# TrainSDC: Characterizing and Mitigating Silent Data Corruption in Large Language Model Training

arXiv: https://arxiv.org/abs/2608.30769 · HTML: https://mlsanity.com/paper/2608.30769/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 17 references that are in the map.
- First posted on arXiv 31 Aug 2026.

## Builds on (most cited first, 10 of 10)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model](https://mlsanity.com/paper/2502.02737/) (2025)
- [A Study of BFLOAT16 for Deep Learning Training](https://mlsanity.com/paper/1905.12322/) (2019)
- [Silent Data Corruptions at Scale](https://mlsanity.com/paper/2102.11245/) (2021)
- [Robust LLM Training Infrastructure at ByteDance](https://mlsanity.com/paper/2509.16293/) (2025)
- [Understanding Silent Data Corruption in LLM Training](https://mlsanity.com/paper/2502.12340/) (2025)
- [ATTNChecker: Highly-Optimized Fault Tolerant Attention for Large Language Model Training](https://mlsanity.com/paper/2410.11720/) (2024)
- [Exploring Silent Data Corruption as a Reliability Challenge in LLM Training](https://mlsanity.com/paper/2604.00726/) (2026)
- [LLM-PRISM: Characterizing Silent Data Corruption from Permanent GPU Faults in LLM Training](https://mlsanity.com/paper/2604.10390/) (2026)

## Abstract (arXiv)

LLM training is increasingly vulnerable to silent data corruption (SDC), yet existing protection methods largely treat Transformer computations uniformly because their vulnerability remains poorly understood. We present the first systematic characterization of SDC vulnerability across major computation interfaces in both the forward and backward passes of Transformer training. Our analysis reveals two distinct error propagation mechanisms: forward-pass vulnerability is highly location dependent, with faults on the Q/K path producing persistent training deviations, whereas backward-pass vulnerability is largely governed by gradient exponent distributions rather than computation locations. Motivated by these observations, we propose TrainSDC, a characterization-guided protection framework consisting of Q/K-path recomputation, residual-gain monitoring, and exponent-aware gradient scaling. Experiments on Llama 3.2-1B and Qwen3-0.6B show that TrainSDC maintains training behavior close to fault-free execution under both sparse and dense fault injection while introducing only 1.65%-6.76% runtime overhead.