Paper Citation Record · LEDGER

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention

As of 7 August 2026, this Paper Citation Record lists 36 of 36 outbound references and 1 inbound Pith citation observation for arXiv:2607.04422.

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

pith.paper-citation-record.v1
2607.04422 v1

Coverage vector

measured 36 of 36 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-07-11T19:17:59.044982Z

measured 37 of 37 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-07T06:34:17.273281+00:00

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-02T07:54:47.738345Z

measured 0 of 1 external citation measurements

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

Source: cited_works

Reference resolution

36 of 36 outbound references displayed

verified exact0
verified fuzzy0
unresolved36
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 65b4cc52-d890-40c6-aeeb-5c56bf7879d1 · outbound

This paper cites Pretraining large language models with nvfp4.arXiv preprint arXiv:2509.25149, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Pretraining large language models with nvfp4.arXiv preprint arXiv:2509.25149, 2025

Reference 1

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5eedd8b1-b69f-4922-9e04-185ae5971f29 · outbound

This paper cites FP4 All the Way: Fully Quantized Training of LLMs.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention FP4 All the Way: Fully Quantized Training of LLMs

Reference 2

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a99b5e2d-9dba-43e2-b6a3-0c365219b1a3 · outbound

This paper cites Quartet: Native fp4 training can be optimal for large language models.arXiv preprint arXiv:2505.14669, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Quartet: Native fp4 training can be optimal for large language models.arXiv preprint arXiv:2505.14669, 2025

Reference 3

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation dffadbd0-1854-48cf-a090-cb6f8beef9c5 · outbound

This paper cites Dissecting outlier dynamics in llm nvfp4 pretraining.arXiv preprint arXiv:2602.02047, 2026.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Dissecting outlier dynamics in llm nvfp4 pretraining.arXiv preprint arXiv:2602.02047, 2026

Reference 4

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2769a908-fc6f-4090-a6d5-de49e1ea5101 · outbound

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

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling

Reference 5

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2f1d7d13-1cc6-4bf0-9599-bfa8ccffa5f2 · outbound

This paper cites Oscillation-Reduced MXFP4 Training for Vision Transformers.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Oscillation-Reduced MXFP4 Training for Vision Transformers

Reference 6

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8a4fd6f2-b7e6-4190-ab69-55c3bbc91a83 · outbound

This paper cites TetraJet-v2: Accurate NVFP4 Training for Large Language Models with Oscillation Suppression and Outlier Control.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention TetraJet-v2: Accurate NVFP4 Training for Large Language Models with Oscillation Suppression and Outlier Control

Reference 7

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 65ef24b8-2162-409c-ac6a-ae944a910e1e · outbound

This paper cites Muon: An optimizer for hidden layers in neural networks, 2024.URL https://kellerjordan.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Muon: An optimizer for hidden layers in neural networks, 2024.URL https://kellerjordan

Reference 8

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b5126e53-ea5b-4606-b6dc-41b2084c5b5e · outbound

This paper cites Muon is Scalable for LLM Training.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Muon is Scalable for LLM Training

Reference 9

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0aad755f-888b-4fdc-b8ad-01d0667bc3ef · outbound

This paper cites Deepseek-v4: Towards highly efficient million-token context intelligence, 2026.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Deepseek-v4: Towards highly efficient million-token context intelligence, 2026

Reference 10

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a6835d34-790e-4fd4-aaf9-d1a499eb48d8 · outbound

This paper cites Achieving low-bit muon through subspace preservation and grid quantization.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Achieving low-bit muon through subspace preservation and grid quantization

Reference 11

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c97efd93-6248-4447-8d7c-0cb980062dcc · outbound

This paper cites COAT: Compressing Optimizer states and Activation for Memory-Efficient FP8 Training.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention COAT: Compressing Optimizer states and Activation for Memory-Efficient FP8 Training

Reference 12

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 76ea06bc-d00c-4ba5-8881-2b28fc0bfb6b · outbound

This paper cites Metis: Training llms with fp4 quantization.arXiv preprint arXiv:2509.00404, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Metis: Training llms with fp4 quantization.arXiv preprint arXiv:2509.00404, 2025

Reference 13

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ce1da27d-e3c0-492a-8234-fe20ed290345 · outbound

This paper cites Attn-qat: 4-bit attention with quantization-aware training.arXiv preprint arXiv:2603.00040, 2026.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Attn-qat: 4-bit attention with quantization-aware training.arXiv preprint arXiv:2603.00040, 2026

Reference 14

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 16c2fa9e-1c31-4668-8a90-5e9e1d80248f · outbound

This paper cites Sageattention3: Microscaling fp4 attention for inference and an exploration of 8-bit training.arXiv preprint arXiv:2505.11594, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Sageattention3: Microscaling fp4 attention for inference and an exploration of 8-bit training.arXiv preprint arXiv:2505.11594, 2025

Reference 15

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 024a417c-01ec-4800-92ba-301d29e9da18 · outbound

This paper cites Microscaling Data Formats for Deep Learning.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Microscaling Data Formats for Deep Learning

Reference 16

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cc3a4a6b-d7e6-4fc0-9f54-c13330b1defd · outbound

This paper cites Root: Robust orthogonalized optimizer for neural network training.arXiv preprint arXiv:2511.20626, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Root: Robust orthogonalized optimizer for neural network training.arXiv preprint arXiv:2511.20626, 2025

