A new 263k TTCW-annotated story dataset shows non-reasoning fine-tuning of Qwen3 models outperforms reasoning-supervised fine-tuning for fixed-format long-form literary review generation.
hub
Title resolution pending
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
hub tools
years
2026 10representative citing papers
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
Image editing models fail zero-shot visual planning on abstract mazes and queen puzzles but generalize after finetuning, yet still cannot match human zero-shot efficiency.
STAR-Teaming uses a Strategy-Response Multiplex Network inside a multi-agent framework to organize attack strategies into semantic communities, delivering higher attack success rates on LLMs at lower computational cost than prior methods.
VLMs as judges exhibit informativeness bias by favoring detailed but image-inconsistent answers; BIRCH mitigates it by first correcting answers against the image, reducing bias up to 17% and improving performance up to 9.8%.
Hallucinations arise from biased latent inference paths rather than missing knowledge, demonstrated via a new diagnostic testbed TrapQA that isolates task-retrieval and key-selection biases.
Interventions in LLM-simulated user experiments induce distribution shifts in latent attributes that create confounding bias, diagnosable with negative control outcomes and partially mitigated by adding setting-relevant persona details.
DASH reduces prefill computation in LLMs and vision-language models by halting tokens with low attention-branch residual updates at a chosen layer, achieving 1.7–2× prefill speedups with minimal accuracy loss.
Orbax supplies a modular JAX-native checkpointing library that claims up to 3.5× faster saves and 2× faster loads than comparable PyTorch solutions.
citing papers explorer
-
When Reasoning Supervision Hurts: TTCW-Based Long-Form Literary Review Generation
A new 263k TTCW-annotated story dataset shows non-reasoning fine-tuning of Qwen3 models outperforms reasoning-supervised fine-tuning for fixed-format long-form literary review generation.
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
Probing Visual Planning in Image Editing Models
Image editing models fail zero-shot visual planning on abstract mazes and queen puzzles but generalize after finetuning, yet still cannot match human zero-shot efficiency.
-
STAR-Teaming: A Strategy-Response Multiplex Network Approach to Automated LLM Red Teaming
STAR-Teaming uses a Strategy-Response Multiplex Network inside a multi-agent framework to organize attack strategies into semantic communities, delivering higher attack success rates on LLMs at lower computational cost than prior methods.
-
When Vision-Language Models Judge Without Seeing: Exposing Informativeness Bias
VLMs as judges exhibit informativeness bias by favoring detailed but image-inconsistent answers; BIRCH mitigates it by first correcting answers against the image, reducing bias up to 17% and improving performance up to 9.8%.
-
Understanding Why Language Models Hallucinate: Testing Reasoning Against Priors
Hallucinations arise from biased latent inference paths rather than missing knowledge, demonstrated via a new diagnostic testbed TrapQA that isolates task-retrieval and key-selection biases.
-
The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study
Interventions in LLM-simulated user experiments induce distribution shifts in latent attributes that create confounding bias, diagnosable with negative control outcomes and partially mitigated by adding setting-relevant persona details.
-
Stability Implies Redundancy: Delta Attention Selective Halting for Efficient Long-Context Prefilling
DASH reduces prefill computation in LLMs and vision-language models by halting tokens with low attention-branch residual updates at a chosen layer, achieving 1.7–2× prefill speedups with minimal accuracy loss.
-
Orbax: Distributed Checkpointing with JAX
Orbax supplies a modular JAX-native checkpointing library that claims up to 3.5× faster saves and 2× faster loads than comparable PyTorch solutions.
- When Audio-Language Models Fail to Leverage Multimodal Context for Dysarthric Speech Recognition