The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
Error code: DatasetGenerationCastError Exception: DatasetGenerationCastError Message: An error occurred while generating the dataset All the data files must have the same columns, but at some point there are 76 new columns ({'mean_input_tokens', 'results_end_to_end_latency_s_min', 'results_request_output_throughput_token_per_s_quantiles_p90', 'stddev_output_tokens', 'results_request_output_throughput_token_per_s_stddev', 'results_number_output_tokens_mean', 'mean_output_tokens', 'results_number_input_tokens_stddev', 'results_num_requests_started', 'results_ttft_s_quantiles_p50', 'results_number_output_tokens_min', 'results_number_input_tokens_quantiles_p75', 'results_number_input_tokens_max', 'results_number_output_tokens_quantiles_p50', 'results_inter_token_latency_s_min', 'results_end_to_end_latency_s_quantiles_p90', 'results_end_to_end_latency_s_max', 'results_number_input_tokens_quantiles_p50', 'results_number_output_tokens_quantiles_p99', 'results_ttft_s_stddev', 'results_inter_token_latency_s_stddev', 'results_number_input_tokens_mean', 'results_end_to_end_latency_s_stddev', 'results_inter_token_latency_s_quantiles_p75', 'results_number_output_tokens_quantiles_p90', 'results_ttft_s_max', 'results_end_to_end_latency_s_quantiles_p50', 'model', 'results_inter_token_latency_s_mean', 'results_ttft_s_quantiles_p75', 'results_end_to_end_latency_s_quantiles_p75', 'results_error_code_frequency', 'results_inter_token_latency_s_quantiles_p25', 'results_end_to_end_latency_s_quantiles_p95', 'results_request_output_throughput_token_per_s_quantiles_p25', 'stddev_input_tokens', 'results_end_to_end_latency_s_mean', 'results_number_input_tokens_quantiles_p99', 'results_error_rate', 'results_number_output_tokens_quantiles_p25', 'results_number_output_tokens_quantiles_p75', 'results_inter_token_latency_s_quantiles_p99', 'results_number_output_tokens_quantiles_p95', 'results_number_output_tokens_stddev', 'results_inter_token_latency_s_quantiles_p50', 'results_request_output_throughput_token_per_s_mean', 'version', 'results_request_output_throughput_token_per_s_quantiles_p99', 'results_number_output_tokens_max', 'results_request_output_throughput_token_per_s_max', 'results_end_to_end_latency_s_quantiles_p99', 'timestamp', 'num_concurrent_requests', 'results_number_errors', 'results_num_completed_requests', 'results_number_input_tokens_quantiles_p90', 'results_request_output_throughput_token_per_s_quantiles_p75', 'results_end_to_end_latency_s_quantiles_p25', 'results_ttft_s_min', 'results_request_output_throughput_token_per_s_quantiles_p95', 'results_inter_token_latency_s_quantiles_p90', 'results_mean_output_throughput_token_per_s', 'results_ttft_s_quantiles_p95', 'results_request_output_throughput_token_per_s_quantiles_p50', 'results_ttft_s_mean', 'results_ttft_s_quantiles_p25', 'results_number_input_tokens_min', 'results_num_completed_requests_per_min', 'results_ttft_s_quantiles_p90', 'results_ttft_s_quantiles_p99', 'results_number_input_tokens_quantiles_p25', 'results_inter_token_latency_s_quantiles_p95', 'results_inter_token_latency_s_max', 'results_request_output_throughput_token_per_s_min', 'results_number_input_tokens_quantiles_p95', 'name'}) and 9 missing columns ({'error_msg', 