Paper Citation Record · LEDGER

Recipes for Pre-training LLMs with MXFP8

As of 11 August 2026, this Paper Citation Record lists 40 of 40 outbound references and 9 inbound Pith citation observations for arXiv:2506.08027.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2506.08027 v2

Coverage vector

measured 40 of 40 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-07T12:12:47.881130Z

measured 49 of 49 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-11T06:34:44.6726+00:00

measured 9 of 9 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-05T10:54:30.832174Z

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: arxiv_reference, observed 2026-07-01T20:26:13.675967Z

Reference resolution

40 of 40 outbound references displayed

verified exact0
verified fuzzy8
unresolved30
parse uncertain0
malformed identifier1
metadata mismatch1

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation e77c830a-1820-49fd-ae46-2e13d579e519 · outbound

This paper cites Ocp microscaling (mx) specification.

Recipes for Pre-training LLMs with MXFP8 Ocp microscaling (mx) specification

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.996595Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation d2778806-4b3d-4b72-8898-02fb0cef4683 · outbound

This paper cites URL https://resources.nvidia.com/ en-us-blackwell-architecture.

Recipes for Pre-training LLMs with MXFP8 URL https://resources.nvidia.com/ en-us-blackwell-architecture

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.848468Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation ea954705-fd04-42aa-92d2-feba37489914 · outbound

This paper cites Microscaling Data Formats for Deep Learning.

Recipes for Pre-training LLMs with MXFP8 Microscaling Data Formats for Deep Learning

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:45.790981Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c91acf53-9320-43cb-890b-4de84268977c · outbound

This paper cites With Shared Microexponents, A Little Shifting Goes a Long Way.

Recipes for Pre-training LLMs with MXFP8 With Shared Microexponents, A Little Shifting Goes a Long Way

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.032229Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 978acedf-b124-4ddd-9747-37dc484cf5b0 · outbound

This paper cites VS-Quant: Per-vector Scaled Quantization for Accurate Low-Precision Neural Network Inference.

Recipes for Pre-training LLMs with MXFP8 VS-Quant: Per-vector Scaled Quantization for Accurate Low-Precision Neural Network Inference

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.101980Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5004dd90-74fe-4df4-9bd4-241d6565ebf4 · outbound

This paper cites IEEE Std 754-2008 , pages 1–70, 2008.

Recipes for Pre-training LLMs with MXFP8 IEEE Std 754-2008 , pages 1–70, 2008

Reference 6

Resolution: metadata mismatch; raw_fallback, observed 2026-08-07T12:12:48.576575Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 97beda48-3844-43fc-b362-5c5cab22791b · outbound

This paper cites FP8 Formats for Deep Learning.

Recipes for Pre-training LLMs with MXFP8 FP8 Formats for Deep Learning

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.118074Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f0aed45f-126b-44e2-9d42-acb6f5b6ea54 · outbound

This paper cites Nemotron-4 15B Technical Report.

Recipes for Pre-training LLMs with MXFP8 Nemotron-4 15B Technical Report

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.131912Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b5193df3-7e9b-400a-9b83-9666c5fe7341 · outbound

This paper cites Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models.

Recipes for Pre-training LLMs with MXFP8 Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.195871Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c911db37-3acf-4e46-ad04-4fe11bdbcc4d · outbound

This paper cites Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism.

Recipes for Pre-training LLMs with MXFP8 Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.269897Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 13320e44-8583-430e-80bf-3fb3072fbb25 · outbound

This paper cites Measuring Massive Multitask Language Understanding.

Recipes for Pre-training LLMs with MXFP8 Measuring Massive Multitask Language Understanding

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.342345Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d51b2a97-8a5a-4e00-acf8-e89e75597ef9 · outbound

This paper cites Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge.

Recipes for Pre-training LLMs with MXFP8 Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.423646Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aa12abb8-26b2-4213-9838-7d379e44d8e6 · outbound

This paper cites RACE: Large-scale ReAding Comprehension Dataset From Examinations.

Recipes for Pre-training LLMs with MXFP8 RACE: Large-scale ReAding Comprehension Dataset From Examinations

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.493873Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation eb1c1a33-5e07-4ef4-b502-57f205bd6a5f · outbound

This paper cites PIQA: Reasoning about Physical Commonsense in Natural Language.

Recipes for Pre-training LLMs with MXFP8 PIQA: Reasoning about Physical Commonsense in Natural Language

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.582966Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 576341e0-8c74-421d-84d1-797c1d9cb0b9 · outbound

This paper cites Winogrande: An adversarial winograd schema challenge at scale, 2019.

Recipes for Pre-training LLMs with MXFP8 Winogrande: An adversarial winograd schema challenge at scale, 2019

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.617897Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ac0d46a1-cc6d-418c-a915-b00240ae8c9c · outbound

This paper cites HellaSwag: Can a Machine Really Finish Your Sentence?.

