Adding regression benchmark for the transformers SHA 2b22cde71e549e194a3b7046b7595030382382e9

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: 20
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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)
+0,16195.125247999998,0.0403,24.8,7.77,25.7

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/0/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:07:47,592][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:07:47,594][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:07:47,874][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:07:47,875][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:07:47,875][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:07:48,018][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:07:48,034][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:07:48,036][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-13 19:08:57,160][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:08:57,162][inference][INFO] - Running inference benchmark
+[2023-08-13 19:09:05,918][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:09:07,187][memory_tracker][INFO] - Peak memory usage: 16195.125247999998 MB
+[2023-08-13 19:09:07,187][inference][INFO] - 	+ Forward pass peak memory: 16195.125247999998 (MB)
+[2023-08-13 19:09:07,188][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:09:07,592][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:09:27,779][inference][INFO] - 	+ Forward pass latency: 4.03e-02 (s)
+[2023-08-13 19:09:27,779][inference][INFO] - 	+ Forward pass throughput: 24.80 (samples/s)
+[2023-08-13 19:09:27,780][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:09:36,265][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:09:59,586][inference][INFO] - 	+ Generation pass latency: 7.77e+00 (s)
+[2023-08-13 19:09:59,589][inference][INFO] - 	+ Generation pass throughput: 25.70 (tokens/s)
+[2023-08-13 19:09:59,589][inference][INFO] - Saving inference results
+[2023-08-13 19:09:59,598][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: 20
+  input_shapes:
+    batch_size: 1
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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)
+0,30317.346815999997,0.0644,15.5,7.37,27.1

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/1/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:10:00,072][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:10:00,074][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:10:00,266][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:10:00,266][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:10:00,266][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:10:00,367][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:10:00,406][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:10:00,407][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-13 19:10:17,670][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:10:17,672][inference][INFO] - Running inference benchmark
+[2023-08-13 19:10:26,262][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:10:26,338][memory_tracker][INFO] - Peak memory usage: 30317.346815999997 MB
+[2023-08-13 19:10:26,338][inference][INFO] - 	+ Forward pass peak memory: 30317.346815999997 (MB)
+[2023-08-13 19:10:26,339][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:10:28,471][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:11:34,720][inference][INFO] - 	+ Forward pass latency: 6.44e-02 (s)
+[2023-08-13 19:11:34,721][inference][INFO] - 	+ Forward pass throughput: 15.50 (samples/s)
+[2023-08-13 19:11:34,722][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:11:42,092][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:12:04,215][inference][INFO] - 	+ Generation pass latency: 7.37e+00 (s)
+[2023-08-13 19:12:04,217][inference][INFO] - 	+ Generation pass throughput: 27.10 (tokens/s)
+[2023-08-13 19:12:04,217][inference][INFO] - Saving inference results
+[2023-08-13 19:12:04,225][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: 20
+  input_shapes:
+    batch_size: 2
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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)
+0,16381.771776,0.0409,48.9,7.98,50.1

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/2/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:12:04,711][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:12:04,712][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:12:04,904][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:12:04,904][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:12:04,904][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:12:05,007][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:12:05,045][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:12:05,046][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-13 19:12:16,454][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:12:16,455][inference][INFO] - Running inference benchmark
+[2023-08-13 19:12:25,068][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:12:25,116][memory_tracker][INFO] - Peak memory usage: 16381.771776 MB
+[2023-08-13 19:12:25,116][inference][INFO] - 	+ Forward pass peak memory: 16381.771776 (MB)
+[2023-08-13 19:12:25,117][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:12:25,600][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:12:49,376][inference][INFO] - 	+ Forward pass latency: 4.09e-02 (s)
+[2023-08-13 19:12:49,377][inference][INFO] - 	+ Forward pass throughput: 48.90 (samples/s)
+[2023-08-13 19:12:49,377][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:12:58,181][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:13:22,117][inference][INFO] - 	+ Generation pass latency: 7.98e+00 (s)
+[2023-08-13 19:13:22,119][inference][INFO] - 	+ Generation pass throughput: 50.10 (tokens/s)
+[2023-08-13 19:13:22,119][inference][INFO] - Saving inference results
+[2023-08-13 19:13:22,126][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: 20
+  input_shapes:
+    batch_size: 2
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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)
+0,30780.817408,0.109,18.3,7.08,56.5

