueError of NonTensorData.where before calling expand

  bugCLA Signedtensorclass
  #2485·peterdsharpe opened Oct 13, 2026Collaborator·

  ·

• [BugFix] Make from_pytree and to_pytree round-trip lists and dicts under int keys

  bugCLA SignedTest
  #2480·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make torch.cat into a lazy-stack out work with empty inputs

  bugCLA SignedTest
  #2479·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make arithmetic between a lazy stack and a batch-shaped tensor work

  bugCLA SignedTest
  #2478·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Stop lazy_stack and .grad from writing into the tensorclasses they read

  bugCLA SigneddocumentationRefactortensorclassTest
  #2477·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  2

• [BugFix] Make NonTensorStack.reshape match torch when a dim has size zero

  bugCLA SignedtensorclassTest
  #2476·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make unflatten raise on a wrong element count without tensor leaves

  bugCLA SignedTest
  #2475·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Make threaded memmap raise on a failed write instead of dropping the entry

  bugCLA SignedTest
  #2474·peterdsharpe opened Oct 13, 2026Collaborator·

  ·
  1

• [BugFix] Stop store[i] from adding new keys to a locked LazyStackedTensorDictStore

  bugCLA SignedstoreTest
  #2473·peterdsharpe opened Oct 13, 2026Collaborator·

  ·