datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
llama-3.2-1b-atlas
llama-3.2-1b-atlas
Llama3-SSL4EO-S12-v1.1-captions
Llama3-SSL4EO-S12-Captions
The captions are aligned with the SSL4EO-S12 v1.1 dataset and were automatically generated using the Llama3-LLaVA-Next-8B model.
Please find more information regarding the generation and evaluation in the Llama3-MS-CLIP paper.
Code: https://github.com/IBM/MS-CLIP
Data Structure
We provide the captions in two versions: As a single compressed Parquet file per split and as CSV files with 256 captions each that match the Zarr Zip files of the… See the full description on the dataset page: https://huggingface.co/datasets/ibm-esa-geospatial/Llama3-SSL4EO-S12-v1.1-captions.hpltv2-llama33-edu-annotation
HPLT version 2.0 educational annotations
This dataset contains annotations derived from HPLT v2 cleaned samples.
There are 500,000 annotations for each language if the source contains at least 500,000 samples.
We prompt Llama-3.3-70B-Instruct to score web pages based on their educational value following FineWeb-Edu classifier.
Note 1: The dataset contains the prompt (using the first 1500 characters of the text sample), the scores, and the full Llama 3 generation. The column "idx"… See the full description on the dataset page: https://huggingface.co/datasets/LumiOpen/hpltv2-llama33-edu-annotation.RULER-llama3-1M
RULER-Llama3-1M
A 1M token version of the RULER dataset based on the Llama-3 chat template.
It is automatically generated based on the scripts available in the RULER repository: https://github.com/NVIDIA/RULER. It is designed for evaluating the performance of Long Language Models (LLMs) on various tasks with varying sequence lengths.
How to Use
from datasets import load_dataset
LENGTH_IN_STRING = ['4k', '8k', '16k', '32k', '64k', '128k', '256k', '512k', '1M']
TASKS =… See the full description on the dataset page: https://huggingface.co/datasets/self-long/RULER-llama3-1M.latenet-v0-activations-llama3.1-70b-base
meta-llama/Llama-3.1-70B — Activation Dataset
Cached activations extracted from meta-llama/Llama-3.1-70B (revision 349b2ddb53ce8f2849a6c168a81980ab25258dac).
Full-sequence activations (80 layers, 8192 dim, float16, all tokens) from meta-llama/Llama-3.1-70B (base) on 23724 LateNet v0 statements (affirmative + negated). Extracted via NDIF. Raw statements only (no chat template). Prompts ordered by negated→generator→pair_id for contiguous domain shards.
Contents… See the full description on the dataset page: https://huggingface.co/datasets/alliedtoasters/latenet-v0-activations-llama3.1-70b-base.got-activations-llama3.1-405b-base
meta-llama/Llama-3.1-405B — Activation Dataset
Cached activations extracted from meta-llama/Llama-3.1-405B (revision unknown).
Contents
Tensor
Layers
Dim
Pooling
Shards
Row Bytes
hidden_layers
0-125
16384
-
12
-
Prompts: 7660
Format version: 1.1
Load with lmprobe
from lmprobe import pull_dataset, load_activation_dataset
# Option 1: Pull into local cache (enables probe training without re-extraction)… See the full description on the dataset page: https://huggingface.co/datasets/latent-lab/got-activations-llama3.1-405b-base.latenet-v0-activations-llama3.1-405b-base
meta-llama/Llama-3.1-405B — Activation Dataset
Cached activations extracted from meta-llama/Llama-3.1-405B (revision b906e4dc842aa489c962f9db26554dcfdde901fe).
LateNet v0 activations for Llama 3.1 405B base (all layers, full sequence)
Contents
Tensor
Layers
Dim
Pooling
Shards
Row Bytes
hidden_layers
0-125
16384
-
20
-
Prompts: 23724
Format version: 2.0
Load with lmprobe
from lmprobe import load_activations, Probe
acts =… See the full description on the dataset page: https://huggingface.co/datasets/alliedtoasters/latenet-v0-activations-llama3.1-405b-base.Llama-3-Magpie-Pro-1M-v0.1
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Llama-3-Magpie-Pro-1M-v0.1.llama-3b-gold-15M-student-generations_SNIS_2048_tune422v1lm-eval-results-hkust-nlp-dart-math-llama3-8b-prop2diff-private
Dataset Card for Evaluation run of hkust-nlp/dart-math-llama3-8b-prop2diff
Dataset automatically created during the evaluation run of model hkust-nlp/dart-math-llama3-8b-prop2diff
The dataset is composed of 62 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 4 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/nyu-dice-lab/lm-eval-results-hkust-nlp-dart-math-llama3-8b-prop2diff-private.llama-3b-gold-15M-student-generations_PRESAMPLING_2048_tune422v1Llama-3-Magpie-Air-3M-v0.1
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Llama-3-Magpie-Air-3M-v0.1.fineweb-edu-pretokenized-llama3-100b
FineWeb-Edu Pretokenized with Llama 3.1 (100B)
This repository contains the sample/100BT subset of FineWeb-Edu, pretokenized for Megatron-LM-style training with meta-llama/Meta-Llama-3.1-8B.
It is a derived, pretokenized version of the upstream data, not an official Hugging Face FineWeb release.
