Unverified50% confidenceFactExact time
Wanda is a post-training pruning method that requires no retraining, determining which parameters can be zeroed out by combining weight magnitude with statistical information from input activations.
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAccelerated LoRA is optimized during training for a specific sampler scheduling strategy and model weights; when the actual sampler's step size distribution, noise schedule, or model quantization method is inconsistent with the training conditions, the shortcut paths learned by the LoRA may fail entirely.63% similarUnverifiedMethods such as freezing LLM backbone parameters during the visual alignment training phase, staged training, data ratio replay, and LoRA can mitigate catastrophic forgetting in multimodal training.62% similarVerifiedParameter-efficient fine-tuning (PEFT) methods such as LoRA and Adapter can compress the number of trainable parameters to less than 1% of the total parameters.62% similarUnverifiedModern training frameworks use mixed-precision training and gradient scaling (loss scaling) techniques to prevent small values from being swallowed by amplifying small gradients into the representable range.62% similarUnverifiedOn infilling tasks, Diffusion-LM outperforms continuous-relaxation-based methods such as COLD and DELOREAN on automatic evaluation metrics, and reaches the level of autoregressive models trained from scratch specifically for this purpose without requiring targeted training.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/949158API
curl https://kongchang.com/api/v1/knowledge/claims/949158MCP
get_claim(id=949158)