Reference 17

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f7fe30bb-a00a-4b95-b865-ecdf7acda074 · outbound

This paper cites Flashattention: Fast and memory-efficient exact attention with io-awareness.Advances in neural information processing systems, 35:16344–16359, 2022.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Flashattention: Fast and memory-efficient exact attention with io-awareness.Advances in neural information processing systems, 35:16344–16359, 2022

Reference 18

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 03d758b3-badb-4f9e-a64b-49d73b40d7aa · outbound

This paper cites FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning

Reference 19

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1738a3a2-3203-4f81-bbd8-8603df924636 · outbound

This paper cites Training LLMs with MXFP4.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Training LLMs with MXFP4

Reference 20

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c7d18e4a-1378-4cfe-b4b7-a315d3b3bcf1 · outbound

This paper cites Svdquant: Absorbing outliers by low-rank components for 4-bit diffusion models.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Svdquant: Absorbing outliers by low-rank components for 4-bit diffusion models

Reference 21

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 650622c4-0d22-4628-b157-63549d6046ea · outbound

This paper cites Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions.SIAM review, 53 (2):217–288, 2011.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Finding structure with randomness: Probabilistic algorithms for constructing approximate matrix decompositions.SIAM review, 53 (2):217–288, 2011

Reference 22

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 933fef17-26ff-4e69-97d7-030e7e2ccaac · outbound

This paper cites NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Model.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Model

Reference 23

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 887afe75-4300-4d96-b5b4-b539a9052f90 · outbound

This paper cites Training and inference with integers in deep neural networks.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Training and inference with integers in deep neural networks

Reference 24

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7a3d4e70-c129-44b3-808f-0f95ebaca6b3 · outbound

This paper cites FP8 Formats for Deep Learning.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention FP8 Formats for Deep Learning

Reference 25

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation dd200e99-2ae9-4764-8e63-308886d09899 · outbound

This paper cites Stable and low-precision training for large-scale vision-language models.Advances in Neural Information Processing Systems, 36:10271–10298, 2023.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Stable and low-precision training for large-scale vision-language models.Advances in Neural Information Processing Systems, 36:10271–10298, 2023

Reference 26

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7ae5b301-63ae-49e1-8d4e-0c59f242f88d · outbound

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

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Scaling FP8 training to trillion-token LLMs

Reference 27

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e631cf3e-e3d2-4b17-9a83-f88810c66b60 · outbound

This paper cites DeepSeek-V3 Technical Report.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention DeepSeek-V3 Technical Report

Reference 28

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 51b48e9c-9ca3-4ca6-ad5c-940bf0d1d5e8 · outbound

This paper cites Quarot: Outlier-free 4-bit inference in rotated llms.Advances in Neural Information Processing Systems, 37:100213–100240, 2024.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Quarot: Outlier-free 4-bit inference in rotated llms.Advances in Neural Information Processing Systems, 37:100213–100240, 2024

Reference 29

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3be8d1c5-0b54-4f28-bbec-c9ef66a52809 · outbound

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

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Optimizing Large Language Model Training Using FP4 Quantization

Reference 30

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0814f2e2-e831-447a-8c90-bd2e633f1f36 · outbound

This paper cites 8-bit Optimizers via Block-wise Quantization.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention 8-bit Optimizers via Block-wise Quantization

Reference 31

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2ba6c902-29f3-4a49-8a28-45ee412a64a5 · outbound

This paper cites Memory efficient optimizers with 4-bit states.Advances in Neural Information Processing Systems, 36:15136–15171, 2023.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Memory efficient optimizers with 4-bit states.Advances in Neural Information Processing Systems, 36:15136–15171, 2023

Reference 32

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bdd81a36-5f49-4ab0-8405-cda1554772e6 · outbound

This paper cites Effective quantization of muon optimizer states.arXiv preprint arXiv:2509.23106, 2025.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Effective quantization of muon optimizer states.arXiv preprint arXiv:2509.23106, 2025

Reference 33

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 80a5171d-09fe-418f-be6c-8cc164953c46 · outbound

This paper cites Sageat- tention: Accurate 8-bit attention for plug-and-play inference acceleration.arXiv preprint arXiv:2410.02367, 2024.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Sageat- tention: Accurate 8-bit attention for plug-and-play inference acceleration.arXiv preprint arXiv:2410.02367, 2024

Reference 34

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b97251a2-4038-489e-aebc-7aae3b804ccc · outbound

This paper cites Sageat- tention2: Efficient attention with thorough outlier smoothing and per-thread int4 quantization.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Sageat- tention2: Efficient attention with thorough outlier smoothing and per-thread int4 quantization

Reference 35

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 957beb04-0120-4da7-8ac8-a7848030d5d1 · outbound

This paper cites variance.

Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention variance

Reference 36

Resolution: unresolved; no resolver link, observed 2026-07-11T19:17:59.044982Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation dd01480b-b33a-40e7-a3e2-bf042a043c86 · inbound

Reference Traces for Auditing Invisible Weight Updates and Guiding Exact-Budget Protection cites this paper.

Reference Traces for Auditing Invisible Weight Updates and Guiding Exact-Budget Protection Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention

Reference 83

Resolution: unresolved; no resolver link, observed 2026-08-02T07:54:47.738345Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.