大模型推理能力 - arXivDaily 专题

2502.10325 2025-02-17 cs.LG cs.AI 62%

Process Reward Models for LLM Agents: Practical Framework and Directions

Sanjiban Choudhury

专题命中测试时计算：reasoning(abstract)；分类 cs.AI、cs.LG

Comments 17 pages, 7 figures

2410.11287 2025-02-12 cs.CL cs.AI 62%

Process Reward Model with Q-Value Rankings

Wendi Li, Yixuan Li

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2410.08020 2025-02-11 cs.LG cs.AI 62%

Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs

Jonas Hübotter, Sascha Bongni, Ido Hakimi, Andreas Krause

专题命中测试时计算：test-time compute(abstract)；分类 cs.AI、cs.LG

Comments accepted in ICLR 2025

2502.04576 2025-02-10 cs.LG cs.CL 62%

Self-Regulation and Requesting Interventions

So Yeon Min, Yue Wu, Jimin Sun, Max Kaufmann, Fahim Tajwar, Yonatan Bisk, Ruslan Salakhutdinov

专题命中测试时计算：test-time compute(abstract)；分类 cs.CL、cs.LG

2501.17725 2025-01-30 cs.AI cs.CL cs.DM math.CO 62%

Using Code Generation to Solve Open Instances of Combinatorial Design Problems

Christopher D. Rosin

专题命中测试时计算：verifier(abstract)；分类 cs.CL、cs.AI

2409.06131 2025-01-30 cs.CL cs.AI 62%

Accelerating Large Language Model Pretraining via LFR Pedagogy: Learn, Focus, and Review

Neha Prakriya, Jui-Nan Yen, Cho-Jui Hsieh, Jason Cong

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2501.15109 2025-01-28 cs.LG cs.AI 62%

Clear Preferences Leave Traces: Reference Model-Guided Sampling for Preference Learning

Nirav Diwan, Tolga Ergen, Dongsub Shim, Honglak Lee

专题命中测试时计算：reasoning(abstract)；分类 cs.AI、cs.LG

2412.13023 2024-12-18 cs.AI cs.LG 62%

Relational Neurosymbolic Markov Models

Lennert De Smet, Gabriele Venturato, Luc De Raedt, Giuseppe Marra

专题命中测试时计算：reasoning(abstract)；分类 cs.AI、cs.LG

Comments Accepted at AAAI 2025

2412.05299 2024-12-17 cs.SE cs.AI cs.CL 62%

Specifications: The missing link to making the development of LLM systems an engineering discipline

Ion Stoica, Matei Zaharia, Joseph Gonzalez, Ken Goldberg, Koushik Sen, Hao Zhang, Anastasios Angelopoulos, Shishir G. Patil, Lingjiao Chen, Wei-Lin Chiang, Jared Q. Davis

专题命中测试时计算：test-time compute(abstract)；分类 cs.CL、cs.AI

2408.09849 2024-12-13 cs.CL cs.AI 62%

Importance Weighting Can Help Large Language Models Self-Improve

Chunyang Jiang, Chi-min Chan, Wei Xue, Qifeng Liu, Yike Guo

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2410.02742 2024-11-13 cs.CL cs.LG cs.RO 62%

Grounding Large Language Models In Embodied Environment With Imperfect World Models

Haolan Liu, Jishen Zhao

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

2410.22480 2024-10-31 cs.CL cs.AI 62%

Scaling LLM Inference with Optimized Sample Compute Allocation

Kexun Zhang, Shang Zhou, Danqing Wang, William Yang Wang, Lei Li

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2410.07114 2024-10-28 cs.CY cs.AI cs.CL 62%

System 2 thinking in OpenAI's o1-preview model: Near-perfect performance on a mathematics exam

Joost de Winter, Dimitra Dodou, Yke Bauke Eisma

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Journal ref Computers 13 (2024) 278

2406.07515 2024-10-28 cs.LG cs.AI stat.ML 62%

Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification

Yunzhen Feng, Elvis Dohmatob, Pu Yang, Francois Charton, Julia Kempe

专题命中测试时计算：verifier(abstract)；分类 cs.AI、cs.LG

2410.15184 2024-10-22 cs.LG cs.AI 62%

Action abstractions for amortized sampling

Oussama Boussif, Léna Néhale Ezzine, Joseph D Viviano, Michał Koziarski, Moksh Jain, Nikolay Malkin, Emmanuel Bengio, Rim Assouel, Yoshua Bengio

专题命中测试时计算：planning(abstract)；分类 cs.AI、cs.LG

2406.02721 2024-10-15 cs.CL cs.AI 62%

Self-Control of LLM Behaviors by Compressing Suffix Gradient into Prefix Controller

Min Cai, Yuchen Zhang, Shichang Zhang, Fan Yin, Dan Zhang, Difan Zou, Yisong Yue, Ziniu Hu

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Website: https://llm-self-control.github.io/

2310.11571 2024-10-15 cs.CL cs.LG 62%

Alexpaca: Learning Factual Clarification Question Generation Without Examples

Matthew Toles, Yukun Huang, Zhou Yu, Luis Gravano

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

2408.16737 2024-10-10 cs.CL cs.AI 62%

Smaller, Weaker, Yet Better: Training LLM Reasoners via Compute-Optimal Sampling

Hritik Bansal, Arian Hosseini, Rishabh Agarwal, Vinh Q. Tran, Mehran Kazemi

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2410.03083 2024-10-07 cs.CL cs.AI 62%

Scaling Parameter-Constrained Language Models with Quality Data

Ernie Chang, Matteo Paltenghi, Yang Li, Pin-Jie Lin, Changsheng Zhao, Patrick Huber, Zechun Liu, Rastislav Rabatin, Yangyang Shi, Vikas Chandra