'request_output_throughput_token_per_s', 'inter_token_latency_s', 'number_total_tokens', 'end_to_end_latency_s', 'error_code', 'number_input_tokens', 'number_output_tokens', 'ttft_s'}). This happened while the json dataset builder was generating data using hf://datasets/ssong1/llmperf-bedrock/raw_data/bedrock-anthropic-claude-instant-v1_1024_0_1024_100_summary.json (at revision 820b8e769eb61c207909825cbf4a4e58f3914a1d) Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations) Traceback: Traceback (most recent call last): File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py", line 2011, in _prepare_split_single writer.write_table(table) File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/arrow_writer.py", line 585, in write_table pa_table = table_cast(pa_table, self._schema) File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/table.py", line 2302, in table_cast return cast_table_to_schema(table, schema) File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/table.py", line 2256, in cast_table_to_schema raise CastError( datasets.table.CastError: Couldn't cast version: timestamp[s] name: string model: string mean_input_tokens: int64 stddev_input_tokens: int64 mean_output_tokens: int64 stddev_output_tokens: int64 num_concurrent_requests: int64 results_inter_token_latency_s_quantiles_p25: double results_inter_token_latency_s_quantiles_p50: double results_inter_token_latency_s_quantiles_p75: double results_inter_token_latency_s_quantiles_p90: double results_inter_token_latency_s_quantiles_p95: double results_inter_token_latency_s_quantiles_p99: double results_inter_token_latency_s_mean: double results_inter_token_latency_s_min: double results_inter_token_latency_s_max: double results_inter_token_latency_s_stddev: double results_ttft_s_quantiles_p25: double results_ttft_s_quantiles_p50: double results_ttft_s_quantiles_p75: double results_ttft_s_quantiles_p90: double results_ttft_s_quantiles_p95: double results_ttft_s_quantiles_p99: double results_ttft_s_mean: double results_ttft_s_min: double results_ttft_s_max: double results_ttft_s_stddev: double results_end_to_end_latency_s_quantiles_p25: double results_end_to_end_latency_s_quantiles_p50: double results_end_to_end_latency_s_quantiles_p75: double results_end_to_end_latency_s_quantiles_p90: double results_end_to_end_latency_s_quantiles_p95: double results_end_to_end_latency_s_quantiles_p99: double results_end_to_end_latency_s_mean: double results_end_to_end_latency_s_min: double results_end_to_end_latency_s_max: double results_end_to_end_latency_s_stddev: double results_request_output ... throughput_token_per_s_quantiles_p99: double results_request_output_throughput_token_per_s_mean: double results_request_output_throughput_token_per_s_min: double results_request_output_throughput_token_per_s_max: double results_request_output_throughput_token_per_s_stddev: double results_number_input_tokens_quantiles_p25: double results_number_input_tokens_quantiles_p50: double results_number_input_tokens_quantiles_p75: double results_number_input_tokens_quantiles_p90: double results_number_input_tokens_quantiles_p95: double results_number_input_tokens_quantiles_p99: double results_number_input_tokens_mean: double results_number_input_tokens_min: string results_number_input_tokens_max: