Pull request #4: S3 cci updateMerge in MLFET/maelstrom-radiation from s3_cci_update to main
* commit '9595855ac9700b1b72b7708a1480d8fd8cb2b8d5':
update version to 0.7
update urls to use new s3 bucket on cci
Pull request #3: JubeMerge in MLFET/maelstrom-radiation from jube to main
* commit 'd1d28b9d337e58354653877b214dae84ef184e86':
fix to cscratch location
Inference jube
jube updates for E4
jube fixes
env setup
jube ready?
Pull request #2: Benchmark metricsMerge in MLFET/maelstrom-radiation from metrics to main
* commit '85c7f016a3e9b9b16aeb1bb2f3fc3cda3567ad42':
Benchmark metrics
Tempory remove parallel processing of netcdfThis was causing HDF5 errors. Further debugging needed before
this can be reintroduced. No large degradations in performance
were noted with parallel=False
Pull request #1: DevelMerge in MLFET/maelstrom-radiation from devel to main
* commit 'fc0443da89ef2af8c30e1b83e8cb81af55997b80':
Refactor benchmarks, add hooks
Updates to models & data
Inference in benchmark
Command line calling for benchmark
New tier definitions
Plotting and attention
Visualising models and other updates
Fix cache/shuffle order
Progress in second benchmarking
Benchmarking and demo notebooks.Plus small quality of life changes in radiation and _tf.
1. gather_fluxes in radiation
2. subsetting sw radiation for only "light side values"