Enable enzyme for Turing example

[yebai]

No description provided.

[github-actions]

Mooncake.jl documentation for PR #788 is available at:
https://chalk-lab.github.io/Mooncake.jl/previews/PR788/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[yebai]

@penelopeysm, can you help take a look at why Enzyme is broken for the Turing broadcasting example? Let's figure out whether this is an incorrect use of the Enzyme API or a genuine limitation of Enzyme.

[github-actions]

Performance Ratio:
Ratio of time to compute gradient and time to compute function.
Warning: results are very approximate! See here for more context.

┌────────────────────────────┬──────────┬──────────┬─────────────┬─────────┬─────────────┬────────┐
│                      Label │   Primal │ Mooncake │ MooncakeFwd │  Zygote │ ReverseDiff │ Enzyme │
│                     String │   String │   String │      String │  String │      String │ String │
├────────────────────────────┼──────────┼──────────┼─────────────┼─────────┼─────────────┼────────┤
│                   sum_1000 │ 100.0 ns │      1.8 │         1.9 │     1.1 │        5.61 │   8.31 │
│                  _sum_1000 │ 941.0 ns │     6.49 │        1.01 │  1420.0 │        32.4 │   1.09 │
│               sum_sin_1000 │  6.57 μs │     2.58 │        1.38 │    1.71 │        10.6 │   2.19 │
│              _sum_sin_1000 │  5.34 μs │     2.97 │        2.14 │   265.0 │        12.9 │   2.43 │
│                   kron_sum │ 333.0 μs │     39.5 │        2.61 │    4.74 │       191.0 │    8.1 │
│              kron_view_sum │ 340.0 μs │     40.0 │         3.3 │    10.7 │       205.0 │   11.8 │
│      naive_map_sin_cos_exp │  2.25 μs │     2.25 │        1.33 │ missing │        7.17 │   2.26 │
│            map_sin_cos_exp │  2.12 μs │     2.66 │        1.44 │    1.49 │        6.06 │   2.36 │
│      broadcast_sin_cos_exp │  2.25 μs │     2.38 │        1.38 │    2.29 │        1.47 │   2.24 │
│                 simple_mlp │ 163.0 μs │     7.02 │        2.97 │     1.8 │        12.6 │   3.29 │
│                     gp_lml │ 190.0 μs │     10.5 │        2.37 │    4.43 │     missing │   5.59 │
│ turing_broadcast_benchmark │  1.95 ms │     4.34 │        3.12 │ missing │        24.2 │   2.03 │
│         large_single_block │ 380.0 ns │     4.53 │        2.03 │  4540.0 │        32.4 │   2.24 │
└────────────────────────────┴──────────┴──────────┴─────────────┴─────────┴─────────────┴────────┘

[sunxd3]

I see no issue.

(Maybe update the description to "enable Enzyme benchmark", should_run_benchmark is not really unused.)

[sunxd3]

maybe let CI run and see if new benchmarks are produced

[yebai]

update the description to "enable Enzyme benchmark", should_run_benchmark is not really unused.)

Yes, there is a generic fallback for should_run_benchmark, so this PR doesn't remove it.