Fix Issues With Benchmark Script

[fabianlim]

The benchscript can still be improved in couple of ways:

Edit: We need to Triage these

• [x] progressively write out the results as each experiment completes.
• [x] also measure the GPU mem usage, and update the device hardware (e.g., A100). See #8
• ~switch from accelerate.yaml to [command line args]~(https://huggingface.co/docs/accelerate/en/package_reference/cli) so we have one yaml less to manage.
• [x] print out the command line, accelerate args to experiment.save_dir so there is a record; there is already one in the stderr, but maybe something more tidy like in a YAML
• ~grep out the FMS accelerate printout from stdout to confirm the correct plugin is activated.~
• [x] making sure the results directory is cleared (but maybe this can be done in run_benchmarks.sh. See #13
• [x] put more logs to stdout to inform what is happening in benchmark script (e.g., directory created, results written, etc)

[fabianlim]

I think most of the things written here are done so closing this