[quant] Quantized flip dispatch

[z-a-f]

Stack from ghstack:

• [quant][WIP] Fixing the CM #47220 [quant][WIP] Fixing the CM
• [quant] Statically quantized LSTM #46229 [quant] Statically quantized LSTM
• [quant] Quantized flip dispatch #46235 [quant] Quantized flip dispatch

Differential Revision: D24689161

[codecov]

Codecov Report

Merging #46235 into gh/z-a-f/75/base will increase coverage by 19.91%.
The diff coverage is n/a.

@@                  Coverage Diff                  @@
##           gh/z-a-f/75/base   #46235       +/-   ##
=====================================================
+ Coverage             48.42%   68.34%   +19.91%     
=====================================================
  Files                  2735      410     -2325     
  Lines                252151    53804   -198347     
=====================================================
- Hits                 122110    36770    -85340     
+ Misses               130041    17034   -113007     

Impacted Files	Coverage Δ
torch/nn/utils/memory_format.py	25.00% <0.00%> (-75.00%)	⬇️
torch/backends/cudnn/rnn.py	0.00% <0.00%> (-54.55%)	⬇️
torch/multiprocessing/_atfork.py	61.53% <0.00%> (-30.77%)	⬇️
torch/cuda/random.py	22.97% <0.00%> (-28.38%)	⬇️
torch/backends/cudnn/__init__.py	59.74% <0.00%> (-27.28%)	⬇️
...enchmark/utils/valgrind_wrapper/timer_interface.py	25.78% <0.00%> (-23.32%)	⬇️
torch/cuda/__init__.py	55.17% <0.00%> (-22.99%)	⬇️
torch/__init__.py	65.25% <0.00%> (-16.61%)	⬇️
torch/futures/__init__.py	84.61% <0.00%> (-11.54%)	⬇️
torch/testing/_internal/common_nn.py	85.53% <0.00%> (-10.77%)	⬇️
... and 2549 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update da26858...3fc2721. Read the comment docs.

[dr-ci]

💊 CI failures summary and remediations

As of commit 3fc2721 (more details on the Dr. CI page):

---

• 1/1 failures introduced in this PR

---

1 job timed out:

• pytorch_windows_vs2019_py36_cuda10.1_test2

---

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group.

See how this bot performed.

This comment has been revised 7 times.

[jerryzh168]

if this pattern becomes common, then we probably want to create a new macro that include both AT_DISPATCH_ALL_TYPES_AND_COMPLEX_AND and AT_DISPATCH_QINT_AND_SUB_BYTE_TYPES

[z-a-f]

I agree. Alternatively, I can copy-paste this code into the quantized folder, and register the quantized kernel there.

[facebook-github-bot]

@z-a-f merged this pull request in 31ebac3.