Recipes for Pre-training LLMs with MXFP8 HellaSwag: Can a Machine Really Finish Your Sentence?

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.646932Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8158c597-b84e-42aa-b16f-721c584852f3 · outbound

This paper cites Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering.

Recipes for Pre-training LLMs with MXFP8 Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.664591Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 47481e3d-4ead-439e-b2a9-1a932ff3223a · outbound

This paper cites Socialiqa: Com- monsense reasoning about social interactions, 2019.

Recipes for Pre-training LLMs with MXFP8 Socialiqa: Com- monsense reasoning about social interactions, 2019

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.697169Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 27417057-aa1c-4011-b874-ef31b5cef6b2 · outbound

This paper cites CommonsenseQA: A question answering challenge targeting commonsense knowledge.

Recipes for Pre-training LLMs with MXFP8 CommonsenseQA: A question answering challenge targeting commonsense knowledge

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.703553Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4324b597-0cae-48ec-97f6-5ad225c24c3c · outbound

This paper cites The Llama 3 Herd of Models.

Recipes for Pre-training LLMs with MXFP8 The Llama 3 Herd of Models

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.730511Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8c86409a-67f0-46ea-a62c-353d2b508dcc · outbound

This paper cites 8-bit numerical formats for deep neural networks, 2022.

Recipes for Pre-training LLMs with MXFP8 8-bit numerical formats for deep neural networks, 2022

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.576743Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation c7d49f70-6b50-4e37-9bf3-15da04787a62 · outbound

This paper cites Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Haowei Zhang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Li, Hui Qu, J.

Recipes for Pre-training LLMs with MXFP8 Zhang, Han Bao, Hanwei Xu, Haocheng Wang, Haowei Zhang, Honghui Ding, Huajian Xin, Huazuo Gao, Hui Li, Hui Qu, J

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.754388Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8abe41fd-7c90-4bf4-af27-f82a37759078 · outbound

This paper cites Ocp 8-bit floating point specification (ofp8).

Recipes for Pre-training LLMs with MXFP8 Ocp 8-bit floating point specification (ofp8)

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.395984Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 073bdb9f-4131-4710-9eb0-861788e9ff79 · outbound

This paper cites Smoothquant: Accurate and efficient post-training quantization for large language models,.

Recipes for Pre-training LLMs with MXFP8 Smoothquant: Accurate and efficient post-training quantization for large language models,

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.253153Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 97cacf03-4797-4ebe-bce0-b18e0910712d · outbound

This paper cites QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving.

Recipes for Pre-training LLMs with MXFP8 QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.074402Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 28f760c5-22f8-4eb8-bbd2-b581dedb1e4c · outbound

This paper cites GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers.

Recipes for Pre-training LLMs with MXFP8 GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.159556Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5a0ae3d2-6aab-4a30-8b62-7852f67537a9 · outbound

This paper cites AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration.

Recipes for Pre-training LLMs with MXFP8 AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.221979Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6873f7de-f463-4129-8d2a-ad77c575243b · outbound

This paper cites Scaling FP8 training to trillion-token LLMs.

Recipes for Pre-training LLMs with MXFP8 Scaling FP8 training to trillion-token LLMs

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.282884Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b808aec9-2b83-4fe6-b3cd-43066f20400f · outbound

This paper cites The llama 4 herd: The beginning of a new era of natively multimodal ai innova- tion.

Recipes for Pre-training LLMs with MXFP8 The llama 4 herd: The beginning of a new era of natively multimodal ai innova- tion

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:49.124436Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 2ad6dd0c-2c4d-450a-8a9c-19ee36aabc99 · outbound

This paper cites Training LLMs with MXFP4.

Recipes for Pre-training LLMs with MXFP8 Training LLMs with MXFP4

Reference 30

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.436064Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation faf32e08-be52-45c7-a1cb-f11c842a2909 · outbound

This paper cites Optimizing Large Language Model Training Using FP4 Quantization.

Recipes for Pre-training LLMs with MXFP8 Optimizing Large Language Model Training Using FP4 Quantization

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.457958Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 40625447-23d6-4b9f-b3ca-81594a08b54c · outbound

This paper cites Transformer engine.

Recipes for Pre-training LLMs with MXFP8 Transformer engine

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T12:12:48.996727Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 0f338323-0bc3-41c1-b8b8-2a77fa8253e5 · outbound

This paper cites DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models.

Recipes for Pre-training LLMs with MXFP8 DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.507189Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cefd3c8d-4ce7-45c9-99ff-acb5919ea260 · outbound

This paper cites Understanding Warmup-Stable-Decay Learning Rates: A River Valley Loss Landscape Perspective.

Recipes for Pre-training LLMs with MXFP8 Understanding Warmup-Stable-Decay Learning Rates: A River Valley Loss Landscape Perspective

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.569722Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation caef3419-ee7f-44e3-ac4c-c9d34222be05 · outbound

This paper cites Nemotron-4 340B Technical Report.

