Adding regression benchmark for the transformers SHA 0b568291d722aa5e39cd4a8fa05c03200dd280ab

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float16
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: ${infer_task:${model}, ${hub_kwargs.revision}}
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/.config/hydra.yaml

@@ -0,0 +1,174 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,16
+      backend.torch_dtype: float16,float32
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=1
+    - backend.torch_dtype=float16
+  job:
+    name: main
+    chdir: true
+    override_dirname: backend.torch_dtype=float16,benchmark.input_shapes.batch_size=1
+    id: '0'
+    num: 0
+    config_name: llama2_1gpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/.config/overrides.yaml

@@ -0,0 +1,2 @@
+- benchmark.input_shapes.batch_size=1
+- backend.torch_dtype=float16

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float16
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s),generate.latency(s),generate.throughput(tokens/s)
+0,15651.96288,0.0312,32.1,5.91,33.8

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/0/main.log

@@ -0,0 +1,26 @@
+[2023-08-17 08:23:47,220][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:23:47,221][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:23:48,836][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-17 08:23:48,837][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:23:48,837][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:23:48,837][utils][INFO] - device_ids to check: {0}
+[2023-08-17 08:23:48,908][utils][INFO] - os.getpid() 601426
+[2023-08-17 08:23:48,909][utils][INFO] - pids_on_device_id set()
+[2023-08-17 08:23:48,909][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:23:48,922][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:23:48,923][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-17 08:25:02,794][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:25:02,796][inference][INFO] - Running inference benchmark
+[2023-08-17 08:25:10,806][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:25:12,068][memory_tracker][INFO] - Peak memory usage: 15651.96288 MB
+[2023-08-17 08:25:12,069][inference][INFO] - 	+ Forward pass peak memory: 15651.96288 (MB)
+[2023-08-17 08:25:12,069][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:25:12,383][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:25:27,654][inference][INFO] - 	+ Forward pass latency: 3.12e-02 (s)
+[2023-08-17 08:25:27,655][inference][INFO] - 	+ Forward pass throughput: 32.10 (samples/s)
+[2023-08-17 08:25:27,655][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-17 08:25:34,245][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-17 08:25:51,980][inference][INFO] - 	+ Generation pass latency: 5.91e+00 (s)
+[2023-08-17 08:25:51,982][inference][INFO] - 	+ Generation pass throughput: 33.80 (tokens/s)
+[2023-08-17 08:25:51,982][inference][INFO] - Saving inference results
+[2023-08-17 08:25:51,991][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float32
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: ${infer_task:${model}, ${hub_kwargs.revision}}
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/.config/hydra.yaml

@@ -0,0 +1,174 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,16
+      backend.torch_dtype: float16,float32
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=1
+    - backend.torch_dtype=float32
+  job:
+    name: main
+    chdir: true
+    override_dirname: backend.torch_dtype=float32,benchmark.input_shapes.batch_size=1
+    id: '1'
+    num: 1
+    config_name: llama2_1gpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/.config/overrides.yaml

@@ -0,0 +1,2 @@
+- benchmark.input_shapes.batch_size=1
+- backend.torch_dtype=float32

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float32
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s),generate.latency(s),generate.throughput(tokens/s)
+0,29317.005311999998,0.0641,15.6,5.55,36.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/1/main.log

@@ -0,0 +1,26 @@
+[2023-08-17 08:25:52,563][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:25:52,564][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:25:53,065][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-17 08:25:53,066][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:25:53,066][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:25:53,066][utils][INFO] - device_ids to check: {0}
+[2023-08-17 08:25:53,162][utils][INFO] - os.getpid() 601426
+[2023-08-17 08:25:53,162][utils][INFO] - pids_on_device_id {601426}
+[2023-08-17 08:25:53,162][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:25:53,184][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:25:53,185][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-17 08:26:10,010][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:26:10,011][inference][INFO] - Running inference benchmark
+[2023-08-17 08:26:17,940][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:26:18,016][memory_tracker][INFO] - Peak memory usage: 29317.005311999998 MB
+[2023-08-17 08:26:18,016][inference][INFO] - 	+ Forward pass peak memory: 29317.005311999998 (MB)
+[2023-08-17 08:26:18,017][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:26:20,149][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:27:10,046][inference][INFO] - 	+ Forward pass latency: 6.41e-02 (s)
+[2023-08-17 08:27:10,046][inference][INFO] - 	+ Forward pass throughput: 15.60 (samples/s)
+[2023-08-17 08:27:10,047][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-17 08:27:15,624][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-17 08:27:32,282][inference][INFO] - 	+ Generation pass latency: 5.55e+00 (s)
+[2023-08-17 08:27:32,283][inference][INFO] - 	+ Generation pass throughput: 36.00 (tokens/s)
+[2023-08-17 08:27:32,283][inference][INFO] - Saving inference results
+[2023-08-17 08:27:32,290][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float16
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: ${infer_task:${model}, ${hub_kwargs.revision}}
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/.config/hydra.yaml