string results_number_input_tokens_stddev: double results_number_output_tokens_quantiles_p25: double results_number_output_tokens_quantiles_p50: double results_number_output_tokens_quantiles_p75: double results_number_output_tokens_quantiles_p90: double results_number_output_tokens_quantiles_p95: double results_number_output_tokens_quantiles_p99: double results_number_output_tokens_mean: double results_number_output_tokens_min: string results_number_output_tokens_max: string results_number_output_tokens_stddev: double results_num_requests_started: int64 results_error_rate: double results_number_errors: int64 results_error_code_frequency: string results_mean_output_throughput_token_per_s: double results_num_completed_requests: int64 results_num_completed_requests_per_min: double timestamp: int64 to {'error_msg': Value(dtype='string', id=None), 'request_output_throughput_token_per_s': Value(dtype='float64', id=None), 'inter_token_latency_s': Value(dtype='float64', id=None), 'number_total_tokens': Value(dtype='int64', id=None), 'end_to_end_latency_s': Value(dtype='float64', id=None), 'error_code': Value(dtype='null', id=None), 'number_input_tokens': Value(dtype='int64', id=None), 'number_output_tokens': Value(dtype='int64', id=None), 'ttft_s': Value(dtype='float64', id=None)} because column names don't match During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1321, in compute_config_parquet_and_info_response parquet_operations = convert_to_parquet(builder) File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 935, in convert_to_parquet builder.download_and_prepare( File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py", line 1027, in download_and_prepare self._download_and_prepare( File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py", line 1122, in _download_and_prepare self._prepare_split(split_generator, **prepare_split_kwargs) File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py", line 1882, in _prepare_split for job_id, done, content in self._prepare_split_single( File "/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py", line 2013, in _prepare_split_single raise DatasetGenerationCastError.from_cast_error( datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset All the data files must have the same columns, but at some point there are 76 new columns ({'mean_input_tokens', 'results_end_to_end_latency_s_min', 'results_request_output_throughput_token_per_s_quantiles_p90', 'stddev_output_tokens', 'results_request_output_throughput_token_per_s_stddev', 'results_number_output_tokens_mean', 'mean_output_tokens', 'results_number_input_tokens_stddev', 'results_num_requests_started', 'results_ttft_s_quantiles_p50', 'results_number_output_tokens_min', 'results_number_input_tokens_quantiles_p75', 'results_number_input_tokens_max', 'results_number_output_tokens_quantiles_p50', 'results_inter_token_latency_s_min', 'results_end_to_end_latency_s_quantiles_p90', 'results_end_to_end_latency_s_max', 'results_number_input_tokens_quantiles_p50', 'results_number_output_tokens_quantiles_p99', 'results_ttft_s_stddev', 'results_inter_token_latency_s_stddev', 'results_number_input_tokens_mean', 'results_end_to_end_latency_s_stddev', 'results_inter_token_latency_s_quantiles_p75', 'results_number_output_tokens_quantiles_p90', 