Gahow Wang 5b7dde1736 test: bf16 test reads f32-cast logits (forward now returns bf16) ...

The `keep bf16 logits` change made forward_batched return bf16 logits
in bf16 mode; the bf16 test's host read must cast to f32 first.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-16 14:29:24 +08:00

..

xtrain-autodiff

perf: keep bf16 logits (no persistent fp32 logits buffer)

2026-06-16 14:20:48 +08:00

xtrain-cuda

cuda: bf16 cuBLAS GemmEx (16BF in/out, fp32 accum) + cast kernels

2026-06-16 14:14:39 +08:00

xtrain-distributed

train: --bf16 flag (fp32-master AMP) + bf16 correctness test

2026-06-16 14:14:55 +08:00

xtrain-model

test: bf16 test reads f32-cast logits (forward now returns bf16)

2026-06-16 14:29:24 +08:00

xtrain-optim

perf: make xtrain-cuda a regular dep of xtrain-optim (GPU AdamW)

2026-06-15 16:53:52 +08:00

xtrain-tensor

autodiff: bf16 mixed-precision path (fp32 master via cast op)

2026-06-16 14:14:48 +08:00

xtrain-train

perf: keep bf16 logits (no persistent fp32 logits buffer)

2026-06-16 14:20:48 +08:00