@@ -0,0 +1,174 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,16
+      backend.torch_dtype: float16,float32
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=16
+    - backend.torch_dtype=float16
+  job:
+    name: main
+    chdir: true
+    override_dirname: backend.torch_dtype=float16,benchmark.input_shapes.batch_size=16
+    id: '2'
+    num: 2
+    config_name: llama2_1gpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/.config/overrides.yaml

@@ -0,0 +1,2 @@
+- benchmark.input_shapes.batch_size=16
+- backend.torch_dtype=float16

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float16
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s),generate.latency(s),generate.throughput(tokens/s)
+0,18841.731072,0.0969,165.0,6.24,513.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/2/main.log

@@ -0,0 +1,26 @@
+[2023-08-17 08:27:32,864][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:27:32,864][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:27:33,719][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-17 08:27:33,719][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:27:33,719][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:27:33,719][utils][INFO] - device_ids to check: {0}
+[2023-08-17 08:27:33,816][utils][INFO] - os.getpid() 601426
+[2023-08-17 08:27:33,817][utils][INFO] - pids_on_device_id {601426}
+[2023-08-17 08:27:33,817][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:27:33,838][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:27:33,839][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-17 08:27:44,695][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:27:44,697][inference][INFO] - Running inference benchmark
+[2023-08-17 08:27:52,490][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:27:52,601][memory_tracker][INFO] - Peak memory usage: 18841.731072 MB
+[2023-08-17 08:27:52,601][inference][INFO] - 	+ Forward pass peak memory: 18841.731072 (MB)
+[2023-08-17 08:27:52,602][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:27:55,210][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:28:35,765][inference][INFO] - 	+ Forward pass latency: 9.69e-02 (s)
+[2023-08-17 08:28:35,766][inference][INFO] - 	+ Forward pass throughput: 165.00 (samples/s)
+[2023-08-17 08:28:35,767][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-17 08:28:42,959][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-17 08:29:01,691][inference][INFO] - 	+ Generation pass latency: 6.24e+00 (s)
+[2023-08-17 08:29:01,692][inference][INFO] - 	+ Generation pass throughput: 513.00 (tokens/s)
+[2023-08-17 08:29:01,692][inference][INFO] - Saving inference results
+[2023-08-17 08:29:01,699][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float32
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: ${infer_task:${model}, ${hub_kwargs.revision}}
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/.config/hydra.yaml

@@ -0,0 +1,174 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,16
+      backend.torch_dtype: float16,float32
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=16
+    - backend.torch_dtype=float32
+  job:
+    name: main
+    chdir: true
+    override_dirname: backend.torch_dtype=float32,benchmark.input_shapes.batch_size=16
+    id: '3'
+    num: 3
+    config_name: llama2_1gpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/.config/overrides.yaml

@@ -0,0 +1,2 @@
+- benchmark.input_shapes.batch_size=16
+- backend.torch_dtype=float32

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: float32
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 200
+experiment_name: llama_1gpu_inference
+model: daryl149/llama-2-7b-chat-hf
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s),generate.latency(s),generate.throughput(tokens/s)
+0,34801.057792,0.683,23.4,12.9,248.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/llama_1gpu_inference/3/main.log

@@ -0,0 +1,26 @@
+[2023-08-17 08:29:02,345][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:29:02,346][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:29:02,843][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-17 08:29:02,843][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:29:02,843][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:29:02,844][utils][INFO] - device_ids to check: {0}
+[2023-08-17 08:29:02,942][utils][INFO] - os.getpid() 601426
+[2023-08-17 08:29:02,942][utils][INFO] - pids_on_device_id {601426}
+[2023-08-17 08:29:02,943][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:29:02,964][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:29:02,965][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-17 08:29:19,467][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:29:19,469][inference][INFO] - Running inference benchmark
+[2023-08-17 08:29:27,471][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:29:28,181][memory_tracker][INFO] - Peak memory usage: 34801.057792 MB
+[2023-08-17 08:29:28,181][inference][INFO] - 	+ Forward pass peak memory: 34801.057792 (MB)
+[2023-08-17 08:29:28,197][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:29:53,507][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:30:51,092][inference][INFO] - 	+ Forward pass latency: 6.83e-01 (s)
+[2023-08-17 08:30:51,093][inference][INFO] - 	+ Forward pass throughput: 23.40 (samples/s)
+[2023-08-17 08:30:51,094][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-17 08:31:04,233][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-17 08:31:30,058][inference][INFO] - 	+ Generation pass latency: 1.29e+01 (s)
+[2023-08-17 08:31:30,061][inference][INFO] - 	+ Generation pass throughput: 248.00 (tokens/s)
+[2023-08-17 08:31:30,061][inference][INFO] - Saving inference results
+[2023-08-17 08:31:30,068][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_bert_inference
+model: hf-internal-testing/tiny-random-bert
+device: cpu
+task: text-classification
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/.config/hydra.yaml

