diff --git a/FIX_LOWERING_FOR_CORE_ATEN_OPS.md b/FIX_LOWERING_FOR_CORE_ATEN_OPS.md
index 37c1fda22110..72f35ade2161 100644
--- a/FIX_LOWERING_FOR_CORE_ATEN_OPS.md
+++ b/FIX_LOWERING_FOR_CORE_ATEN_OPS.md
@@ -27,7 +27,7 @@ but produced a different result than torch eager mode.
 
 If the test uses 16-bit floats (float16, bfloat16); This is very likely
 that the tolerances that we give to `torch.allclose` to compare was to 
-strict. You can relax it a bit. Take a look at docs/fixing_core_aten_ops_log.md of one such example.
+strict. You can relax it a bit. Take a look at [this issue](https://github.com/pytorch/xla/issues/5934) of one such example.
 
 If the result torchxla produces is totally different than what torch produces, that means it's a bug in lowering code; and probably need
 more work. Feel free to tag more people (such as qihqi to look).
diff --git a/docs/fixing_core_aten_ops_log.md b/docs/fixing_core_aten_ops_log.md
deleted file mode 100644
index 80ec9ddf609e..000000000000
--- a/docs/fixing_core_aten_ops_log.md
+++ /dev/null
@@ -1,79 +0,0 @@
-
-# Issue beging worked https://github.com/pytorch/xla/issues/5902
-
-qihqi
-
-## 1. Uncomment and rerun the test
-
-```
-LD_LIBRARY_PATH=/mnt/hanq/miniconda3/envs/torch310/lib/:/usr/lib/x86_64-linux-gnu/ PJRT_DEVICE=CPU XLA_STABLEHLO_COMPILE=1 XLA_HLO_DEBUG=1 XLA_IR_DEBUG=1 pytest test/test_core_aten_ops.py -k test_aten_tan_1
-```
-
-output:
-```
-=========================== short test summary info ============================
-[torch_xla_diff:0.001] SUBFAIL test/test_core_aten_ops.py::AtenOpTest::test_aten_tan_1 - AssertionError: False is not true
-[stablehlo_diff: 0.001] SUBFAIL test/test_core_aten_ops.py::AtenOpTest::test_aten_tan_1 - AssertionError: False is not true
-================= 2 failed, 1 passed, 514 deselected in 5.51s ==================
-I0000 00:00:1700690393.569658 2513762 tfrt_cpu_pjrt_client.cc:352] TfrtCpuClient destroyed.
-(torch310) hanq@hanq-compile-2:/mnt/hanq/git/qihqi/pytorch/xla$
-```
-
-This means that the accuracy is not good.
-
-Break line here
-```
-(torch310) hanq@hanq-compile-2:/mnt/hanq/git/qihqi/pytorch/xla$ git diff
-diff --git a/test/test_core_aten_ops.py b/test/test_core_aten_ops.py
-index 46a18494d..ff055ee38 100644
---- a/test/test_core_aten_ops.py
-+++ b/test/test_core_aten_ops.py
-@@ -36,6 +36,7 @@ def run_export_and_compare(testcase, func, args, kwargs, atol=1e-3):
-                                      lambda x: x.to(device=device), kwargs)
-       res_xla = func(*args2, **kwargs2)
-       with testcase.subTest('torch_xla_diff:' + str(atol)):
-+        import pdb; pdb.set_trace()
-         diff_output(testcase, res, res_xla, atol)
-```
-
-Rerun, print out the difference:
-```
-(Pdb) p res - res_xla.cpu()
-tensor([[ 0.0000e+00,  0.0000e+00, -4.8828e-04,  0.0000e+00,  0.0000e+00,
-          0.0000e+00,  0.0000e+00,  6.1035e-05,  0.0000e+00,  0.0000e+00],
-        [-4.8828e-04,  0.0000e+00,  0.0000e+00,  9.7656e-04,  0.0000e+00,
-          1.2207e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
-        [ 0.0000e+00, -1.5259e-05,  0.0000e+00,  0.0000e+00,  0.0000e+00,
-          4.8828e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  1.2207e-04],
-        [ 0.0000e+00,  2.4414e-04,  0.0000e+00, -1.9531e-03,  0.0000e+00,
-          0.0000e+00,  0.0000e+00,  0.0000e+00, -3.0518e-05,  0.0000e+00],
-        [ 0.0000e+00, -4.8828e-04, -2.4414e-04,  0.0000e+00,  0.0000e+00,
-          0.0000e+00,  0.0000e+00, -6.1035e-05,  0.0000e+00,  0.0000e+00],
-        [ 0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00, -1.9531e-03,
-          0.0000e+00,  0.0000e+00,  1.9531e-03,  0.0000e+00,  0.0000e+00],
-        [ 0.0000e+00,  0.0000e+00, -1.9531e-03,  0.0000e+00,  0.0000e+00,
-          2.4414e-04,  9.7656e-04,  1.2207e-04,  0.0000e+00,  0.0000e+00],
-        [ 4.8828e-04,  0.0000e+00,  0.0000e+00, -7.8125e-03,  1.2207e-04,
-         -9.7656e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00],
-        [ 0.0000e+00,  1.5625e-02,  0.0000e+00,  0.0000e+00, -4.8828e-04,
-         -1.2207e-04,  0.0000e+00,  0.0000e+00, -4.8828e-04, -3.9062e-03],
-        [ 0.0000e+00, -1.2207e-04,  0.0000e+00,  0.0000e+00,  0.0000e+00,
-          0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00,  0.0000e+00]],
-       dtype=torch.float16)
-```
-The result looks good enough; This means that probably we are being too strict in
-test; setting a larger tolerance probably will work.
-```
-(Pdb) p torch.max(torch.abs(res - res_xla.cpu()))
-tensor(0.0156, dtype=torch.float16)
-
-```
-printing out the difference shows that roughly 0.01 atol with a slightly larger
-`rtol` probably work.
-
-```
-(Pdb) torch.allclose(res, res_xla.cpu(), atol=0.01, rtol=0.001)
-True
-```
-Now it's time to PR:
-https://github.com/pytorch/xla/pull/5915
\ No newline at end of file