✨Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs
📝 Summary:
This paper optimizes LLM chain-of-thought reasoning by addressing redundant reflections and overthinking. It uses a graph-based framework to convert CoT into a DAG and applies dual pruning strategies to remove inefficient reflection patterns. This approach reduces reasoning tokens by 42% while ma...
🔹 Publication Date: Published on Apr 7
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.05643
• PDF: https://arxiv.org/pdf/2604.05643
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #ChainOfThought #AI #GraphAlgorithms #Reasoning
📝 Summary:
This paper optimizes LLM chain-of-thought reasoning by addressing redundant reflections and overthinking. It uses a graph-based framework to convert CoT into a DAG and applies dual pruning strategies to remove inefficient reflection patterns. This approach reduces reasoning tokens by 42% while ma...
🔹 Publication Date: Published on Apr 7
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.05643
• PDF: https://arxiv.org/pdf/2604.05643
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #ChainOfThought #AI #GraphAlgorithms #Reasoning
✨SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding
📝 Summary:
SPEED-Bench is introduced as a new benchmark for Speculative Decoding SD evaluation. It provides diverse semantic domains and realistic serving regimes to address limitations of existing benchmarks. This enables accurate measurement of SD performance in production environments, setting a unified ...
🔹 Publication Date: Published on Feb 10
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.09557
• PDF: https://arxiv.org/pdf/2604.09557
• Project Page: https://huggingface.co/blog/nvidia/speed-bench
• Github: https://github.com/NVIDIA/Model-Optimizer
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#SpeculativeDecoding #AIBenchmarks #LLMs #DeepLearning #ModelOptimization
📝 Summary:
SPEED-Bench is introduced as a new benchmark for Speculative Decoding SD evaluation. It provides diverse semantic domains and realistic serving regimes to address limitations of existing benchmarks. This enables accurate measurement of SD performance in production environments, setting a unified ...
🔹 Publication Date: Published on Feb 10
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.09557
• PDF: https://arxiv.org/pdf/2604.09557
• Project Page: https://huggingface.co/blog/nvidia/speed-bench
• Github: https://github.com/NVIDIA/Model-Optimizer
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#SpeculativeDecoding #AIBenchmarks #LLMs #DeepLearning #ModelOptimization
✨TRACER: Trace-Based Adaptive Cost-Efficient Routing for LLM Classification
📝 Summary:
TRACER trains ML surrogates using LLM classification production traces. These cost-efficient surrogates activate only if they agree with the original LLM above a threshold, saving significant costs. TRACER also provides interpretability for its routing decisions and achieves high coverage.
🔹 Publication Date: Published on Apr 16
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14531
• PDF: https://arxiv.org/pdf/2604.14531
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #MachineLearning #CostEfficiency #AI #Interpretability
📝 Summary:
TRACER trains ML surrogates using LLM classification production traces. These cost-efficient surrogates activate only if they agree with the original LLM above a threshold, saving significant costs. TRACER also provides interpretability for its routing decisions and achieves high coverage.
🔹 Publication Date: Published on Apr 16
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.14531
• PDF: https://arxiv.org/pdf/2604.14531
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #MachineLearning #CostEfficiency #AI #Interpretability
✨PRL-Bench: A Comprehensive Benchmark Evaluating LLMs' Capabilities in Frontier Physics Research
📝 Summary:
PRL-Bench is a new benchmark evaluating LLMs' end-to-end capabilities in theoretical and computational physics research. It uses 100 curated papers to assess exploration-oriented, long-horizon workflows. Current LLMs perform poorly, revealing a significant gap in autonomous scientific discovery.
🔹 Publication Date: Published on Apr 16
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15411
• PDF: https://arxiv.org/pdf/2604.15411
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #PhysicsResearch #ScientificDiscovery #AI #Benchmarking
📝 Summary:
PRL-Bench is a new benchmark evaluating LLMs' end-to-end capabilities in theoretical and computational physics research. It uses 100 curated papers to assess exploration-oriented, long-horizon workflows. Current LLMs perform poorly, revealing a significant gap in autonomous scientific discovery.
🔹 Publication Date: Published on Apr 16
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15411
• PDF: https://arxiv.org/pdf/2604.15411
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #PhysicsResearch #ScientificDiscovery #AI #Benchmarking
✨GTA-2: Benchmarking General Tool Agents from Atomic Tool-Use to Open-Ended Workflows
📝 Summary:
GTA-2 is a new benchmark for General Tool Agents, covering both atomic and real-world, open-ended workflows. It shows frontier models struggle significantly, especially on workflows. The study emphasizes that execution frameworks are crucial for performance, more so than just model capacity.
