Open links in new tab
  1. Fish Passage Center Adult Dam Count Queries The FPC Adult pages contain dam count data for salmon and steelhead, adult passage Pit Tag detections, Fishermen's Links webpage, fishing forums …

  2. TSPulse: Tiny Pre-Trained Models with Disentangled Representations...

    Jan 26, 2026 · TSPulse presents an ultra-lightweight 1M-parameter re-trained model for time-series diagnostics (classiication, anomaly detection, imputation, similarity sarch) using disentangled …

  3. Learning Efficient Latent Reasoning with Abstract Chain-of-Thought

    Mar 1, 2026 · For the Qwen3-8B and Granite-3.3-8B language models, Abstract-CoT achieves up to 12 × fewer reasoning tokens while achieving comparable performance across mathematical reasoning, …

  4. Historic salmon runs | IFish Fishing Forum

    May 12, 2011 · The McCall Hatchery stock was originally established from adults trapped at Little Goose and Lower Granite Dams during the mid- to late 1970s. The original gene pool was likely made up of …

  5. Forum - OpenReview

    Promoting openness in scientific communication and the peer-review process

  6. SafeConstellations: Mitigating Over-Refusals in LLMs ... - OpenReview

    Jan 5, 2026 · LLMs increasingly exhibit over-refusal behavior, where safety mechanisms cause models to reject benign instructions that seemingly resemble harmful content. This phenomenon diminishes …

  7. Advancing LLM Safe Alignment with Safety Representation Ranking

    Sep 18, 2025 · 2b) The use of GPT-2 as a ranker for this safety based evaluation is incorrect. There is a wealth of literature developing guard models that can be used as safety evaluator for both training …

  8. TSP : T PRE-TRAINEDMODELS WITHDISEN REPRESENTATIONS FORRAPIDTIME-SERIES TSPULSE: TINY PRE-TRAINED MODELS WITH DISEN-TANGLED REPRESENTATIONS FOR …

  9. ABSTRACT Function calling significantly extends the application boundary of large language models (LLMs), where high-quality and diverse training data is critical for unlock-ing this capability. However, …

  10. ComputerRL: Scaling End-to-End Online Reinforcement Learning for...

    Jan 26, 2026 · This paper introduces ComputerRL, an end-to-end reinforcement learning framework for autonomous desktop agents that integrates a machine-oriented API–GUI interaction paradigm with a …