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/3/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:13:22,606][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:13:22,607][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:13:22,802][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:13:22,802][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:13:22,802][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:13:22,907][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:13:22,945][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:13:22,946][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-13 19:13:40,313][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:13:40,315][inference][INFO] - Running inference benchmark
+[2023-08-13 19:13:48,817][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:13:48,940][memory_tracker][INFO] - Peak memory usage: 30780.817408 MB
+[2023-08-13 19:13:48,940][inference][INFO] - 	+ Forward pass peak memory: 30780.817408 (MB)
+[2023-08-13 19:13:48,941][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:13:52,763][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:15:03,423][inference][INFO] - 	+ Forward pass latency: 1.09e-01 (s)
+[2023-08-13 19:15:03,424][inference][INFO] - 	+ Forward pass throughput: 18.30 (samples/s)
+[2023-08-13 19:15:03,424][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:15:10,517][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:15:31,769][inference][INFO] - 	+ Generation pass latency: 7.08e+00 (s)
+[2023-08-13 19:15:31,770][inference][INFO] - 	+ Generation pass throughput: 56.50 (tokens/s)
+[2023-08-13 19:15:31,771][inference][INFO] - Saving inference results
+[2023-08-13 19:15:31,779][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/4/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: 20
+  input_shapes:
+    batch_size: 4
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/4/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)
+0,17002.528768,0.0321,125.0,6.28,127.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/4/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:15:32,284][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:15:32,285][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:15:32,477][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:15:32,477][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:15:32,477][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:15:32,580][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:15:32,619][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:15:32,620][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-13 19:15:43,987][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:15:43,989][inference][INFO] - Running inference benchmark
+[2023-08-13 19:15:52,515][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:15:52,564][memory_tracker][INFO] - Peak memory usage: 17002.528768 MB
+[2023-08-13 19:15:52,565][inference][INFO] - 	+ Forward pass peak memory: 17002.528768 (MB)
+[2023-08-13 19:15:52,565][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:15:53,307][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:16:40,349][inference][INFO] - 	+ Forward pass latency: 3.21e-02 (s)
+[2023-08-13 19:16:40,350][inference][INFO] - 	+ Forward pass throughput: 125.00 (samples/s)
+[2023-08-13 19:16:40,351][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:16:47,605][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:17:12,734][inference][INFO] - 	+ Generation pass latency: 6.28e+00 (s)
+[2023-08-13 19:17:12,736][inference][INFO] - 	+ Generation pass throughput: 127.00 (tokens/s)
+[2023-08-13 19:17:12,736][inference][INFO] - Saving inference results
+[2023-08-13 19:17:12,749][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/5/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: 20
+  input_shapes:
+    batch_size: 4
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/5/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)
+0,31481.266175999997,0.187,21.4,7.7,104.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/5/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:17:13,238][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:17:13,239][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:17:13,430][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:17:13,430][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:17:13,431][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:17:13,533][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:17:13,573][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:17:13,573][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-13 19:17:30,896][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:17:30,898][inference][INFO] - Running inference benchmark
+[2023-08-13 19:17:39,448][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:17:39,661][memory_tracker][INFO] - Peak memory usage: 31481.266175999997 MB
+[2023-08-13 19:17:39,661][inference][INFO] - 	+ Forward pass peak memory: 31481.266175999997 (MB)
+[2023-08-13 19:17:39,666][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:17:46,593][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:19:01,255][inference][INFO] - 	+ Forward pass latency: 1.87e-01 (s)
+[2023-08-13 19:19:01,257][inference][INFO] - 	+ Forward pass throughput: 21.40 (samples/s)
+[2023-08-13 19:19:01,257][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:19:09,042][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:19:32,146][inference][INFO] - 	+ Generation pass latency: 7.70e+00 (s)
+[2023-08-13 19:19:32,148][inference][INFO] - 	+ Generation pass throughput: 104.00 (tokens/s)
+[2023-08-13 19:19:32,148][inference][INFO] - Saving inference results
+[2023-08-13 19:19:32,155][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/6/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: 20
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/6/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)
+0,19498.139648,0.0989,162.0,6.46,495.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/6/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:19:32,708][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:19:32,710][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:19:32,903][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:19:32,903][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:19:32,903][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:19:33,005][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:19:33,042][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:19:33,043][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float16 on device: cuda
+[2023-08-13 19:19:44,510][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:19:44,512][inference][INFO] - Running inference benchmark
+[2023-08-13 19:19:53,130][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:19:53,241][memory_tracker][INFO] - Peak memory usage: 19498.139648 MB
+[2023-08-13 19:19:53,241][inference][INFO] - 	+ Forward pass peak memory: 19498.139648 (MB)
+[2023-08-13 19:19:53,241][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:19:55,898][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:20:50,042][inference][INFO] - 	+ Forward pass latency: 9.89e-02 (s)
+[2023-08-13 19:20:50,043][inference][INFO] - 	+ Forward pass throughput: 162.00 (samples/s)
+[2023-08-13 19:20:50,043][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:20:56,729][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:21:22,585][inference][INFO] - 	+ Generation pass latency: 6.46e+00 (s)
+[2023-08-13 19:21:22,586][inference][INFO] - 	+ Generation pass throughput: 495.00 (tokens/s)
+[2023-08-13 19:21:22,586][inference][INFO] - Saving inference results
+[2023-08-13 19:21:22,593][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/7/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: 20
+  input_shapes:
+    batch_size: 16
+    sequence_length: 200
+    num_choices: 4
+    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: togethercomputer/LLaMA-2-7B-32K
+device: cuda
+task: text-generation
+hub_kwargs:
+  revision: main
+  cache_dir: null
+  force_download: false
+  local_files_only: false
+environment:
+  optimum_version: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/7/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)
+0,35824.467968,0.683,23.4,13.0,246.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/llama_1gpu_inference/7/main.log