@@ -0,0 +1,172 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,4
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=1
+  job:
+    name: main
+    chdir: true
+    override_dirname: benchmark.input_shapes.batch_size=1
+    id: '0'
+    num: 0
+    config_name: bert_cpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/.config/overrides.yaml

@@ -0,0 +1 @@
+- benchmark.input_shapes.batch_size=1

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_bert_inference
+model: hf-internal-testing/tiny-random-bert
+device: cpu
+task: text-classification
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s)
+0,466.591744,0.00311,322.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/0/main.log

@@ -0,0 +1,20 @@
+[2023-08-17 08:22:40,089][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:22:40,090][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:22:41,460][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForSequenceClassification for task text-classification and model_type bert
+[2023-08-17 08:22:41,460][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:22:41,461][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:22:41,461][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:22:41,461][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:22:41,461][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: None on device: cpu
+[2023-08-17 08:22:42,093][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:22:42,094][inference][INFO] - Running inference benchmark
+[2023-08-17 08:22:42,218][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-17 08:22:42,219][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:22:42,278][inference][INFO] - 	+ Forward pass peak memory: 466.591744 (MB)
+[2023-08-17 08:22:42,279][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-17 08:22:42,281][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:22:42,322][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:22:57,508][inference][INFO] - 	+ Forward pass latency: 3.11e-03 (s)
+[2023-08-17 08:22:57,511][inference][INFO] - 	+ Forward pass throughput: 322.00 (samples/s)
+[2023-08-17 08:22:57,511][inference][INFO] - Saving inference results
+[2023-08-17 08:22:57,528][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 4
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_bert_inference
+model: hf-internal-testing/tiny-random-bert
+device: cpu
+task: text-classification
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/.config/hydra.yaml

@@ -0,0 +1,172 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params:
+      benchmark.input_shapes.batch_size: 1,4
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task:
+    - benchmark.input_shapes.batch_size=4
+  job:
+    name: main
+    chdir: true
+    override_dirname: benchmark.input_shapes.batch_size=4
+    id: '1'
+    num: 1
+    config_name: bert_cpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/.config/overrides.yaml

@@ -0,0 +1 @@
+- benchmark.input_shapes.batch_size=4

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: true
+  warmup_runs: 10
+  benchmark_duration: 15
+  input_shapes:
+    batch_size: 4
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_bert_inference
+model: hf-internal-testing/tiny-random-bert
+device: cpu
+task: text-classification
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s)
+0,467.91475199999996,0.00343,1170.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_bert_inference/1/main.log

@@ -0,0 +1,20 @@
+[2023-08-17 08:22:57,918][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:22:57,919][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:22:58,394][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForSequenceClassification for task text-classification and model_type bert
+[2023-08-17 08:22:58,395][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:22:58,395][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:22:58,395][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:22:58,395][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:22:58,395][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: None on device: cpu
+[2023-08-17 08:22:58,532][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:22:58,532][inference][INFO] - Running inference benchmark
+[2023-08-17 08:22:58,658][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-17 08:22:58,660][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-17 08:22:58,704][inference][INFO] - 	+ Forward pass peak memory: 467.91475199999996 (MB)
+[2023-08-17 08:22:58,705][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-17 08:22:58,706][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:22:58,742][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:23:13,906][inference][INFO] - 	+ Forward pass latency: 3.43e-03 (s)
+[2023-08-17 08:23:13,908][inference][INFO] - 	+ Forward pass throughput: 1170.00 (samples/s)
+[2023-08-17 08:23:13,909][inference][INFO] - Saving inference results
+[2023-08-17 08:23:13,920][backend][INFO] - Cleaning backend

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/.config/config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: ${is_inference:${benchmark.name}}
+  eval_mode: ${is_inference:${benchmark.name}}
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: false
+  warmup_runs: 10
+  benchmark_duration: 10
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_gpt2_inference
+model: hf-internal-testing/tiny-random-gpt2
+device: cpu
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/.config/hydra.yaml

