THE INTERNET IS THINKING
Earlier sources: Language models
Language models learn patterns in text and other representations to predict or generate sequences. Follow model releases, training methods, reasoning, evaluation, and deployment research; a benchmark result is specific to its test conditions.
SYMTRI organizes observations from public sources into a living map. Connections reflect shared subject matter; they do not establish causation or endorse a source.
Open in the map ↗446 classified signals in the fourteen-day window. These counts include signals awaiting readable evidence.
Earlier sources
Up to ten sources with readable evidence, ordered by their feed timestamp. These archive sources can be older than the live fourteen-day window. This list is a sample, not a measure of total activity. Dates distinguish paper publication, repository creation, model/dataset activity, Hacker News posts, and KEV catalog additions. “First archived” records when SYMTRI first stored the signal.
- arXiv · Published · First archived
Enhancing Diffusion Language Models with Autoregressive Post-Training Weights ↗
Diffusion language models (dLLMs) have emerged as a promising alternative to autoregressive (AR) language models, offering flexible token-update orders and parallel decoding. Recent dLLMs are often initialized from pretrained AR models before diffusion conversion in order to inherit their learned representations. After the conversion, however, they typically ignore the extensive post-training ecosystem of their AR ancestors. In this work, we show that these existing AR post-training weight updates can instead be effectively recycled to enhance diffusion models. Despite the changes by AR-to-dif…
- arXiv · Published · First archived
DirectSpeech2LLM: A Simple End-to-End Framework to Mitigate Prompt Overfitting in Speech-LLMs ↗
Speech-LLMs often exhibit prompt overfitting, where models solely trained on automatic speech recognition (ASR) instruction fail to generalize to new instructions such as speech translation and continue to behave primarily as ASR system. We propose DirectSpeech2LLM, a simple end-to-end framework that preserves the instruction-following ability of the LLM on unseen tasks when conditioned on speech. It computes distance-based CTC loss over the frozen LLM embedding matrix and uses greedy CTC labels to derive geometrically and temporally aligned speech embeddings respectively as an input to the LL…
- Hacker News · Posted on Hacker News · First archived
KI-LLM Has Reached the Pareto Frontier of Recklessness ↗
Hacker News discussion.
- arXiv · Published · First archived
Language Carries the Expert's Impression: Instrument-Anchored LLM Judges Transfer Counseling-Quality Assessment and Beat In-Domain Training ↗
Automatic assessment of communication quality in dyadic counseling conversations is bottlenecked by data: expert-rated corpora are small and expensive to grow. We study cross-domain transfer of expert overall-impression prediction across three German corpora of simulated counseling (two general-practice medical, one school-related parent-teacher; $n=195$ expert-rated sessions, one corpus after scale equating). Training on the other domains beats training in-domain: leave-one-domain-out transfer reaches nested Spearman $ρ= 0.54$ against $\le 0.48$ within the target domain, a paired session-leve…
- OpenAlex · Published · First archived
GP-DHT: A dual-head transformer with supervised contrastive learning for shared-gene pretraining and single-cell GRN prediction ↗
Gene Regulatory Networks (GRNs) are crucial for understanding cell fate decisions and disease mechanisms. However, inferring GRNs from single-cell RNA sequencing remains challenging due to noise, data sparsity, and species-specific distribution variations. We propose GP-DHT (Gene Pair–Dual Head Transformer), a single-cell GRN inference framework that pre-trains gene representations on a unified shared gene space using the BEELINE datasets of humans and mice. It models gene-cell expression relationships through a multi-relational heterograph and predicts transcription factor (TF)-target regulat…
- OpenAlex · Published · First archived
Ontology-Guided Multimodal Large Language Model for Semantic Recognition of Ship Navigation Behaviors ↗
Automatic Identification System (AIS) data provide sequential motion observations for ship-behavior analysis, yet high-level navigational behavioral semantics cannot be reliably interpreted from low-level motion features alone, especially considering ship-environment spatial interactions and their temporal dependencies. To mitigate this limitation, this study proposes an ontology-guided multimodal large language model for semantic recognition of ship navigation behaviors. Built on a three-layer hierarchical behavioral model, the approach unifies motion states, spatial interactions, and tempora…
- OpenAlex · Published · First archived
LLM‐guided semantic‐aware trajectory optimization for UAV path planning using environmental context ↗
Summary Autonomous unmanned aerial vehicle path planning in urban environments requires both geometric safety and semantic awareness. Conventional planners based on fixed costs or predefined risk models struggle to reflect mission‐dependent semantic priorities. We present a large language model (LLM)‐guided semantic‐aware planning framework in which the LLM serves as a bounded high‐level decision module. OpenStreetMap data are transformed into class‐wise semantic potential fields, and a route‐relative context and natural‐language mission description are used to infer validated semantic modulat…
- OpenAlex · Published · First archived
Vision–Language Model (VLM)-Based Interactive Visual Querying for Construction Ergonomic Risk Reasoning ↗
Construction workers are frequently exposed to postural ergonomic risks, yet conventional AI-based ergonomic risk assessment (ERA) systems generally provide risk classifications or scores rather than interpretable and interactive explanations. This study investigates whether domain adaptation can improve a generic vision–language model’s ability to identify and explain construction ergonomic risks from images. A dataset with 1900 construction image–text pairs was curated. The domain-adapted model was compared with the baseline model. Performance was evaluated using visual question answering (V…
- OpenAlex · Published · First archived
PRISM-LoRA: Principal-Direction-Guided Single-LoRA Merging Framework for Continual Learning of Large Language Models ↗
Continual learning of large language models requires incorporating new knowledge while limiting catastrophic forgetting; however, many low-rank adaptation (LoRA)-based methods retain task-specific modules or separate learning spaces that grow with the task sequence. We propose PRISM-LoRA, a replay-free framework that repeatedly trains a single LoRA module, merges its update into the backbone, and reinitializes it for the next task. Unlike LoRA-based continual-learning approaches that retain task-specific modules or allocate separate task-specific learning spaces, PRISM-LoRA uses the directiona…
- OpenAlex · Published · First archived
A New Metacognitive Confidence Calculation Approach Based on Large Language Models for Natural Language Explanations ↗
Traditional methods used to measure metacognitive confidence rely on superficial features such as word count and ambiguity and therefore fail to adequately reflect the level of metacognitive confidence in statements. In this study, in addition to traditional text-based confidence measures, we used a Large Language Model (LLM) to directly analyze confidence and its ratio in statements. LLM-based confidence scores were calculated on statements from the e-SNLI and CoS-E open-source datasets, and the prediction performance obtained with traditional features was compared. The results indicate that…
Live window recorded . This is the ingestion observation time, not the publication or discovery date of each source.
How to assess evidence about Language models
Check each original source for its methods, release status, limitations, and supporting data. A repository announcement, a preprint, and a published study provide different kinds of evidence. SYMTRI’s subject classification helps discovery; it does not independently validate the source’s claims.
See the collection and counting methodology ↗ · Read the weekly synthesis ↗
Related threads
Explore related regions
These navigation suggestions are part of the map’s editorial structure, rather than measured evidence of a connection.