@@ -0,0 +1,23 @@
+[2023-08-13 19:21:23,174][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:21:23,175][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:21:23,357][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type llama
+[2023-08-13 19:21:23,358][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:21:23,358][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:21:23,462][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:21:23,499][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:21:23,500][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: float32 on device: cuda
+[2023-08-13 19:21:40,982][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:21:40,983][inference][INFO] - Running inference benchmark
+[2023-08-13 19:21:49,497][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:21:50,216][memory_tracker][INFO] - Peak memory usage: 35824.467968 MB
+[2023-08-13 19:21:50,216][inference][INFO] - 	+ Forward pass peak memory: 35824.467968 (MB)
+[2023-08-13 19:21:50,232][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:22:15,578][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:23:33,502][inference][INFO] - 	+ Forward pass latency: 6.83e-01 (s)
+[2023-08-13 19:23:33,503][inference][INFO] - 	+ Forward pass throughput: 23.40 (samples/s)
+[2023-08-13 19:23:33,503][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:23:47,161][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:24:13,131][inference][INFO] - 	+ Generation pass latency: 1.30e+01 (s)
+[2023-08-13 19:24:13,133][inference][INFO] - 	+ Generation pass throughput: 246.00 (tokens/s)
+[2023-08-13 19:24:13,133][inference][INFO] - Saving inference results
+[2023-08-13 19:24:13,141][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: 10
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 4
+    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: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/pytorch_bert_inference/0/inference_results.csv

@@ -0,0 +1,2 @@
+,forward.peak_memory(MB),forward.latency(s),forward.throughput(samples/s)
+0,460.23065599999995,0.00324,309.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/pytorch_bert_inference/0/main.log