@@ -0,0 +1,170 @@
+hydra:
+  run:
+    dir: runs/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+  sweep:
+    dir: sweeps/${oc.env:COMMIT_DATE_GMT}_${oc.env:COMMIT_SHA}/${experiment_name}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(purple)sHYDRA%(reset)s] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+      colorlog:
+        (): colorlog.ColoredFormatter
+        format: '[%(cyan)s%(asctime)s%(reset)s][%(blue)s%(name)s%(reset)s][%(log_color)s%(levelname)s%(reset)s]
+          - %(message)s'
+        log_colors:
+          DEBUG: purple
+          INFO: green
+          WARNING: yellow
+          ERROR: red
+          CRITICAL: red
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: colorlog
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: MULTIRUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=MULTIRUN
+    task: []
+  job:
+    name: main
+    chdir: true
+    override_dirname: ''
+    id: '0'
+    num: 0
+    config_name: gpt2_cpu_inference
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/user/transformers-regression
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: optimum_benchmark
+      schema: pkg
+      provider: main
+    - path: hydra_plugins.hydra_colorlog.conf
+      schema: pkg
+      provider: hydra-colorlog
+    - path: /home/user/transformers-regression/configs
+      schema: file
+      provider: command-line
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/user/transformers-regression/sweeps/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0
+    choices:
+      benchmark: inference
+      backend: pytorch
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: colorlog
+      hydra/hydra_logging: colorlog
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/.config/overrides.yaml

@@ -0,0 +1 @@
+[]

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/hydra_config.yaml

@@ -0,0 +1,66 @@
+backend:
+  name: pytorch
+  version: 2.0.1+cu117
+  _target_: optimum_benchmark.backends.pytorch.PyTorchBackend
+  inter_op_num_threads: null
+  intra_op_num_threads: null
+  initial_isolation_check: true
+  continous_isolation_check: true
+  delete_cache: false
+  no_weights: false
+  torch_dtype: null
+  device_map: null
+  load_in_8bit: false
+  load_in_4bit: false
+  bettertransformer: false
+  torch_compile: false
+  torch_compile_config:
+    fullgraph: false
+    dynamic: false
+    backend: inductor
+    mode: null
+    options: null
+    disable: false
+  amp_autocast: false
+  amp_dtype: null
+  disable_grad: true
+  eval_mode: true
+benchmark:
+  name: inference
+  _target_: optimum_benchmark.benchmarks.inference.InferenceBenchmark
+  seed: 42
+  memory: false
+  warmup_runs: 10
+  benchmark_duration: 10
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 1
+    width: 64
+    height: 64
+    num_channels: 3
+    point_batch_size: 3
+    nb_points_per_image: 2
+    feature_size: 80
+    nb_max_frames: 3000
+    audio_sequence_length: 16000
+  new_tokens: 100
+experiment_name: pytorch_gpt2_inference
+model: hf-internal-testing/tiny-random-gpt2
+device: cpu
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.1
+  transformers_version: 4.32.0.dev0
+  accelerate_version: 0.21.0
+  diffusers_version: null
+  python_version: 3.10.12
+  system: Linux
+  cpu: ' Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz'
+  cpu_count: 96
+  cpu_ram_mb: 1204539.797504

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.latency(s),forward.throughput(samples/s),generate.latency(s),generate.throughput(tokens/s)
+0,0.00351,285.0,0.523,191.0

raw_results/2023-08-16_10:49:29_0b568291d722aa5e39cd4a8fa05c03200dd280ab/pytorch_gpt2_inference/0/main.log

@@ -0,0 +1,20 @@
+[2023-08-17 08:23:18,740][benchmark][INFO] - Configuring inference benchmark
+[2023-08-17 08:23:18,741][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-17 08:23:20,207][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type gpt2
+[2023-08-17 08:23:20,208][backend][INFO] - Configuring pytorch backend
+[2023-08-17 08:23:20,208][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-17 08:23:20,208][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-17 08:23:20,208][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-17 08:23:20,209][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: None on device: cpu
+[2023-08-17 08:23:20,868][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-17 08:23:20,868][inference][INFO] - Running inference benchmark
+[2023-08-17 08:23:21,077][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-17 08:23:21,115][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-17 08:23:31,218][inference][INFO] - 	+ Forward pass latency: 3.51e-03 (s)
+[2023-08-17 08:23:31,220][inference][INFO] - 	+ Forward pass throughput: 285.00 (samples/s)
+[2023-08-17 08:23:31,221][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-17 08:23:31,717][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-17 08:23:42,178][inference][INFO] - 	+ Generation pass latency: 5.23e-01 (s)
+[2023-08-17 08:23:42,179][inference][INFO] - 	+ Generation pass throughput: 191.00 (tokens/s)
+[2023-08-17 08:23:42,179][inference][INFO] - Saving inference results
+[2023-08-17 08:23:42,192][backend][INFO] - Cleaning backend