Recipes for Pre-training LLMs with MXFP8 Nemotron-4 340B Technical Report

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.627819Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation be684594-0e13-4231-98fc-0ab4833e81eb · outbound

This paper cites an unresolved cited work.

Recipes for Pre-training LLMs with MXFP8 Unresolved cited work

Reference 38

Resolution: unresolved; raw_fallback, observed 2026-08-07T12:12:48.896269Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 802c0213-cdba-4429-bcf3-8cbef9468f7f · outbound

This paper cites an unresolved cited work.

Recipes for Pre-training LLMs with MXFP8 Unresolved cited work

Reference 39

Resolution: unresolved; raw_fallback, observed 2026-08-07T12:12:48.785664Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 91e9d15e-83dd-480e-89f7-a4fca49d9d1b · outbound

This paper cites By construction amax/destmax never exceeds 2127 (which is the largest value representable in UE8M0) with FP8, FP6 or FP4 formats.

Recipes for Pre-training LLMs with MXFP8 By construction amax/destmax never exceeds 2127 (which is the largest value representable in UE8M0) with FP8, FP6 or FP4 formats

Reference 40

Resolution: malformed identifier; raw_fallback, observed 2026-08-07T12:12:48.698298Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation ff2704d2-c5dc-4343-a25d-6be97f90f558 · outbound

This paper cites SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models.

Recipes for Pre-training LLMs with MXFP8 SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:47.016932Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 80d5696e-8679-4c52-a166-52d569aefcda · outbound

This paper cites DeepSeek-V3 Technical Report.

Recipes for Pre-training LLMs with MXFP8 DeepSeek-V3 Technical Report

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-07T12:12:46.816128Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation d238393d-c84c-4717-8e97-4a6fc1071923 · inbound

A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models cites this paper.

A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models Recipes for Pre-training LLMs with MXFP8

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-04T14:51:29.915567Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 886ff8dc-5d2d-408b-bfc2-f9ebd0f55fcf · inbound

Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling cites this paper.

Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling Recipes for Pre-training LLMs with MXFP8

Reference 5

Resolution: verified exact; arxiv_id, observed 2026-05-17T02:23:52.769839Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 5bc29d3f-438b-4a33-b922-6dfd36cb0907 · inbound

StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models cites this paper.

StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models Recipes for Pre-training LLMs with MXFP8

Reference 27

Resolution: verified exact; arxiv_id, observed 2026-05-10T12:15:22.229993Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation a50e61fd-5ca4-4c24-8e7c-b3e0301cdfbb · inbound

OSP-Next: Efficient High-Quality Video Generation with Sparse Sequence Parallelism, HiF8 Quantization, and Reinforcement Learning cites this paper.

OSP-Next: Efficient High-Quality Video Generation with Sparse Sequence Parallelism, HiF8 Quantization, and Reinforcement Learning Recipes for Pre-training LLMs with MXFP8

Reference 21

Resolution: verified exact; arxiv_id, observed 2026-06-29T13:13:27.553824Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation 81579add-036b-46f1-b340-2027e20a9d29 · inbound

Stochastic Rounding Increases Small Singular Values cites this paper.

Stochastic Rounding Increases Small Singular Values Recipes for Pre-training LLMs with MXFP8

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-07-01T20:26:13.679444Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-11T06:34:44.6726+00:00.

Observation cc3f4a9d-f5f0-4d3c-9c6d-a1906fda7d6a · inbound

SOAP, Muon, and Beyond: Pushing LLM Pretraining Scales cites this paper.

SOAP, Muon, and Beyond: Pushing LLM Pretraining Scales Recipes for Pre-training LLMs with MXFP8

Reference 73

Resolution: unresolved; no resolver link, observed 2026-08-02T06:45:18.277482Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 35ce8b5d-d972-4296-9c31-5c3f40a2dbc8 · inbound

ACRL: Adaptive Control of Training-Inference Discrepancy for Stable Reinforcement Learning cites this paper.

ACRL: Adaptive Control of Training-Inference Discrepancy for Stable Reinforcement Learning Recipes for Pre-training LLMs with MXFP8

Reference 26

Resolution: unresolved; no resolver link, observed 2026-07-31T23:13:13.763909Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6bc22111-1ccd-450d-9be1-e93ba5b19779 · inbound

Stable FP4 Training via Transposition-Invariant Block Quantization cites this paper.

Stable FP4 Training via Transposition-Invariant Block Quantization Recipes for Pre-training LLMs with MXFP8

Reference 13

Resolution: unresolved; no resolver link, observed 2026-07-31T05:04:00.365982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4524645d-eafa-4379-bd0b-c0b96d1facf0 · inbound

Heterogeneity-Aware Microscaling for Efficient Low-Bit LLM Inference cites this paper.

Heterogeneity-Aware Microscaling for Efficient Low-Bit LLM Inference Recipes for Pre-training LLMs with MXFP8

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-05T10:54:30.832174Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.