Bump tolerances for per_sample_grads tutorial

Author: mikaylagawarecki

intermediate_source/per_sample_grads.py

@@ -169,7 +169,10 @@ def compute_loss(params, buffers, sample, target):
 # results of hand processing each one individually:
 
 for per_sample_grad, ft_per_sample_grad in zip(per_sample_grads, ft_per_sample_grads.values()):
-    assert torch.allclose(per_sample_grad, ft_per_sample_grad, atol=3e-3, rtol=1e-5)
+    print((per_sample_grad - ft_per_sample_grad).max(), (per_sample_grad - ft_per_sample_grad).mean())
+    # assert torch.allclose(per_sample_grad, ft_per_sample_grad, atol=1.5e-2, rtol=1e-5)
+
+assert False
 
 ######################################################################
 # A quick note: there are limitations around what types of functions can be