Training & Fine-tuning
A
Showing 1–30 of 118
-
GitHub、「Copilot Fusion」正式リリース。タスクに応じて最適なAIモデル選択で高性能かつ低コストに。GitHub Universe 2026GitHub launches Copilot Fusion, picking the best AI model per taskAt GitHub Universe 2026, held October 28-29 in San Francisco, GitHub announced the general release of GitHub Copilot Fusion. The feature selects the best-suited AI model for each task, aiming for higher performance at lower cost.
-
【開催レポート】Sakana AIシンポジウム：Jürgen Schmidhuberが語るAIのこれまでとこれから、日本から挑む言語モデル・世界モデル・Physical AISakana AI unveils in-house pretrained LLM series at Tokyo symposiumSakana AI held a symposium in Tokyo on Oct 26, 2026, drawing 400+ attendees. Advisor Jürgen Schmidhuber gave the keynote and spoke with CEO David Ha. The company announced it will develop pretrained models in-house, unveiling the 'Sakana Sawara' (tentative) language model series; the small model is planned for spring 2027.
-
AI学習への利用拒否をrobots.txtで一括指定へ OpenAIやGoogleなどが対応表明OpenAI, Google, Microsoft back one robots.txt AI opt-outOpenAI, Google and Microsoft will support a shared way for sites to state AI-training preferences. An Anthropic staffer says, personally, it will implement it too. One robots.txt token would cover several AI crawlers.
-
Build a Validated Kubernetes Runtime for NVIDIA DynamoNVIDIA details how to build a validated Kubernetes runtime for DynamoTeams building AI applications need a repeatable path from a model configuration to a working inference endpoint. NVIDIA walks through building a validated Kubernetes runtime for Dynamo, its distributed inference framework, aiming to standardize how a chosen model setup becomes a running inference service and to cut operational effort.
-
Latent Reasoning in Continuous Space for Unified Multimodal Models
-
RoboArchitect: Automating the Design of Robotic Agents with Persistent Memory
-
CRoS: Causality Reward on Scrambled CoTs Improves Accuracy and Reasoning Faithfulness
-
There Is No Free Loop: When Does Looping a Trained Model Pay for Its Inference Compute?
-
When Does Jacobian Fine-Tuning Stabilize Neural PDE Emulators?
-
TurnTurk-GPT: Text-Based Turn-Taking Prediction for Turkish
-
SecPO: Principled Adversarial Training for Prompt Injection Security
-
ID2Pattern: Pretraining Generalizable Identity Behavior Patterns for Transaction Risk Foundation Models
-
Chronos-2: A foundation model for multivariate time series forecasting
-
Shock Loading Accesses Reaction Pathways beyond Accelerated Thermolysis
-
Using Machine Learning to Measure Self-Efficacy Belief Through Learners' Interaction Traces in Technology-Enhanced Learning
-
General-purpose foundation models surpass specialized ones on chemical property prediction tasks
-
ReuScope: Auditing Data Reuse and Evaluation Sensitivity in Capsule Vision 2024
-
RECAST: REconstruction of Chain-of-Thought via Auditable Structured Traces
-
Towards Out-of-Distribution Uncertainty Quantification for Machine-Learned Interatomic Potentials via Pre-Training
-
Implicit Assumptions About Tasks in Continual Learning
-
Different Models, Shared Failures: Auditing Frozen-Encoder Pools for Medical Segmentation
-
Stepwise-Querying: Distilling Reasoning without Access to Teacher Chain-of-Thought
-
Latent-Lens: Visual Perception in Small Language Models Through Latent Communication
-
UpLoop: Training Looped Transformers via Upcycling at No Extra Pretraining Compute
-
Advancing Clinical Trial Outcome Prediction with Domain-Specific Large Language Models
-
Precision--Recall Trade-off on the Inference Level for LLMs in Grammatical Error Correction
-
Saidzoda-Tajik-LFM-2.6B: Adapting a 2.6B Hybrid Model to Tajik Tool Calling, and What Continued Pretraining Takes Away
-
BYOD: Build Your Own DLM - Efficient Autoregressive-to-Diffusion Language Model Conversion
-
Enhancing Knowledge Entity Relation Extraction in Abstracts and Full-text Content of Academic Articles Using Large Language Models
-
KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise