Datasets:

indahl
/

model-data-pack

MODEL Data Pack
Heuremen LLC · https://indahl.ai/hermes/model

Attribution. The full terms are in LICENSE.txt and this notice says the same.

MLCommons results
  MLPerf® Inference v5.0, v5.1, v6.0 and v6.1 results, datacenter and edge suites, closed and open divisions, as published by MLCommons Association in summary_results.json of the repositories inference_results_v5.0, inference_results_v5.1 and inference_results_v6.0 and in summary.json of the repository inference_results_v6.1 at github.com/mlcommons, retrieved 28 October 2026. MLPerf® Training v5.0, v5.1 and v6.0 results as published in the result logs of the repositories training_results_v5.0, training_results_v5.1 and training_results_v6.0, retrieved 28 October 2026. MLPerf® Storage v3.0 results as published in the repository storage_results_v3.0, retrieved 28 October 2026.
  Licence: Apache License 2.0. https://www.apache.org/licenses/LICENSE-2.0 . The text is in APACHE-2.0.txt.
  Applies to these files:
  - json/accelerators_summary.json
  - json/storage_v30_results.json
  - csv/accelerators.csv
  - csv/accelerators_summary.csv
  - csv/storage_v30_rows.csv
  - csv/storage_v30_meta.csv
  - csv/claims_mlcommons_files.csv
  Changes by Heuremen LLC: accelerators.csv keeps only closed-division datacenter results that have an accelerator count and a throughput unit, and adds the column per_accelerator, which is the result divided by the number of accelerators in the entry. accelerators_summary.json and the CSV made from it give the count, minimum, median, maximum and best entry for each group. storage_v30_results.json is our aggregation of the Storage v3.0 result files into one row per system and workload, where for training the value is the mean of the runs after the first. The per-accelerator values, ratios and percentages are our own computations. They are not primary MLPerf® metrics and are not verified by MLCommons Association.

Trademarks and verification
  Unverified results have not been through an MLPerf® review and may use measurement methodologies and/or workload implementations that are inconsistent with the MLPerf® specification for verified results. The per-accelerator values, ratios and percentages in this pack are our own computations from the MLCommons results. Result not verified by MLCommons Association.
  The MLPerf® and MLCOMMONS® trademarks and design versions thereof are registered and unregistered trademarks and service marks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use is strictly prohibited. See www.mlcommons.org for more information.
  Heuremen LLC is a member of MLCommons, and nothing in this pack is endorsed by MLCommons Association.
  Where this pack names an MLPerf result, the row carries the benchmark suite and version, the model and scenario, and the MLPerf result ID, and the date and source of retrieval are given above. The accelerators.csv rows are Closed division, datacenter suite. MLPerf results may only be compared with compatible MLPerf results, and the MLPerf Results Messaging Guidelines at https://github.com/mlcommons/policies/blob/master/MLPerf_Results_Messaging_Guidelines.adoc apply to anyone who repeats them.

Vendor and other sources
  Sentences in the statement column of csv/claims.csv and in json/claims.json are quoted from the page or document named in the source_ columns of the same row, fetched on 2026-10-28. They belong to the owners of those pages and keep their terms. They are quoted so that each statement can be set beside the published results.
  Applies to these files:
  - json/claims.json
  - csv/claims.csv
  - csv/claims_sources.csv
  The columns compared, in_file, outcome and note of these files are Heuremen LLC's own text.

Heuremen LLC
  The README, SOURCES.md, NOTICE.txt, LICENSE.txt, TESTED.md, START_HERE.html, verify.py, the build code, the compilation and the CSV conversion, the readings in json/mac_run.jsonl and csv/mac_run.csv, and these files may be used, copied and changed with attribution to Heuremen LLC and https://indahl.ai/hermes/model:
  - json/mac_run.jsonl
  - csv/mac_run.csv
  - csv/claims_tally.csv
  - csv/claims_vendor_page_hash.csv
  - csv/claims_meta.csv
  For the text files (the README, SOURCES.md, NOTICE.txt, LICENSE.txt, TESTED.md and START_HERE.html) and for the columns compared, in_file, outcome and note of the claims files, that permission does not include use as AI training data. The reservation against text and data mining and AI training covers that text only. It does not apply to the MLCommons results or to the sentences quoted from the sources.

