✨Test-Time Spectrum-Aware Latent Steering for Zero-Shot Generalization in Vision-Language Models

📝 Summary:
VLMs degrade under test-time domain shifts. Spectrum-Aware Test-Time Steering STS is a lightweight method that adapts VLM latent representations by steering them using textual embedding subspaces, without backpropagation. STS surpasses state-of-the-art, offering faster inference and less memory.

🔹 Publication Date: Published on Nov 12

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.09809
• PDF: https://arxiv.org/pdf/2511.09809

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#VisionLanguageModels #ZeroShotGeneralization #DomainAdaptation #DeepLearning #AI

✨An Empirical Study on Preference Tuning Generalization and Diversity Under Domain Shift

📝 Summary:
Preference tuning performance degrades under domain shift. This study found pseudo-labeling adaptation strategies effectively reduce performance degradation in summarization and question-answering tasks across various alignment objectives.

🔹 Publication Date: Published on Jan 9

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.05882
• PDF: https://arxiv.org/pdf/2601.05882
• Github: https://github.com/ckarouzos/prefadap

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#PreferenceTuning #DomainAdaptation #NLP #MachineLearning #AIResearch

✨YaPO: Learnable Sparse Activation Steering Vectors for Domain Adaptation

📝 Summary:
YaPO learns sparse steering vectors for LLMs using Sparse Autoencoders, enabling more effective and stable control than dense methods. This leads to disentangled, interpretable directions for fine-grained alignment across various behaviors, without degrading general knowledge. YaPO offers a gener...

🔹 Publication Date: Published on Jan 13

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.08441
• PDF: https://arxiv.org/pdf/2601.08441
• Project Page: https://mbzuai-paris.github.io/YaPO/
• Github: https://github.com/MBZUAI-Paris/YaPO

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#LLMs #DomainAdaptation #SparseLearning #MachineLearning #AI

✨Mecellem Models: Turkish Models Trained from Scratch and Continually Pre-trained for the Legal Domain

📝 Summary:
Mecellem models are a framework for specialized Turkish legal language models. They feature a scratch-trained encoder achieving top retrieval rankings with efficiency, and a continually pre-trained decoder for legal domain adaptation, reducing legal text perplexity.

🔹 Publication Date: Published on Jan 22

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2601.16018
• PDF: https://arxiv.org/pdf/2601.16018
• Project Page: https://huggingface.co/collections/newmindai/mecellem-models
• Github: https://github.com/newmindai/mecellem-models

🔹 Models citing this paper:
• https://huggingface.co/newmindai/Mursit-Base-TR-Retrieval
• https://huggingface.co/newmindai/Mursit-Base
• https://huggingface.co/newmindai/Mursit-Large-TR-Retrieval

✨ Datasets citing this paper:
• https://huggingface.co/datasets/newmindai/caselaw-retrieval
• https://huggingface.co/datasets/newmindai/contract-retrieval
• https://huggingface.co/datasets/newmindai/regulation-retrieval

✨ Spaces citing this paper:
• https://huggingface.co/spaces/newmindai/Mizan

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#LegalAI #TurkishNLP #LLM #InformationRetrieval #DomainAdaptation

✨FORGE:Fine-grained Multimodal Evaluation for Manufacturing Scenarios

📝 Summary:
FORGE introduces a multimodal manufacturing dataset, revealing that MLLM performance is limited by domain-specific knowledge, not visual grounding. Fine-tuning on FORGEs annotations significantly improves accuracy, offering a path for domain-adapted MLLMs.

🔹 Publication Date: Published on Apr 8

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.07413
• PDF: https://arxiv.org/pdf/2604.07413
• Project Page: https://ai4manufacturing.github.io/forge-web/
• Github: https://github.com/AI4Manufacturing/FORGE

✨ Datasets citing this paper:
• https://huggingface.co/datasets/AI4Manufacturing/forge

==================================

For more data science resources:
✓ https://xn--r1a.website/DataScienceT

#FORGE #MLLM #ManufacturingAI #MultimodalAI #DomainAdaptation