@@ -0,0 +1,20 @@
+[2023-08-13 19:24:17,896][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:24:17,897][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:24:18,071][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForSequenceClassification for task text-classification and model_type bert
+[2023-08-13 19:24:18,071][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:24:18,071][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:24:18,071][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:24:18,073][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:24:18,073][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: None on device: cpu
+[2023-08-13 19:24:18,666][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:24:18,667][inference][INFO] - Running inference benchmark
+[2023-08-13 19:24:18,804][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-13 19:24:18,806][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:24:18,860][inference][INFO] - 	+ Forward pass peak memory: 460.23065599999995 (MB)
+[2023-08-13 19:24:18,862][dummy_input][INFO] - Generating dummy input for: ['input_ids', 'attention_mask', 'token_type_ids']
+[2023-08-13 19:24:18,863][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:24:18,896][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:24:29,007][inference][INFO] - 	+ Forward pass latency: 3.24e-03 (s)
+[2023-08-13 19:24:29,009][inference][INFO] - 	+ Forward pass throughput: 309.00 (samples/s)
+[2023-08-13 19:24:29,009][inference][INFO] - Saving inference results
+[2023-08-13 19:24:29,023][backend][INFO] - Cleaning backend

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/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: true
+  warmup_runs: 10
+  benchmark_duration: 10
+  input_shapes:
+    batch_size: 1
+    sequence_length: 16
+    num_choices: 4
+    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: 1.11.0
+  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/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/pytorch_gpt2_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)
+0,463.59756799999997,0.00336,298.0,0.481,208.0

raw_results/2023-08-13_17:47:04_2b22cde71e549e194a3b7046b7595030382382e9/pytorch_gpt2_inference/0/main.log

@@ -0,0 +1,22 @@
+[2023-08-13 19:24:33,409][benchmark][INFO] - Configuring inference benchmark
+[2023-08-13 19:24:33,410][benchmark][INFO] - 	+ Setting seed(42)
+[2023-08-13 19:24:33,589][pytorch][INFO] - 	+ Infered AutoModel class AutoModelForCausalLM for task text-generation and model_type gpt2
+[2023-08-13 19:24:33,589][backend][INFO] - Configuring pytorch backend
+[2023-08-13 19:24:33,589][backend][INFO] - 	+ Checking initial device isolation
+[2023-08-13 19:24:33,589][backend][INFO] - 	+ Checking contineous device isolation
+[2023-08-13 19:24:33,591][pytorch][INFO] - 	+ Disabling gradients
+[2023-08-13 19:24:33,591][pytorch][INFO] - 	+ Loading pretrained model weights in dtype: None on device: cpu
+[2023-08-13 19:24:34,236][pytorch][INFO] - 	+ Turning on eval mode
+[2023-08-13 19:24:34,237][inference][INFO] - Running inference benchmark
+[2023-08-13 19:24:34,429][inference][INFO] - 	+ Tracking forward pass peak memory
+[2023-08-13 19:24:34,479][inference][INFO] - 	+ Forward pass peak memory: 463.59756799999997 (MB)
+[2023-08-13 19:24:34,480][inference][INFO] - 	+ Warming up the forward pass
+[2023-08-13 19:24:34,514][inference][INFO] - 	+ Tracking forward pass latency and throughput
+[2023-08-13 19:24:44,616][inference][INFO] - 	+ Forward pass latency: 3.36e-03 (s)
+[2023-08-13 19:24:44,619][inference][INFO] - 	+ Forward pass throughput: 298.00 (samples/s)
+[2023-08-13 19:24:44,620][inference][INFO] - 	+ Warming up the generation pass
+[2023-08-13 19:24:45,113][inference][INFO] - 	+ Tracking generation latency and throughput
+[2023-08-13 19:24:55,225][inference][INFO] - 	+ Generation pass latency: 4.81e-01 (s)
+[2023-08-13 19:24:55,225][inference][INFO] - 	+ Generation pass throughput: 208.00 (tokens/s)
+[2023-08-13 19:24:55,226][inference][INFO] - Saving inference results
+[2023-08-13 19:24:55,239][backend][INFO] - Cleaning backend