Tag
The article introduces the Contrastive Language Model (CLM), which is 9x faster than Jev for System One models, and provides a guide on using Jev in building custom AI harnesses with Pi SDK.
The article likely presents research on contrastive language models, exploring the use of contrastive learning techniques in language model development.
The tweet shares links to Contrastive Language Models on Hugging Face, highlighting recent updates to models like CLM-v0.1-8B for text ranking and deepswe-clm-heads-8k.
Contrastive World Models propose a new approach for learning latent dynamics without pixel reconstruction, using a contrastive objective to improve robustness and efficiency in visually complex environments for model-based reinforcement learning.
LE4Mob is an inductive, distance-aware, and general-purpose location embedding framework for enhancing human mobility modeling in tasks such as next location prediction and commuter flow generation.
The article introduces Contrastive-LM (CLM), a new language model trained with contrastive learning for connecting states and actions, offering significant improvements in latency and performance on computer-use, gaming, and tool-calling tasks.
This paper introduces contrastive self-distillation to improve reasoning in AI models by separating correctness from behavioral shifts, demonstrating enhanced performance and stability.
This paper proposes Centroid-Guided Contrastive Loss (CGCL) for structured fraudulent job posting detection, unifying classification and clustering in latent space to achieve state-of-the-art performance.
This paper presents a parameter-efficient adaptation framework using LoRA and contrastive learning to enhance vision transformers for automatic target recognition in synthetic aperture sonar imagery.
HERMES is a graph-based framework that constructs personalized knowledge graphs from clinical notes using LLM-guided extraction and contrastive logic modeling, and applies graph attention networks for improved patient outcome prediction, outperforming text-only baselines on MIMIC datasets.
The paper introduces Ovis-Embedding, a state-of-the-art omni-modal embedding model that uses a shared backbone to encode text, image, video, and audio in a common representation space, achieving top performance on benchmarks like MMEB-v3 and MVEB.
AntennaFlow is a three-stage generative flow model framework that jointly addresses phase acquisition and offset correction challenges in antenna testing, enabling fast, phaseless, and offset-vector-free near-field to far-field reconstruction from sparse amplitude-only measurements.
ReMoMask-2 improves text-to-motion generation by embedding retrieval directly into the generator's latent space, eliminating representation gaps and achieving state-of-the-art results on benchmarks like KIT-ML and SnapMoGen.
This paper proposes a zero-shot learning framework for multivariate IoT traffic anomaly detection using adversarial and contrastive learning within a variational autoencoder, enabling domain adaptation without labeled data and demonstrating strong performance across diverse datasets.
This paper empirically studies contrastive pretraining with synthetic semantic supervision for code embeddings in small transformers, showing significant gains over baselines and competitiveness with larger models.
The paper proposes Prior-Guided Tuning (PGT) and Contrastive Prior Steering (CPS) to use natural language priors as auxiliary learning signals, improving low-resource LLM training performance on tasks like AmbiMath, Jigsaw, and MNLI/HANS.
PEARL proposes a path-entity aligned relational learning framework that uses contextual subgraphs and LLM-guided retrieval for inductive knowledge graph completion, achieving state-of-the-art results on benchmarks.
This study proposes a two-stage framework for sentence-level depression symptom recognition using candidate generation and definition-guided verification, achieving best accuracy and F1 scores among evaluated methods.
CORE introduces a distillation method that transfers compositional ranking judgments from a reranker to an embedding model using a Rank-KL objective, enhancing compositional retrieval performance across benchmarks without compromising standard tasks.
The study demonstrates that different representation learning objectives recover distinct latent structures from psychometric data, with contrastive objectives enhancing teacher-child retrieval but PCA-based methods better preserving behavioral phenotype organization.