🔹 Publication Date: Published on Apr 17
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15715
• PDF: https://arxiv.org/pdf/2604.15715
• Github: https://github.com/open-compass/GTA
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AIAgents #BenchmarkingAI #LLMs #AIWorkflows #AIResearch
📝 Summary:
GTA-2 is a new benchmark for General Tool Agents, covering both atomic and real-world, open-ended workflows. It shows frontier models struggle significantly, especially on workflows. The study emphasizes that execution frameworks are crucial for performance, more so than just model capacity.
🔹 Publication Date: Published on Apr 17
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.15715
• PDF: https://arxiv.org/pdf/2604.15715
• Github: https://github.com/open-compass/GTA
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AIAgents #BenchmarkingAI #LLMs #AIWorkflows #AIResearch
✨ShadowPEFT: Shadow Network for Parameter-Efficient Fine-Tuning
📝 Summary:
ShadowPEFT is a new parameter-efficient fine-tuning framework that uses a depth-shared shadow module for layer-level refinement. This shifts adaptation from distributed weight perturbations to a shared layer-space process, matching or outperforming LoRA with reduced overhead and increased flexibi...
🔹 Publication Date: Published on Apr 21
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.19254
• PDF: https://arxiv.org/pdf/2604.19254
• Project Page: https://github.com/ShadowLLM/shadow-peft
• Github: https://github.com/ShadowLLM/shadow-peft
🔹 Models citing this paper:
• https://huggingface.co/shadow-llm/Qwen3-4B-GSM8k-Shadow
• https://huggingface.co/shadow-llm/Qwen3-4B-SquadV2-Shadow
• https://huggingface.co/shadow-llm/Qwen3-4B-MMLU-Shadow
✨ Datasets citing this paper:
• https://huggingface.co/datasets/shadow-llm/robot-dog-skills
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#PEFT #FineTuning #MachineLearning #AI #LLMs
📝 Summary:
ShadowPEFT is a new parameter-efficient fine-tuning framework that uses a depth-shared shadow module for layer-level refinement. This shifts adaptation from distributed weight perturbations to a shared layer-space process, matching or outperforming LoRA with reduced overhead and increased flexibi...
🔹 Publication Date: Published on Apr 21
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.19254
• PDF: https://arxiv.org/pdf/2604.19254
• Project Page: https://github.com/ShadowLLM/shadow-peft
• Github: https://github.com/ShadowLLM/shadow-peft
🔹 Models citing this paper:
• https://huggingface.co/shadow-llm/Qwen3-4B-GSM8k-Shadow
• https://huggingface.co/shadow-llm/Qwen3-4B-SquadV2-Shadow
• https://huggingface.co/shadow-llm/Qwen3-4B-MMLU-Shadow
✨ Datasets citing this paper:
• https://huggingface.co/datasets/shadow-llm/robot-dog-skills
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#PEFT #FineTuning #MachineLearning #AI #LLMs
arXiv.org
ShadowPEFT: Shadow Network for Parameter-Efficient Fine-Tuning
Parameter-efficient fine-tuning (PEFT) reduces the training cost of full-parameter fine-tuning for large language models (LLMs) by training only a small set of task-specific parameters while...
✨WebGen-R1: Incentivizing Large Language Models to Generate Functional and Aesthetic Websites with Reinforcement Learning
📝 Summary:
WebGen-R1 is a reinforcement learning framework enabling small language models to generate functional and aesthetically pleasing multi-page websites. It uses structured generation and a novel cascaded multimodal reward for structural integrity, functional feedback, and aesthetic supervision. WebG...
🔹 Publication Date: Published on Apr 22
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.20398
• PDF: https://arxiv.org/pdf/2604.20398
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#ReinforcementLearning #LLMs #WebsiteGeneration #AI #WebDevelopment
📝 Summary:
WebGen-R1 is a reinforcement learning framework enabling small language models to generate functional and aesthetically pleasing multi-page websites. It uses structured generation and a novel cascaded multimodal reward for structural integrity, functional feedback, and aesthetic supervision. WebG...
🔹 Publication Date: Published on Apr 22
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.20398
• PDF: https://arxiv.org/pdf/2604.20398
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#ReinforcementLearning #LLMs #WebsiteGeneration #AI #WebDevelopment
arXiv.org
WebGen-R1: Incentivizing Large Language Models to Generate...
While Large Language Models (LLMs) excel at function-level code generation, project-level tasks such as generating functional and visually aesthetic multi-page websites remain highly challenging....
✨Encoder-Free Human Motion Understanding via Structured Motion Descriptions
📝 Summary:
Structured Motion Description SMD converts human motion into natural language, enabling large language models LLMs to reason about it directly. This encoder-free method achieves state-of-the-art performance on motion question answering and captioning.