'results_ttft_s_max', 'results_end_to_end_latency_s_quantiles_p50', 'model', 'results_inter_token_latency_s_mean', 'results_ttft_s_quantiles_p75', 'results_end_to_end_latency_s_quantiles_p75', 'results_error_code_frequency', 'results_inter_token_latency_s_quantiles_p25', 'results_end_to_end_latency_s_quantiles_p95', 'results_request_output_throughput_token_per_s_quantiles_p25', 'stddev_input_tokens', 'results_end_to_end_latency_s_mean', 'results_number_input_tokens_quantiles_p99', 'results_error_rate', 'results_number_output_tokens_quantiles_p25', 'results_number_output_tokens_quantiles_p75', 'results_inter_token_latency_s_quantiles_p99', 'results_number_output_tokens_quantiles_p95', 'results_number_output_tokens_stddev', 'results_inter_token_latency_s_quantiles_p50', 'results_request_output_throughput_token_per_s_mean', 'version', 'results_request_output_throughput_token_per_s_quantiles_p99', 'results_number_output_tokens_max', 'results_request_output_throughput_token_per_s_max', 'results_end_to_end_latency_s_quantiles_p99', 'timestamp', 'num_concurrent_requests', 'results_number_errors', 'results_num_completed_requests', 'results_number_input_tokens_quantiles_p90', 'results_request_output_throughput_token_per_s_quantiles_p75', 'results_end_to_end_latency_s_quantiles_p25', 'results_ttft_s_min', 'results_request_output_throughput_token_per_s_quantiles_p95', 'results_inter_token_latency_s_quantiles_p90', 'results_mean_output_throughput_token_per_s', 'results_ttft_s_quantiles_p95', 'results_request_output_throughput_token_per_s_quantiles_p50', 'results_ttft_s_mean', 'results_ttft_s_quantiles_p25', 'results_number_input_tokens_min', 'results_num_completed_requests_per_min', 'results_ttft_s_quantiles_p90', 'results_ttft_s_quantiles_p99', 'results_number_input_tokens_quantiles_p25', 'results_inter_token_latency_s_quantiles_p95', 'results_inter_token_latency_s_max', 'results_request_output_throughput_token_per_s_min', 'results_number_input_tokens_quantiles_p95', 'name'}) and 9 missing columns ({'error_msg', 'request_output_throughput_token_per_s', 'inter_token_latency_s', 'number_total_tokens', 'end_to_end_latency_s', 'error_code', 'number_input_tokens', 'number_output_tokens', 'ttft_s'}). This happened while the json dataset builder was generating data using hf://datasets/ssong1/llmperf-bedrock/raw_data/bedrock-anthropic-claude-instant-v1_1024_0_1024_100_summary.json (at revision 820b8e769eb61c207909825cbf4a4e58f3914a1d) Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
request_output_throughput_token_per_s
float64 | inter_token_latency_s
float64 | error_msg
string | error_code
null | end_to_end_latency_s
float64 | number_input_tokens
int64 | number_output_tokens
int64 | ttft_s
float64 | number_total_tokens
int64 |
---|---|---|---|---|---|---|---|---|
80.709977 | 0.01101 | null | 9.366872 | 1,024 | 850 | 1.296926 | 1,874 |
|
85.792044 | 0.009911 | null | 7.529836 | 1,024 | 759 | 1.16885 | 1,783 |
|
32.07669 | 0.024916 | null | 1.65229 | 1,024 | 66 | 1.24543 | 1,090 |
|
53.063796 | 0.016905 | null | 12.060954 | 1,024 | 713 | 1.222925 | 1,737 |
|
74.019346 | 0.012365 | null | 7.06572 | 1,024 | 571 | 1.172058 | 1,595 |
|
76.539988 | 0.010764 | null | 3.958715 | 1,024 | 367 | 1.181337 | 1,391 |
|