	MODEL Data Pack
	Heuremen LLC · https://indahl.ai/hermes/model

	Attribution. The full terms are in LICENSE.txt and this notice says the same.

	MLCommons results
	MLPerf® Inference v5.0, v5.1, v6.0 and v6.1 results, datacenter and edge suites, closed and open divisions, as published by MLCommons Association in summary_results.json of the repositories inference_results_v5.0, inference_results_v5.1 and inference_results_v6.0 and in summary.json of the repository inference_results_v6.1 at github.com/mlcommons, retrieved 28 October 2026. MLPerf® Training v5.0, v5.1 and v6.0 results as published in the result logs of the repositories training_results_v5.0, training_results_v5.1 and training_results_v6.0, retrieved 28 October 2026. MLPerf® Storage v3.0 results as published in the repository storage_results_v3.0, retrieved 28 October 2026.
	Licence: Apache License 2.0. https://www.apache.org/licenses/LICENSE-2.0 . The text is in APACHE-2.0.txt.
	Applies to these files:
	- json/accelerators_summary.json
	- json/storage_v30_results.json
	- csv/accelerators.csv
	- csv/accelerators_summary.csv
	- csv/storage_v30_rows.csv
	- csv/storage_v30_meta.csv
	- csv/claims_mlcommons_files.csv
	Changes by Heuremen LLC: accelerators.csv keeps only closed-division datacenter results that have an accelerator count and a throughput unit, and adds the column per_accelerator, which is the result divided by the number of accelerators in the entry. accelerators_summary.json and the CSV made from it give the count, minimum, median, maximum and best entry for each group. storage_v30_results.json is our aggregation of the Storage v3.0 result files into one row per system and workload, where for training the value is the mean of the runs after the first. The per-accelerator values, ratios and percentages are our own computations. They are not primary MLPerf® metrics and are not verified by MLCommons Association.

	Trademarks and verification
	Unverified results have not been through an MLPerf® review and may use measurement methodologies and/or workload implementations that are inconsistent with the MLPerf® specification for verified results. The per-accelerator values, ratios and percentages in this pack are our own computations from the MLCommons results. Result not verified by MLCommons Association.
	The MLPerf® and MLCOMMONS® trademarks and design versions thereof are registered and unregistered trademarks and service marks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use is strictly prohibited. See www.mlcommons.org for more information.
	Heuremen LLC is a member of MLCommons, and nothing in this pack is endorsed by MLCommons Association.
	Where this pack names an MLPerf result, the row carries the benchmark suite and version, the model and scenario, and the MLPerf result ID, and the date and source of retrieval are given above. The accelerators.csv rows are Closed division, datacenter suite. MLPerf results may only be compared with compatible MLPerf results, and the MLPerf Results Messaging Guidelines at https://github.com/mlcommons/policies/blob/master/MLPerf_Results_Messaging_Guidelines.adoc apply to anyone who repeats them.

	Vendor and other sources
	Sentences in the statement column of csv/claims.csv and in json/claims.json are quoted from the page or document named in the source_ columns of the same row, fetched on 2026-10-28. They belong to the owners of those pages and keep their terms. They are quoted so that each statement can be set beside the published results.
	Applies to these files:
	- json/claims.json
	- csv/claims.csv
	- csv/claims_sources.csv
	The columns compared, in_file, outcome and note of these files are Heuremen LLC's own text.

	Heuremen LLC
	The README, SOURCES.md, NOTICE.txt, LICENSE.txt, TESTED.md, START_HERE.html, verify.py, the build code, the compilation and the CSV conversion, the readings in json/mac_run.jsonl and csv/mac_run.csv, and these files may be used, copied and changed with attribution to Heuremen LLC and https://indahl.ai/hermes/model:
	- json/mac_run.jsonl
	- csv/mac_run.csv
	- csv/claims_tally.csv
	- csv/claims_vendor_page_hash.csv
	- csv/claims_meta.csv
	For the text files (the README, SOURCES.md, NOTICE.txt, LICENSE.txt, TESTED.md and START_HERE.html) and for the columns compared, in_file, outcome and note of the claims files, that permission does not include use as AI training data. The reservation against text and data mining and AI training covers that text only. It does not apply to the MLCommons results or to the sentences quoted from the sources.