datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
kernelbench-mega-traces
KernelBench-Mega agent traces
Coding agents writing full GPU megakernels across Blackwell / H100 / B200, scored as speedup over reference; contamination-audited (23 verified cells).
Each .jsonl file is one agent run in Claude-Code session format, viewable with the agent trace viewer. Filename = run id; manifest.csv maps each run to model / harness / problem / GPU / score.
23 agent traces · live leaderboard: https://kernelbench.com/mega
Secrets redacted. Full reasoning for… See the full description on the dataset page: https://huggingface.co/datasets/Infatoshi/kernelbench-mega-traces.quran-asr-mega-corpusdolma3_dolmino_megatron_tokenize
Dolma 3 / Dolmino Megatron-LM indexed dataset
This repository contains immutable Megatron-LM indexed datasets (.bin and
.idx) produced from pinned Dolma 3 and Dolmino releases. It intentionally
contains no training checkpoints, experiment outputs, logs, or dataset caches.
The indexed payloads were derived from these pinned public datasets:
allenai/dolma3_mix-150B-1025@afa92bfb22366821c5e6cd427cdd036b34b713ef… See the full description on the dataset page: https://huggingface.co/datasets/Mr-Philo/dolma3_dolmino_megatron_tokenize.lm-eval-results-Eurdem-megatron_2.1_MoE_2x7B-private
Dataset Card for Evaluation run of Eurdem/megatron_2.1_MoE_2x7B
Dataset automatically created during the evaluation run of model Eurdem/megatron_2.1_MoE_2x7B
The dataset is composed of 62 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.… See the full description on the dataset page: https://huggingface.co/datasets/nyu-dice-lab/lm-eval-results-Eurdem-megatron_2.1_MoE_2x7B-private.requestsLunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 3 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8.9-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8.9
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8.9
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8.9-details.megamath-web-pro-max-splittedprithivMLmods__Megatron-Corpus-14B-Exp-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Corpus-14B-Exp
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Corpus-14B-Exp
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Corpus-14B-Exp-details.cd_hparam_search_whisper_th_megaspeech_v3Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v4-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v4
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v4
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v4-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9.2-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9.2
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9.2
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9.2-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8.7-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8.7
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8.7
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v8.7-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v5-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v5
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v5
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v5-details.requests_debugprithivMLmods__Megatron-Corpus-14B-Exp.v2-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Corpus-14B-Exp.v2
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Corpus-14B-Exp.v2
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Corpus-14B-Exp.v2-details.aws-prototyping__MegaBeam-Mistral-7B-512k-details
Dataset Card for Evaluation run of aws-prototyping/MegaBeam-Mistral-7B-512k
Dataset automatically created during the evaluation run of model aws-prototyping/MegaBeam-Mistral-7B-512k
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/aws-prototyping__MegaBeam-Mistral-7B-512k-details.prithivMLmods__Megatron-Opus-14B-2.0-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Opus-14B-2.0
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Opus-14B-2.0
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Opus-14B-2.0-details.agentic-trace-megacorpus-10tb
Agentic Trace Megacorpus — ~2.06 TB (aggregated by reference)
A reference-aggregation of 432 public trace datasets (agentic coding,
reasoning, SWE, tool-use) totaling ~2.06 TB,
assembled for MiniMax-M3 post-training. Nothing is re-hosted by value here yet —
the loader streams directly from each source repo. Byte mirroring into this repo
(to fill the 10 TB quota) is done Hub→Hub via mirror_to_hub.py.
Composition (TB by relevance group)
group
TB
agentic… See the full description on the dataset page: https://huggingface.co/datasets/srbwin/agentic-trace-megacorpus-10tb.prithivMLmods__Megatron-Opus-14B-2.1-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Opus-14B-2.1
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Opus-14B-2.1
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Opus-14B-2.1-details.ZeroXClem__Llama-3.1-8B-AthenaSky-MegaMix-details
Dataset Card for Evaluation run of ZeroXClem/Llama-3.1-8B-AthenaSky-MegaMix
Dataset automatically created during the evaluation run of model ZeroXClem/Llama-3.1-8B-AthenaSky-MegaMix
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/ZeroXClem__Llama-3.1-8B-AthenaSky-MegaMix-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9.1-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9.1
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9.1
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9.1-details.Junhoee__Qwen-Megumin-details
Dataset Card for Evaluation run of Junhoee/Qwen-Megumin
Dataset automatically created during the evaluation run of model Junhoee/Qwen-Megumin
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Junhoee__Qwen-Megumin-details.prithivMLmods__Megatron-Opus-14B-Exp-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Opus-14B-Exp
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Opus-14B-Exp
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Opus-14B-Exp-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v3-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v3
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v3
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 2 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v3-details.CultriX__Qwen2.5-14B-MegaMerge-pt2-details
Dataset Card for Evaluation run of CultriX/Qwen2.5-14B-MegaMerge-pt2
Dataset automatically created during the evaluation run of model CultriX/Qwen2.5-14B-MegaMerge-pt2
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/CultriX__Qwen2.5-14B-MegaMerge-pt2-details.prithivMLmods__Megatron-Opus-7B-Exp-details
Dataset Card for Evaluation run of prithivMLmods/Megatron-Opus-7B-Exp
Dataset automatically created during the evaluation run of model prithivMLmods/Megatron-Opus-7B-Exp
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/prithivMLmods__Megatron-Opus-7B-Exp-details.Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9-details
Dataset Card for Evaluation run of Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9
Dataset automatically created during the evaluation run of model Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v9
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Lunzima__NQLSG-Qwen2.5-14B-MegaFusion-v9-details.shelflife-dataamazon__MegaBeam-Mistral-7B-300k-details
Dataset Card for Evaluation run of amazon/MegaBeam-Mistral-7B-300k
Dataset automatically created during the evaluation run of model amazon/MegaBeam-Mistral-7B-300k
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/amazon__MegaBeam-Mistral-7B-300k-details.