59.497714 | 0.014155 | null | 11.445818 | 1,024 | 808 | 1.183751 | 1,832 |
|
77.100792 | 0.01087 | null | 4.280112 | 1,024 | 393 | 1.160499 | 1,417 |
|
69.68627 | 0.011976 | null | 6.859314 | 1,024 | 572 | 1.179717 | 1,596 |
|
63.723908 | 0.013348 | null | 5.947532 | 1,024 | 445 | 1.215837 | 1,469 |
|
72.890522 | 0.011173 | null | 5.460243 | 1,024 | 488 | 1.232071 | 1,512 |
|
85.407083 | 0.00978 | null | 4.460988 | 1,024 | 455 | 1.26758 | 1,479 |
|
89.814971 | 0.009159 | null | 4.642879 | 1,024 | 506 | 1.174789 | 1,530 |
|
53.886441 | 0.016375 | null | 3.544491 | 1,024 | 216 | 1.220313 | 1,240 |
|
70.811874 | 0.012504 | null | 8.473155 | 1,024 | 677 | 1.283001 | 1,701 |
|
65.432664 | 0.012743 | null | 11.492731 | 1,024 | 901 | 1.324202 | 1,925 |
|
37.786261 | 0.023529 | null | 20.457171 | 1,024 | 869 | 1.219398 | 1,893 |
|
30.015202 | 0.029718 | null | 23.854579 | 1,024 | 802 | 2.155391 | 1,826 |
|
16.055926 | 0.05334 | null | 10.83712 | 1,024 | 203 | 1.889337 | 1,227 |
|
16.528309 | 0.050152 | null | 16.456614 | 1,024 | 328 | 1.460065 | 1,352 |
|
92.596693 | 0.009434 | null | 8.423627 | 1,024 | 892 | 1.588464 | 1,916 |
|
62.948711 | 0.013339 | null | 7.879431 | 1,024 | 590 | 1.163141 | 1,614 |
|
37.374297 | 0.022191 | null | 4.735875 | 1,024 | 213 | 2.199076 | 1,237 |
|
53.875938 | 0.015837 | null | 10.301445 | 1,024 | 650 | 1.179561 | 1,674 |
|
103.315625 | 0.008574 | null | 6.688243 | 1,024 | 779 | 1.303697 | 1,803 |
|
68.865003 | 0.01272 | null | 5.184056 | 1,024 | 407 | 1.225452 | 1,431 |
|
75.572458 | 0.011215 | null | 3.149295 | 1,024 | 279 | 1.173232 | 1,303 |
|
85.355607 | 0.01032 | null | 10.637848 | 1,024 | 1,030 | 1.157395 | 2,054 |
|
99.610734 | 0.008874 | null | 6.876769 | 1,024 | 774 | 1.168326 | 1,798 |
|
69.199754 | 0.011711 | null | 5.043371 | 1,024 | 430 | 1.255332 | 1,454 |
|
59.626878 | 0.014051 | null | 9.391738 | 1,024 | 668 | 1.220992 | 1,692 |
|
58.009859 | 0.014669 | null | 9.291524 | 1,024 | 633 | 1.153024 | 1,657 |
|
100.220617 | 0.008426 | null | 6.126484 | 1,024 | 726 | 1.183953 | 1,750 |
|
65.022949 | 0.013012 | null | 12.903137 | 1,024 | 991 | 1.181941 | 2,015 |
|
55.902366 | 0.016034 | null | 7.155332 | 1,024 | 445 | 1.236226 | 1,469 |
|
64.519107 | 0.013969 | null | 4.463794 | 1,024 | 318 | 1.152611 | 1,342 |
|
65.747195 | 0.013013 | null | 8.07639 | 1,024 | 619 | 1.166912 | 1,643 |
|
98.187761 | 0.008479 | null | 6.222771 | 1,024 | 733 | 1.152345 | 1,757 |
|
67.901878 | 0.012692 | null | 3.357786 | 1,024 | 264 | 1.230695 | 1,288 |
|
65.649574 | 0.012607 | null | 5.620752 | 1,024 | 445 | 1.93973 | 1,469 |
|
60.156785 | 0.014265 | null | 12.533914 | 1,024 | 878 | 1.153848 | 1,902 |
|
61.867365 | 0.014002 | null | 5.915882 | 1,024 | 422 | 1.156762 | 1,446 |
|
62.90672 | 0.01359 | null | 11.954208 | 1,024 | 879 | 1.273736 | 1,903 |
|
77.796138 | 0.010843 | null | 3.804816 | 1,024 | 349 | 1.231282 | 1,373 |
|
69.927269 | 0.011941 | null | 4.047062 | 1,024 | 338 | 1.353932 | 1,362 |
|
55.536745 | 0.01492 | null | 5.095725 | 1,024 | 341 | 1.175492 | 1,365 |
|
66.107427 | 0.012792 | null | 14.385676 | 1,024 | 1,124 | 1.129887 | 2,148 |
|
71.554572 | 0.011574 | null | 4.891372 | 1,024 | 422 | 1.282573 | 1,446 |
|
52.935421 | 0.015506 | null | 3.513715 | 1,024 | 226 | 1.186333 | 1,250 |
|
69.544767 | 0.012529 | null | 9.404015 | 1,024 | 750 | 1.242136 | 1,774 |
