We lifted 1 functions out of this paper's own repositories and ran 0 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| bluorion-com/ZClip | canonical | 0 of 1 |
| Function | Status | Where it lives |
|---|---|---|
| is_fsdp_model | Not yet run | bluorion-com/ZClip/zclip/zclip.py code served (permissive licence) · get_code("0f018b4877545a2d") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Training large language models (LLMs) presents numerous challenges, including gradient instability and loss spikes. These phenomena can lead to catastrophic divergence, requiring costly checkpoint restoration and data batch skipping. Traditional gradient clipping techniques, such as constant or norm-based methods, fail to address these issues effectively due to their reliance on fixed thresholds or heuristics, leading to inefficient learning and requiring frequent manual intervention. In this work, we propose ZClip, an adaptive gradient clipping algorithm that dynamically adjusts the clipping threshold based on statistical properties of gradient norms over time. Unlike prior reactive strategies, ZClip proactively adapts to training dynamics without making any prior assumptions on the scale and the temporal evolution of gradient norms. At its core, it leverages z-score-based anomaly detection to identify and mitigate large gradient spikes, preventing malignant loss spikes while not interfering with convergence otherwise. Our code is available at: https://github.com/bluorion-com/ZClip.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2504.02507")
get_code_for_paper("2504.02507")
have("2504.02507")
Connect an agent — have() is free.