🔹 Publication Date: Published on Apr 23
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.21668
• PDF: https://arxiv.org/pdf/2604.21668
• Project Page: https://yaozhang182.github.io/motion-smd/
• Github: https://yaozhang182.github.io/motion-smd/
🔹 Models citing this paper:
• https://huggingface.co/zyyy12138/motion-smd-lora
✨ Datasets citing this paper:
• https://huggingface.co/datasets/zyyy12138/motion-smd-data
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#HumanMotionUnderstanding #LLMs #NLP #AI #DeepLearning
📝 Summary:
Structured Motion Description SMD converts human motion into natural language, enabling large language models LLMs to reason about it directly. This encoder-free method achieves state-of-the-art performance on motion question answering and captioning.
🔹 Publication Date: Published on Apr 23
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.21668
• PDF: https://arxiv.org/pdf/2604.21668
• Project Page: https://yaozhang182.github.io/motion-smd/
• Github: https://yaozhang182.github.io/motion-smd/
🔹 Models citing this paper:
• https://huggingface.co/zyyy12138/motion-smd-lora
✨ Datasets citing this paper:
• https://huggingface.co/datasets/zyyy12138/motion-smd-data
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#HumanMotionUnderstanding #LLMs #NLP #AI #DeepLearning
arXiv.org
Encoder-Free Human Motion Understanding via Structured Motion Descriptions
The world knowledge and reasoning capabilities of text-based large language models (LLMs) are advancing rapidly, yet current approaches to human motion understanding, including motion question...
❤1
✨AutoResearchBench: Benchmarking AI Agents on Complex Scientific Literature Discovery
📝 Summary:
AutoResearchBench is a new benchmark evaluating AI agents on complex scientific literature discovery tasks. It features deep and wide research, demanding in-depth comprehension and fine-grained information use. Even powerful LLMs show very low accuracy, highlighting the significant challenge for ...
🔹 Publication Date: Published on Apr 28
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.25256
• PDF: https://arxiv.org/pdf/2604.25256
• Project Page: https://cheryou.github.io/autoresearchbench.github.io/
• Github: https://github.com/CherYou/AutoResearchBench
✨ Datasets citing this paper:
• https://huggingface.co/datasets/Lk123/AutoResearchBench
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AI #LLMs #Benchmarking #ScientificResearch #AIAgents
📝 Summary:
AutoResearchBench is a new benchmark evaluating AI agents on complex scientific literature discovery tasks. It features deep and wide research, demanding in-depth comprehension and fine-grained information use. Even powerful LLMs show very low accuracy, highlighting the significant challenge for ...
🔹 Publication Date: Published on Apr 28
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.25256
• PDF: https://arxiv.org/pdf/2604.25256
• Project Page: https://cheryou.github.io/autoresearchbench.github.io/
• Github: https://github.com/CherYou/AutoResearchBench
✨ Datasets citing this paper:
• https://huggingface.co/datasets/Lk123/AutoResearchBench
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#AI #LLMs #Benchmarking #ScientificResearch #AIAgents
❤1
✨SignRoundV2: Closing the Performance Gap in Extremely Low-Bit Post-Training Quantization for LLMs
📝 Summary:
SignRoundV2 is a post-training quantization method for LLMs. It achieves competitive, near full-precision accuracy even at extremely low-bits like 2-bits. This is done via layer-wise bit allocation and pre-tuning scale search.
🔹 Publication Date: Published on Dec 4, 2025
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.04746
• PDF: https://arxiv.org/pdf/2512.04746
• Project Page: https://github.com/intel/auto-round
• Github: https://github.com/intel/auto-round
🔹 Models citing this paper:
• https://huggingface.co/Intel/MiroThinker-v1.5-30B-gguf-q2ks-mixed-AutoRound
• https://huggingface.co/Intel/DeepSeek-R1-0528-Qwen3-8B-int4-AutoRound
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #Quantization #DeepLearning #AI #MachineLearning
📝 Summary:
SignRoundV2 is a post-training quantization method for LLMs. It achieves competitive, near full-precision accuracy even at extremely low-bits like 2-bits. This is done via layer-wise bit allocation and pre-tuning scale search.
🔹 Publication Date: Published on Dec 4, 2025
🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.04746
• PDF: https://arxiv.org/pdf/2512.04746
• Project Page: https://github.com/intel/auto-round
• Github: https://github.com/intel/auto-round
🔹 Models citing this paper:
• https://huggingface.co/Intel/MiroThinker-v1.5-30B-gguf-q2ks-mixed-AutoRound
• https://huggingface.co/Intel/DeepSeek-R1-0528-Qwen3-8B-int4-AutoRound
==================================
For more data science resources:
✓ https://xn--r1a.website/DataScienceT
#LLMs #Quantization #DeepLearning #AI #MachineLearning