|
74.763363 | 0.011162 | null | 4.775066 | 1,024 | 427 | 1.244265 | 1,451 |
|
36.920591 | 0.02496 | null | 2.952282 | 1,024 | 118 | 1.441923 | 1,142 |
|
57.297283 | 0.015504 | null | 4.816982 | 1,024 | 310 | 1.166936 | 1,334 |
|
60.2727 | 0.013196 | null | 2.754149 | 1,024 | 208 | 1.185945 | 1,232 |
|
47.745166 | 0.017544 | null | 5.78069 | 1,024 | 329 | 1.84757 | 1,353 |
|
65.366222 | 0.012924 | null | 3.059684 | 1,024 | 236 | 1.214574 | 1,260 |
|
110.382373 | 0.008095 | null | 7.084464 | 1,024 | 874 | 1.182321 | 1,898 |
|
null | null | null | null | null | null | null | null | null |
59.605989 | 0.013723 | null | 4.261317 | 957 | 310 | 1.54734 | 1,267 |
|
35.023098 | 0.026349 | null | 1.798813 | 1,090 | 68 | 1.601813 | 1,158 |
|
54.495916 | 0.015222 | null | 4.23885 | 1,145 | 278 | 1.25157 | 1,423 |
|
71.464665 | 0.012198 | null | 9.291305 | 1,131 | 761 | 1.282382 | 1,892 |
|
49.206911 | 0.017683 | null | 1.971268 | 947 | 111 | 1.248526 | 1,058 |
|
66.824149 | 0.012658 | null | 3.905774 | 944 | 308 | 1.207053 | 1,252 |
|
57.286718 | 0.014215 | null | 3.264282 | 1,116 | 229 | 1.182101 | 1,345 |
|
50.743723 | 0.016372 | null | 3.52753 | 944 | 215 | 1.174207 | 1,159 |
|
45.518452 | 0.018593 | null | 9.490657 | 1,100 | 510 | 1.188139 | 1,610 |
|
25.526443 | 0.034759 | null | 2.232979 | 1,073 | 64 | 1.490863 | 1,137 |
|
62.937897 | 0.013654 | null | 3.749728 | 874 | 274 | 1.318778 | 1,148 |
|
103.93937 | 0.008574 | null | 8.351022 | 1,160 | 973 | 1.300075 | 2,133 |
|
85.488814 | 0.009869 | null | 6.035877 | 961 | 611 | 1.362104 | 1,572 |
|
72.095836 | 0.011846 | null | 3.217939 | 1,038 | 271 | 1.236247 | 1,309 |
|
41.693718 | 0.021209 | null | 4.485088 | 850 | 211 | 1.381482 | 1,061 |
|
82.141598 | 0.010176 | null | 5.015729 | 850 | 492 | 1.178206 | 1,342 |
|
61.75684 | 0.013694 | null | 13.71508 | 1,217 | 1,001 | 1.391986 | 2,218 |
|
44.225399 | 0.019092 | null | 4.228339 | 903 | 221 | 1.549759 | 1,124 |
|
59.100547 | 0.014513 | null | 6.886569 | 1,150 | 473 | 2.033308 | 1,623 |
|
99.594705 | 0.008915 | null | 6.606777 | 938 | 740 | 1.240329 | 1,678 |
|
73.217555 | 0.011393 | null | 3.264244 | 1,083 | 286 | 1.281553 | 1,369 |
|
63.240005 | 0.013096 | null | 14.911447 | 896 | 1,138 | 1.325655 | 2,034 |
|
35.929462 | 0.024576 | null | 3.423374 | 879 | 139 | 1.208694 | 1,018 |
|
68.864612 | 0.012065 | null | 7.405836 | 994 | 613 | 1.392854 | 1,607 |
|
72.953275 | 0.011191 | null | 3.399436 | 978 | 303 | 1.193455 | 1,281 |
|
17.776435 | 0.052378 | null | 1.631373 | 1,002 | 31 | 1.350369 | 1,033 |
|
79.185617 | 0.011379 | null | 12.628556 | 1,026 | 1,109 | 1.180356 | 2,135 |
|
81.140707 | 0.010344 | null | 5.200842 | 1,150 | 502 | 1.551933 | 1,652 |
|
73.38689 | 0.010882 | null | 3.120448 | 990 | 286 | 1.155602 | 1,276 |
|
72.092687 | 0.011572 | null | 3.273564 | 982 | 282 | 1.200175 | 1,264 |
|
70.341317 | 0.012095 | null | 3.696263 | 998 | 305 | 1.239015 | 1,303 |
|
75.982194 | 0.011066 | null | 3.329727 | 1,082 | 300 | 1.161832 | 1,382 |
|
44.242763 | 0.018871 | null | 2.893129 | 1,242 | 153 | 1.341739 | 1,395 |
|
43.5751 | 0.018947 | null | 4.268493 | 1,069 | 225 | 2.580754 | 1,294 |
|
63.75876 | 0.013142 | null | 7.747955 | 923 | 588 | 1.184635 | 1,511 |
|
48.014866 | 0.017654 | null | 8.851425 | 946 | 501 | 1.289251 | 1,447 |