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted to EMNLP 2024 Industry Track, 18 pages, 9 figures, 4 tables

2405.13014 2024-09-20 cs.CL cs.AI 62%

QCRD: Quality-guided Contrastive Rationale Distillation for Large Language Models

Wei Wang, Zhaowei Li, Qi Xu, Yiqing Cai, Hang Song, Qi Qi, Ran Zhou, Zhida Huang, Tao Wang, Li Xiao

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2408.15562 2024-08-29 cs.CL cs.LG 62%

Boosting Lossless Speculative Decoding via Feature Sampling and Partial Alignment Distillation

Lujun Gui, Bin Xiao, Lei Su, Weipeng Chen

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

Comments The work was not submitted to AAAI 2025

2408.08869 2024-08-20 cs.CL cs.LG 62%

PEDAL: Enhancing Greedy Decoding with Large Language Models using Diverse Exemplars

Sumanth Prabhu

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

2407.03651 2024-07-16 cs.CL cs.AI 62%

Evaluating Language Model Context Windows: A "Working Memory" Test and Inference-time Correction

Amanda Dsouza, Christopher Glaze, Changho Shin, Frederic Sala

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

2407.03678 2024-07-08 cs.CL cs.LG 62%

Improving Self Consistency in LLMs through Probabilistic Tokenization

Ashutosh Sathe, Divyanshu Aggarwal, Sunayana Sitaram

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

Comments ICML 2024 Workshop on LLMs and Cognition

2407.02056 2024-07-03 cs.CL cs.AI 62%

Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language Generation

Xinglin Wang, Yiwei Li, Shaoxiong Feng, Peiwen Yuan, Boyuan Pan, Heda Wang, Yao Hu, Kan Li

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted to ACL2024 Main Conference

2309.17272 2024-07-03 cs.CL cs.AI cs.SE 62%

Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency

Baizhou Huang, Shuai Lu, Weizhu Chen, Xiaojun Wan, Nan Duan

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments ACL 2024

2406.07887 2024-06-13 cs.LG cs.CL 62%

An Empirical Study of Mamba-based Language Models

Roger Waleffe, Wonmin Byeon, Duncan Riach, Brandon Norick, Vijay Korthikanti, Tri Dao, Albert Gu, Ali Hatamizadeh, Sudhakar Singh, Deepak Narayanan, Garvit Kulshreshtha, Vartika Singh, Jared Casper, Jan Kautz, Mohammad Shoeybi, Bryan Catanzaro

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.LG

2406.05322 2024-06-11 cs.CL cs.AI 62%

Teaching-Assistant-in-the-Loop: Improving Knowledge Distillation from Imperfect Teacher Models in Low-Budget Scenarios

Yuhang Zhou, Wei Ai

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted by ACL 2024 Findings

2402.19471 2024-05-03 cs.CL cs.AI 62%

Loose LIPS Sink Ships: Asking Questions in Battleship with Language-Informed Program Sampling

Gabriel Grand, Valerio Pepe, Jacob Andreas, Joshua B. Tenenbaum

专题命中测试时计算：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted to CogSci 2024

2309.16108 2024-04-22 cs.CV cs.AI cs.LG 62%

Channel Vision Transformers: An Image Is Worth 1 x 16 x 16 Words

Yujia Bao, Srinivasan Sivanandan, Theofanis Karaletsos

专题命中测试时计算：reasoning(abstract)；分类 cs.AI、cs.LG

1. 测试时计算 1990 篇

Process Reward Models for LLM Agents: Practical Framework and Directions

Process Reward Model with Q-Value Rankings

Efficiently Learning at Test-Time: Active Fine-Tuning of LLMs

Self-Regulation and Requesting Interventions

Using Code Generation to Solve Open Instances of Combinatorial Design Problems

Accelerating Large Language Model Pretraining via LFR Pedagogy: Learn, Focus, and Review

Clear Preferences Leave Traces: Reference Model-Guided Sampling for Preference Learning

Relational Neurosymbolic Markov Models

Specifications: The missing link to making the development of LLM systems an engineering discipline

Importance Weighting Can Help Large Language Models Self-Improve

Grounding Large Language Models In Embodied Environment With Imperfect World Models

Scaling LLM Inference with Optimized Sample Compute Allocation

System 2 thinking in OpenAI's o1-preview model: Near-perfect performance on a mathematics exam

Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification

Action abstractions for amortized sampling

Self-Control of LLM Behaviors by Compressing Suffix Gradient into Prefix Controller

Alexpaca: Learning Factual Clarification Question Generation Without Examples

Smaller, Weaker, Yet Better: Training LLM Reasoners via Compute-Optimal Sampling

Scaling Parameter-Constrained Language Models with Quality Data

QCRD: Quality-guided Contrastive Rationale Distillation for Large Language Models

Boosting Lossless Speculative Decoding via Feature Sampling and Partial Alignment Distillation

PEDAL: Enhancing Greedy Decoding with Large Language Models using Diverse Exemplars

Evaluating Language Model Context Windows: A "Working Memory" Test and Inference-time Correction

Improving Self Consistency in LLMs through Probabilistic Tokenization

Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language Generation

Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistency

An Empirical Study of Mamba-based Language Models

Teaching-Assistant-in-the-Loop: Improving Knowledge Distillation from Imperfect Teacher Models in Low-Budget Scenarios

Loose LIPS Sink Ships: Asking Questions in Battleship with Language-Informed Program Sampling

Channel Vision Transformers: An Image Is Worth 1 x 16 x 16 Words