audit_knnshapley_bert.yml

$SUITE_VARS$:

  _python_pre: "export PYTHONPATH={{ exp_code_dir }}/utils && . {{ exp_code_dir }}/.venv/bin/activate"
  _python_path: "[% my_run._python_pre %] && {{ exp_code_dir }}/.venv/bin/python"
  cmd_mpspdz: "[% my_run._python_path %] -m  python_utils.scripts.experiment_runner --player-number <PLAYER-ID>"

  n_parties: 3

  $INCLUDE_VARS$: dataset.yml

  commit_output: False
  convert_ring_bits: 31
  convert_n_threads: 36
  convert_chunk_size: 150000

  sleep_time: 30
  _runtime_info:
    budget: 10000

  consistency_args:
    abs_path_to_code_dir: "{{ exp_consistency_dir }}"
    hosts_file: "{{ exp_consistency_hosts_file }}"
    pc: kzg
    pp_args: "[% my_run.dataset_info[my_run.mpc.script_args.dataset].max_input_size %]"
    gen_pp: True
    type: $FACTOR$
    use_split: True

  mpc:
    player_0_hostname: "[% exp_host_lst | json_query('[?host_type==`large_memory`].private_dns_name') | default(['<UNDEFINED-DNS>'], true) | first %]"
    abs_path_to_code_dir: "{{ exp_code_dir }}"
    player_count: "[% my_run.n_parties %]"

    protocol_setup: $FACTOR$
    compiler_args: $FACTOR$ #'100000000'] # budget was '1000000'1million # also look at B 5 -> cannot be used together with R
    domain: $FACTOR$
    custom_prime: $FACTOR$

    script_args:

      consistency_check: "[% my_run.consistency_args.type %]"
      sha3_approx_factor: "[% my_run.dataset_info[my_run.mpc.script_args.dataset].sha3_approx_factor_full | default(1) %]" # we prove full dataset

      n_samples: 1
      n_input_parties: "[% my_run.n_parties %]"

      n_s_bert: 2500
#        $FACTOR$: [ 5000, 2500, 1000 ]

      dataset:
        $FACTOR$: [ glue-qnli ]

      debug: False  # for a final benchmarking run, can also use `debug: False`, this should then only release no secret info
      emulate: False

      # in a network with latency, a high batch size for ml.py is key to a good performance (60k raised insufficient memory error)
      batch_size: 2 #1024 #256 # was 128 before   # 60000 -> insufficient memory
      n_batches: 1
#      n_threads: 36 #128 #64 # TODO before was 32 -> see if this helps (128 raises too many files runtime error)

#      audit_trigger_idx: 0 # select single audit trigger (from all)

    stage: # "compile" in run and then in next run "run"
      $FACTOR$: [ compile, run ] # ensure that this is last factor


audit_sample_knnshapley:
  n_repetitions: 1
  host_types:
    large_memory:
      n: 3
      check_status: True
      init_roles:
        - setup-base
        - setup-consistency
        - setup-spdz-secrets
      $CMD$:
        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 0) %]"
        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 1) %]"
        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 2) %]"

  base_experiment:
    consistency_args: # need this so that this comes before compile, run factors, ie mpc dict should always be last
      type:
        $FACTOR$: [ sha3s, cerebro, pc ]
#        $FACTOR$: [ pc ]
    mpc:
      script_name: audit_sample_knnshapley

    network_type: lan
    mpc_type: $FACTOR$


  factor_levels:

#    - mpc_type: mal
#      mpc: # should convert
#        protocol_setup: "sy-rep-ring-party"
#        domain: ring_split
#        compiler_args: [ '-R', '64', '-C', '--budget', "[% my_run._runtime_info.budget %]", '-Z', '3', '-D' ]
#        custom_prime: null

    - mpc_type: sh
      mpc: # should convert
        protocol_setup: "semi_honest_3"
        domain: ring_split
        compiler_args: [ '-R', '64', '-C', '--budget', "[% my_run._runtime_info.budget %]", '-Z', '3', '-D' ]
        custom_prime: null


#audit_sample_knnshapley_mal:
#  n_repetitions: 1
#  host_types:
#    large_memory:
#      n: 3
#      check_status: True
#      init_roles:
#        - setup-base
#        - setup-consistency
#        - setup-spdz-secrets
#      $CMD$:
#        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 0) %]"
#        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 1) %]"
#        - main: "[% my_run.cmd_mpspdz | replace('<PLAYER-ID>', 2) %]"
#
#  base_experiment:
#    consistency_args: # need this so that this comes before compile, run factors, ie mpc dict should always be last
#      type:
#        $FACTOR$: [ sha3s, cerebro, pc ]
#    #        $FACTOR$: [ pc ]
#    mpc:
#      script_name: audit_sample_knnshapley
#
#    network_type: lan
#    mpc_type: $FACTOR$
#
#
#  factor_levels:
#
#        - mpc_type: mal
#          mpc: # should convert
#            protocol_setup: "sy-rep-ring-party"
#            domain: ring_split
#            compiler_args: [ '-R', '64', '-C', '--budget', "[% my_run._runtime_info.budget %]", '-Z', '3', '-D' ]
#            custom_prime: null



$ETL$: {}