tinygrad/tinygrad
chenyu 97b05f567e
revert the .detach() in layernorm (#4904)
* revert the .detach() in layernorm

it's only correct in LayerNorm where input is the data, and not correct in GroupNorm and InstanceNorm that reused layernorm.
Added backward tests for weights, bias and input for these norms.

* bigger atol for llvm

* relax backward more
2024-06-10 18:02:05 -04:00
..
codegen process replay in all of CI (#4884) 2024-06-10 14:49:29 -04:00
engine retire hsa (#4885) 2024-06-09 11:33:03 +03:00
nn hotfix: import datasets in nn init 2024-06-10 11:33:50 +02:00
renderer replace set with dedup (#4901) 2024-06-10 18:20:38 +02:00
runtime replace set with dedup (#4901) 2024-06-10 18:20:38 +02:00
shape expand ShapeTracker.invert a bit (#4864) 2024-06-07 14:26:02 -04:00
__init__.py add mnist tutorial 2024-04-25 16:08:32 +08:00
device.py retire hsa (#4885) 2024-06-09 11:33:03 +03:00
dtype.py move acc_dt out of lazy (#4382) 2024-05-06 07:41:25 -07:00
function.py CMPEQ -> CMPNE and make it safe to pad (#4818) 2024-06-03 18:02:15 -04:00
helpers.py tqdm replacement (#4846) 2024-06-09 23:46:03 +02:00
lazy.py cleanup lazy const fold for binary (#4900) 2024-06-10 10:46:58 -04:00
multi.py move image into tensor.py. delete features (#4603) 2024-05-15 10:50:25 -07:00
ops.py unbind Variable pre LazyOp (#4873) 2024-06-08 08:16:38 -04:00
tensor.py revert the .detach() in layernorm (#4904) 2024-06-10 18:02:05 -04:00