jax

Fixing flax.nnx partial fine-tuning: align wrt filter and value_and_grad via nnx.DiffState

Jan. 13, 05:00

JAX vs NumPy Performance for Array Reshaping: Why Simple Ops Seem Slower and How JIT Helps

Dec. 22, 03:00

Why def _(): Still Runs in Python: Single-Underscore Names, Decorators, and JAX Control Flow

Dec. 17, 05:00

JAX Performance: Choosing Python for Loop vs jax.lax.scan, JIT Trade-offs and Unroll Tuning

Dec. 10, 11:00

JAX precision pitfalls: Python lists vs jnp.array and dtype consistency in training loops

Dec. 9, 13:00

Reconstructing Images from Patches in JAX: Fast reshape/transpose, no Python loops

Dec. 3, 07:00

JAX vmap vs Broadcasting: Clear, Safe Vectorization and Batching (with jnp.histogram)

Oct. 21, 02:00

JAX vmap numeric drift across batch sizes on GPU float32: causes, reproduction, fixes

Oct. 18, 01:00

JAX jit and Object State: Fixing TracerArrayConversionError from In-method Attribute Mutation

Oct. 17, 08:00

JAX JIT, scan, and pytrees: avoiding recompilation with lambda captures and stable cache keys

Oct. 3, 07:00

JAX lax.scan parameterization: closure (lambda/def) vs carry — performance parity and clarity

Oct. 2, 21:00

Hyperparameter Scheduling in JAX/Flax with Optax: Use Optimizer Step to Drive Schedules (Flax 0.11)

Sep. 21, 17:00

1

By continuing to use this website, you agree to our Cookie Policy and Privacy Policy.

User Agreement Cookie Policy Privacy Policy About Contact

© 2026 Python Troubles