|
71.606931 | 0.011418 | null | 3.421456 | 930 | 298 | 1.164875 | 1,228 |
|
67.756211 | 0.012469 | null | 6.080623 | 953 | 487 | 1.285378 | 1,440 |
|
69.975344 | 0.012014 | null | 4.95889 | 906 | 412 | 1.218765 | 1,318 |
|
43.3116 | 0.020465 | null | 12.144553 | 1,051 | 593 | 1.287809 | 1,644 |
|
68.703792 | 0.012574 | null | 5.137999 | 1,041 | 408 | 1.286154 | 1,449 |
|
47.219235 | 0.018963 | null | 15.86218 | 1,162 | 836 | 1.229736 | 1,998 |
Utilizing the LLMPerf, we have benchmarked a selection of LLM inference providers. Our analysis focuses on evaluating their performance, reliability, and efficiency under the following key metrics:
- Output tokens throughput, which represents the average number of output tokens returned per second. This metric is important for applications that require high throughput, such as summarization and translation, and easy to compare across different models and providers.
- Time to first token (TTFT), which represents the duration of time that LLM returns the first token. TTFT is especially important for streaming applications, such as chatbots.
Time to First Token (seconds)
For streaming applications, the TTFT is how long before the LLM returns the first token.
Framework | Model | Median | Mean | Min | Max | P25 | P75 | P95 | P99 |
---|---|---|---|---|---|---|---|---|---|
bedrock | claude-instant-v1 | 1.21 | 1.29 | 1.12 | 2.19 | 1.17 | 1.27 | 1.89 | 2.17 |
Output Tokens Throughput (tokens/s)
The output tokens throughput is measured as the average number of output tokens returned per second. We collect results by sending 100 requests to each LLM inference provider, and calculate the mean output tokens throughput based on 100 requests. A higher output tokens throughput indicates a higher throughput of the LLM inference provider.
Framework | Model | Median | Mean | Min | Max | P25 | P75 | P95 | P99 |
---|---|---|---|---|---|---|---|---|---|
bedrock | claude-instant-v1 | 65.64 | 65.98 | 16.05 | 110.38 | 57.29 | 75.57 | 99.73 | 106.42 |
Run Configurations
testscript token_benchmark_ray.py
For each provider, we perform:
- Total number of requests: 100
- Concurrency: 1
- Prompt's token length: 1024
- Expected output length: 1024
- Tested models: claude-instant-v1-100k
python token_benchmark_ray.py \
--model bedrock/anthropic.claude-instant-v1 \
--mean-input-tokens 1024 \
--stddev-input-tokens 0 \
--mean-output-tokens 1024 \
--stddev-output-tokens 100 \
--max-num-completed-requests 100 \
--num-concurrent-requests 1 \
--llm-api litellm
We ran the LLMPerf clients from an on-premise Kubernetes Bastion host. The results were up-to-date of January 19, 2023, 3pm KST. You could find the detailed results in the raw_data folder.
Caveats and Disclaimers
- The endpoints provider backend might vary widely, so this is not a reflection on how the software runs on a particular hardware.
- The results may vary with time of day.
- The results (e.g. measurement of TTFT) depend on client location, and can also be biased by some providers lagging on the first token in order to increase ITL.
- The results is only a proxy of the system capabilities and is also impacted by the existing system load and provider traffic.
- The results may not correlate with users’ workloads.
- Downloads last month
- 0