Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 16 of 16 for “"GRPO"”.
-
POST-TRAINING OF VISION-LANGUAGE AGENTS FOR DECENTRALIZED AUTONOMOUS VEHICLE COORDINATION USING GENERALIZABLE MULTI-AGENT REWARDS
… extend AR1’s Group Relative Policy Optimization (GRPO) post-training by augmenting thereward with Expectation Alignment (ELIGN), an intrinsic social term that penalizes mismatch between a learned neighbor-expectation model and the realized shared next observation. To make ELIGN applicable to AR1’s …
-
Didžiųjų kalbos modelių loginio mąstymo gebėjimų tobulinimas lietuvių kalbai taikant sustiprinimo mokymą: metodologija, duomenų paruošimas ir palyginamoji analizė /
… was trained using three different algorithms: GRPO, DAPO, and LogicRL Reinforce++. The primary objective of the experiment was to incentivize the model to generate accurate, deep, and coherent reasoning chains to successfully solve complex logical tasks. The experimental results revealed that …
-
Object Segmentation Reasoning
… Learning via Group Relative Policy Optimization (GRPO) to optimize spatial precision. The thesis concludes with a critical comparison of these methodologies. While the Multi-Module architecture currently offers superior stability and immediate deployment utility, the End-to-End approach …
-
Generating humor with information-theoretic reward signals - A system trained to write funny cartoon captions and humorous news headlines, optimized with reinforcement learning using human preference reward models and an information-theoretic approach to humor theory (:
… with Group Relative Policy Optimization (GRPO) against the combined reward. Across both tasks, the aligned generator does not produce funnier output than the fine-tuned baseline. Reward arms are either statistically indistinguishable from the SFT initialization or collapse into degenerate, …
-
NanoMem: Iterative Evidence Synthesis for Temporal-Causal Reasoning in Agent Memory
… retrieval round. The Synthesizer is trained with GRPO using multi-level rewards covering answer correctness, evidence compactness, output format, and a per-round sufficiency verdict that provides direct process supervision for the stop-or-continue decision. A new diagnostic benchmark, TimeMemEval, …
-
LEGAL REASONING WITH SEARCH-AUGMENTED REINFORCEMENT LEARNING: A MULTI-TASK FRAMEWORK
… patterns. Implemented with Qwen 2.5 3B and GRPO optimization on 8 A100 GPUs, the system achieves a 10.5 percentage point improvement over baselines on LegalBench (59.6% vs. 49.7%), 97% jurisdictional compliance, and 0.72 Spearman correlation between model attention and expert fact importance …
-
Test-time fine-tuning via diverse rewards: a conditional variational reward model approach on the abstraction and reasoning corpus
… through Group Relative Policy Optimization (GRPO) during test time. This dynamic fine-tuning process allows the model to adapt “on the fly,” improving its reasoning and abstraction capabilities with minimal examples and computational overhead. Extensive experiments and ablation studies …
-
Neurosymbolic ViT-LoMoE : toward interpretable and knowledge-consistent visual recognition
… pre-training, supervised LoMoE optimization, and GRPO-based reinforcement-learning fine-tuning of the router which is evaluated from scratch on four plant pathology benchmarks covering both controlled and in-field imaging conditions. Across all backbone configurations, ViT-LoMoE consistently …
-
When process rewards fail : synthetic negatives and specification gaming in PRM-based RL
… scale. At Phase 2 — Qwen2.5-7B-Instruct, 500 GRPO steps, three seeds, n_test = 500 — pure process rewards lose roughly thirty accuracy points relative to outcome RL while process_correctness approaches 1.0, with length and redundancy inflation. Composing the PRM with other signals prevents …
-
Novel Reinforcement Learning Framework For Alleviating The Germline Bias In Autoregressive Antibody Language Models
… to the original reinforcement learning GRPO algorithm for efficient RL training in the context of antibodies.
-
Multi-Layer OTN Simulation and LLM-Driven Root Cause Analysis: From Alarm Propagation to Reinforcement-Optimized Diagnostic Agents
… adapters and Group Relative Policy Optimization (GRPO) on Qwen 2.5-3B using composite reward functions. A ReAct agent framework wraps the fine-tuned model with five diagnostic tools; a category-based triage layer routes queries to specialist prompts (Fiber, XCON, or Line), narrowing the search …
-
Learning dynamics of deep learning -- force analysis of deep neural networks
… across finetuning methods such as SFT, DPO, and GRPO. We also highlight the critical role of negative gradients. In particular, we identify the "squeezing effect": a counterintuitive phenomenon caused by improperly applied gradient ascent. Representation learning: We explore the dynamics of …
-
Towards Smarter Trading: An AI Trading Framework Combining Reinforcement Learning and Large Language Model
… applies Group Relative Policy Optimization (GRPO) directly to a general-purpose instruction-tuned LLM, enabling the model to develop trading competence purely through reward-driven exploration, without curated professional signals. The model ingests multi-source market observations — …
-
Advancing multi-session personalization in LLMs via direct policy
… Hierarchical Group Relative Policy Optimization (GRPO) pipeline to overcome the instability and improve the sample efficiency of the indirect supervision method, which is common for retriever-generator Joint Training. In addition, this study addresses the lack of long horizon, multi-session, …
-
Toward Reliable Supervision for Foundation Models
… into Group Relative Policy Optimization (GRPO) while preserving verifier authority: verifier rewards drive learning when they are informative, and PRM-based dense supervision is used only when verifier rewards are degenerate. Step scores are converted into future-cumulated token-level …
-
GPS-Driven evacuation analysis and trajectory synthesis using Large Language Models
… and destination-aware prompting. Through GRPO-based reinforcement fine-tuning, the framework achieves over 85% next-step prediction accuracy and generates trajectories that closely match real spatial distributions, providing a scalable and privacy-preserving tool for evacuation planning …