Keras 3.14.0 on Python PyPI

Highlights

Orbax Checkpoint Integration: Full support for Orbax checkpoints, including sharding, remote paths, and step recovery.
Quantization Upgrades: Added support for Activation-aware Weight Quantization (AWQ) and Asymmetric INT4 Sub-Channel Quantization.
Batch Renormalization in BatchNorm: Added batch renormalization feature to the BatchRenormalization layer.
New Optimizer: Added ScheduleFreeAdamW optimizer.
Gated Attention: Introduced optional Gated Attention support in MultiHeadAttention and GroupedQueryAttention layers.

New Features and Operations

Multi-Backend Operations

NaN-aware NumPy Operations: Added support for nanmin, nanmax, nanmean, nanmedian, nanvar, nanstd, nanprod, nanargmin, nanargmax, and nanquantile in keras.ops.numpy.
New Math & Linear Algebra Operators: Added nextafter, ptp, view, sinc, fmod, i0, fliplr, flipud, rad2deg, geomspace, depth_to_space, space_to_depth, and fold.

Preprocessing and Layers

CLAHE Layer: Added Contrast Limited Adaptive Histogram Equalization preprocessing layer.
Adapt Support for Iterables: Preprocessing layers now support Python iterables in the adapt() method, which allows the direct use of Grain datasets.

OpenVINO Backend Support

The OpenVINO backend received a massive update, implementing a wide array of NumPy and Neural Network operations to achieve feature parity with other backends:

NumPy Operations: vander, trapezoid, corrcoef, correlate, flip, diagonal, cbrt, hypot, trace, kron, argpartition, logaddexp2, ldexp, select, round, vstack, hsplit, vsplit, tile, nansum, tensordot, exp2, trunc, gcd, unravel_index, inner, cumprod, searchsorted, hanning, diagflat, norm, histogram, lcm, allclose, real, imag, isreal, kaiser, shuffle, einsum, quantile, conj, randint, in_top_k, signbit, gamma, heaviside, var, std, inv, solve, cholesky_inverse, fft, fft2, ifft2, rfft, irfft, stft, istft, scatter, binomial, unfold, QR decomposition, view, and more.
Neural Network Operations: Added support for separable_conv, conv_transpose, adaptive_average_pool, adaptive_max_pool, RNN, LSTM, and GRU.
Control Flow Operations: Implemented cond, scan, associative_scan, map, switch, fori_loop, and vectorized_map.

Bug Fixes and Improvements

Backend Specific Improvements

PyTorch: Dynamic shapes support in export, device selection improvements, and bug fixes to the CuDNN based LSTM and GRU implementation.
JAX: Improved RNG handling in FlaxLayer and JaxLayer, variable jitting improvements, and direct JAX-to-ONNX export.
NumPy: Enabled masking support for the NumPy backend.

Other Improvements

Fixed multiple symbolic shape bugs across layers like Conv1DTranspose, IndexLookup, and TextVectorization.
Fixed activity regularizer normalization by batch size.
Improved Sequential error messages for incompatible layers.
Minimized memory usage issues in sparse_categorical_crossentropy.

New Contributors

We would like to thank our new contributors for making their first contribution to the Keras project:

@vaidik-gupta made their first contribution in #21939
@HyperPS made their first contribution in #21880
@calad0i made their first contribution in #21959
@KarSri7694 made their first contribution in #21963
@MarcosAsh made their first contribution in #21961
@orbin123 made their first contribution in #21935
@ayulockedin made their first contribution in #21985
@Shi-pra-19 made their first contribution in #21987
@mahi21tha made their first contribution in #21989
@PES2UG23CS205 made their first contribution in #21984
@samudraneel05 made their first contribution in #22017
@Junead04 made their first contribution in #21784
@nexeora made their first contribution in #22051
@bittoby made their first contribution in #22048
@0xManan made their first contribution in #22035
@sharpenteeth made their first contribution in #22079
@maitry63 made their first contribution in #22068
@Kh9705 made their first contribution in #22110
@timon0305 made their first contribution in #22112
@goyaladitya05 made their first contribution in #22131
@Sikandar1310291 made their first contribution in #22014
@haroon10725 made their first contribution in #22159
@andersendsa made their first contribution in #22155
@Rahuldrabit made their first contribution in #22146
@jerryxyj made their first contribution in #22178
@aaishwarymishra made their first contribution in #22173
@Sujanian1304 made their first contribution in #22236
@CityBoy-Claude made their first contribution in #22243
@rstar327 made their first contribution in #22252
@daehyun99 made their first contribution in #22289
@kysolvik made their first contribution in #22290
@ItzCobaltboy made their first contribution in #22158
@cpuguy96 made their first contribution in #22284
@0xRozier made their first contribution in #22218
@tanguyguyot made their first contribution in #22327
@AlanPonnachan made their first contribution in #21953
@shriramThakare3 made their first contribution in #22306
@Eruis2579 made their first contribution in #22350
@satheeshbhukya made their first contribution in #22388
@sam-shubham made their first contribution in #22265
@Passavee-Losripat made their first contribution in #22404
@ChiragSW made their first contribution in #22439
@rishi-sangare made their first contribution in #22407
@Caslyn made their first contribution in #22488
@Abineshabee made their first contribution in #22469
@dagecko made their first contribution in #22555

Full Changelog: v3.13.2...v3.14.0

Keras 3.14.0 v3.14.0 on Python PyPI