thinc/notes.txt at v8.3.x · explosion/thinc · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Release Notes: thinc v8.3.13

Bug Fixes
---------
- Add clear error message for float64 dtype mismatch in BLIS gemm (#461).
- Document that Model.from_disk requires matching architecture (#751).
- Document that Loss.get_grad computes gradient w.r.t. logits, not
  post-softmax probabilities (#901).
- Remove stale comments about torch.cuda.amp.autocast migration (#967).
- Fix numpy deprecation warning filter syntax in pyproject.toml.
- Ignore NumPy 2.4 align deprecation from pickled dtype in test data.

Build & CI
----------
- Pin all GitHub Actions and dependencies to commit SHAs for
  supply-chain security.
- Cancel in-progress CI builds when a new release tag is pushed.
- Update black requirement to >=24.0 and reformat codebase.

Housekeeping
------------
- Remove test_types.py (pydantic create_model validation tests superseded
  by confection validate_type).