Dataset summary
140 indexed shards
97,270,686 non-empty documents
97,458,793,013 tokens
English web text from FineWeb-Edu sample/100BT
Source dataset revision:… See the full description on the dataset page: https://huggingface.co/datasets/ZhuofengLi/fineweb-edu-pretokenized-llama3-100b.llama-3.2-3b-atlas
llama-3.2-3b-atlas
needle-llama3-16x524kgot-activations-llama3.1-70b-base
meta-llama/Llama-3.1-70B — Activation Dataset
Cached activations extracted from meta-llama/Llama-3.1-70B (revision 349b2ddb53ce8f2849a6c168a81980ab25258dac).
Geometry of Truth curated dataset activations for Llama 3.1 70B base
Contents
Tensor
Layers
Dim
Pooling
Shards
Row Bytes
hidden_layers
0-79
8192
-
4
-
Prompts: 7660
Format version: 2.0
Load with lmprobe
from lmprobe import load_activations, Probe
acts =… See the full description on the dataset page: https://huggingface.co/datasets/alliedtoasters/got-activations-llama3.1-70b-base.llama3.2-3b-instruct-atlas
juiceb0xc0de/llama3.2-3b-instruct-atlas
A brain atlas for meta-llama/Llama-3.2-3B-Instruct, the 3B instruction-tuned member of the Llama 3.2 family. This is not a chat dataset or a benchmark - it is an internal-mechanics map built by running activations through a corpus of prompts and scoring what each layer, component, head, and feature direction is doing.
If you want to know where an instruction-tuned model keeps its register machinery, which directions survive a causal test… See the full description on the dataset page: https://huggingface.co/datasets/juiceb0xc0de/llama3.2-3b-instruct-atlas.openthoughts3_100k_llama3_eval_5554
mlfoundations-dev/openthoughts3_100k_llama3_eval_5554
Precomputed model outputs for evaluation.
Evaluation Results
Summary
Metric
AIME24
AMC23
MATH500
MMLUPro
JEEBench
GPQADiamond
LiveCodeBench
CodeElo
CodeForces
HLE
HMMT
AIME25
LiveCodeBenchv5
Accuracy
37.0
75.2
83.8
11.2
45.2
45.1
44.4
13.8
18.3
9.7
19.3
30.3
31.9
AIME24
Average Accuracy: 37.00% ± 1.29%
Number of Runs: 10
Run
Accuracy
Questions Solved
Total Questions… See the full description on the dataset page: https://huggingface.co/datasets/mlfoundations-dev/openthoughts3_100k_llama3_eval_5554.Codeforces-Python_Submissions_reformatted_deduped_llama3.3_nomic_fdetails_Omartificial-Intelligence-Space__al-baka-llama3-8b-experimental
Dataset Card for Evaluation run of Omartificial-Intelligence-Space/al-baka-llama3-8b-experimental
Dataset automatically created during the evaluation run of model Omartificial-Intelligence-Space/al-baka-llama3-8b-experimental.
The dataset is composed of 136 configuration, each one coresponding 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… See the full description on the dataset page: https://huggingface.co/datasets/OALL/details_Omartificial-Intelligence-Space__al-baka-llama3-8b-experimental.details_MaziyarPanahi__calme-2.3-llama3-70b
Dataset Card for Evaluation run of MaziyarPanahi/calme-2.3-llama3-70b
Dataset automatically created during the evaluation run of model MaziyarPanahi/calme-2.3-llama3-70b.
The dataset is composed of 136 configuration, each one coresponding 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/OALL/details_MaziyarPanahi__calme-2.3-llama3-70b.luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-pausefilled
luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-pausefilled
Every non-first chunk of every document carries a thought: the gpt-5.6-luna reasoning thought
where one was generated, and a content-free pause thought everywhere else.
luna chunk <|reserved_special_token_1|> luna reasoning <|reserved_special_token_2|>
filler chunk <|reserved_special_token_1|> 256x <|reserved_special_token_0|> <|reserved_special_token_2|>
The filler is 258 tokens. Chunk 0 is excluded… See the full description on the dataset page: https://huggingface.co/datasets/JackHsieh/luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-pausefilled.llama3.2-1b-instruct-atlas
juiceb0xc0de/llama3.2-1b-instruct-atlas
A brain atlas for meta-llama/Llama-3.2-1B-Instruct, the 1B instruction-tuned member of the Llama 3.2 family. This is not a chat dataset or a benchmark - it is an internal-mechanics map built by running activations through a corpus of prompts and scoring what each layer, component, head, and feature direction is doing.
If you want to know where a small instruction-tuned model keeps its register machinery, which directions survive a causal… See the full description on the dataset page: https://huggingface.co/datasets/juiceb0xc0de/llama3.2-1b-instruct-atlas.llama3_aug_math_train_n45Llama-3.2-1B-Instruct-best-of-N-completionsMagpie-Reasoning-V2-250K-CoT-Llama3
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Llama3.Llama-3.3-70B-Instruct-eval-logs-and-scoresopen-hermes-2.5-sft-mixture-llama3-inference-retrieval-tokensllama-3.2-3B-f1-instruct-eval-logs-and-scoresLlama-3.1-8B-Instruct-eval-logs-and-scores