From Imitation to Discrimination: Progressive Curriculum Learning for Robust Web Navigation

• URL: http://arxiv.org/abs/2604.12666v1
• Date: Tue, 14 Apr 2026 12:37:45 GMT
• Title: From Imitation to Discrimination: Progressive Curriculum Learning for Robust Web Navigation
• Abstract summary: We introduce the Triton dataset (590k instances) and a progressive training curriculum.<n>We produce three models: Triton-SFT-32B for basic imitation, Triton-ORPO-32B for robust discrimination, and Triton-GRPO-32B for long-horizon consistency.<n> Empirical evaluation on Mind2Web demonstrates that Triton-GRPO-32B achieves state-of-the-art performance among open-source models.
• Score: 18.965820679356366
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Text-based web agents offer computational efficiency for autonomous web navigation, yet developing robust agents remains challenging due to the noisy and heterogeneous nature of real-world HTML. Standard Supervised Fine-Tuning (SFT) approaches fail in two critical dimensions: they lack discrimination capabilities to reject plausible but incorrect elements in densely populated pages, and exhibit limited generalization to unseen website layouts. To address these challenges, we introduce the Triton dataset (590k instances) and a progressive training curriculum. Triton is constructed via Structural-Semantic Hard Negative Mining, which explicitly mines topologically similar distractors, and a Dual-Agent Consensus pipeline that synthesizes diverse cross-domain tasks with strict verification. Building upon this foundation, our progressive curriculum produces three models: Triton-SFT-32B for basic imitation, Triton-ORPO-32B for robust discrimination via Odds Ratio Preference Optimization, and Triton-GRPO-32B for long-horizon consistency through Group Relative Policy Optimization. Empirical evaluation on Mind2Web demonstrates that Triton-GRPO-32B achieves state-of-the-art performance among open-source models with 58.7% Step Success Rate, surpassing GPT-4.5 (42.4%) and Claude-4.5 (41.4%) by over 16%, validating that specialized data curriculum outweighs raw parameter scale for web navigation.

Related papers

• AECNav: Active Evidence Consolidation for Efficient Zero-Shot Open-Vocabulary Object Navigation [10.327375579404402]
  Zero-shot object-goal navigation (ZSON) in open-vocabulary scenarios is challenging.<n>Currently, the task still suffers from high latency and limited accuracy due to redundant perception pipelines.<n>We present AECNav, a training-free pipeline built on three components.
  arXiv  Detail & Related papers  (2026-08-11T11:40:05Z)
• EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents [64.96332056338923]
  EEVEE is the first multi-dataset test-time prompt learning framework for LLM agents.<n>It enables test-time prompt learning under real-world task streams.<n>EEVEE improves average multi-benchmark scores by 10.38 and 24.32 points over Qwen3-4B-Instruct and DeepSeek-V3.2.
  arXiv  Detail & Related papers  (2026-06-09T17:57:16Z)
• Structured Distillation of Web Agent Capabilities Enables Generalization [33.41723014134473]
  We introduce Agent-as-Annotators, a framework that structures synthetic trajectory generation for web agents by analogy to human annotation roles.<n>Using Gemini 3 Pro as teacher, we generate 3,000 trajectories across six web environments.<n>The resulting model achieves 41.5% on WebArena, surpassing closed-source models such as Claude 3.5 Sonnet (36.0%) and GPT-4o (31.5%) under the same evaluation protocol.
  arXiv  Detail & Related papers  (2026-04-09T04:04:15Z)
• Portfolio Optimization Proxies under Label Scarcity and Regime Shifts via Bayesian and Deterministic Students under Semi-Supervised Sandwich Training [0.0]
  This paper proposes a machine learning assisted portfolio optimization framework designed for low data environments and regime uncertainty.<n>We construct a teacher student learning pipeline in which a Conditional Value at Risk (CVaR) generates supervisory labels, and neural models are trained using both real and synthetically augmented data.<n>Student models can match or outperform the CVaR teacher in several settings, while achieving improved robustness under regime shifts and reduced turnover.
  arXiv  Detail & Related papers  (2026-04-04T06:42:38Z)
• DecepGPT: Schema-Driven Deception Detection with Multicultural Datasets and Robust Multimodal Learning [64.33887406863899]
  Multimodal deception detection aims to identify deceptive behavior by analyzing audiovisual cues for forensics and security.<n>Existing benchmarks provide only binary labels without intermediate reasoning cues.<n>We construct reasoning datasets with structured cue-level descriptions and reasoning chains.<n>With 1695 samples, it is the largest non-laboratory deception detection dataset.
  arXiv  Detail & Related papers  (2026-03-25T04:06:36Z)
• AgentCPM-Explore: Realizing Long-Horizon Deep Exploration for Edge-Scale Agents [75.67445299298949]
  AgentCPM-Explore is a compact 4B agent model with high knowledge density and strong exploration capability.<n>We introduce a holistic training framework featuring parameter-space model fusion, reward signal denoising, and contextual information refinement.<n>AgentCPM-Explore achieves state-of-the-art (SOTA) performance among 4B-class models, matches or surpasses 8B-class SOTA models on four benchmarks, and even outperforms larger-scale models such as Claude-4.5-Sonnet or DeepSeek-v3.2 in five benchmarks.
  arXiv  Detail & Related papers  (2026-02-06T08:24:59Z)
• Towards a Science of Scaling Agent Systems [79.64446272302287]
  We formalize a definition for agent evaluation and characterize scaling laws as the interplay between agent quantity, coordination structure, modelic, and task properties.<n>We derive a predictive model using coordination metrics, that cross-validated R2=0, enabling prediction on unseen task domains.<n>We identify three effects: (1) a tool-coordination trade-off: under fixed computational budgets, tool-heavy tasks suffer disproportionately from multi-agent overhead, and (2) a capability saturation: coordination yields diminishing or negative returns once single-agent baselines exceed 45%.
  arXiv  Detail & Related papers  (2025-12-09T06:52:21Z)
• Teaching Language Models to Reason with Tools [73.21700643314917]
  We present emphHint-Engineering, a new data synthesis strategy that strategically injects diverse hints at optimal points within reasoning paths.<n>CoRT significantly enhances efficiency, reducing token usage by approximately 30% for the 32B model and 50% for the 1.5B model.
  arXiv  Detail & Related papers  (2025-10-23T08:41:44Z)
• ReSeFlow: Rectifying SE(3)-Equivariant Policy Learning Flows [7.360373380580255]
  We introduce the rectification to the SE(3)-diffusion models and propose the ReSeFlow, providing fast, geodesic-consistent, least-computational policy generation.<n>We find that the proposed ReSeFlow with only one inference step can achieve better performance with lower geodesic distance than the baseline methods.
  arXiv  Detail & Related papers  (2025-09-20T06:32:36Z)
• Too Helpful, Too Harmless, Too Honest or Just Right? [19.134202394422285]
  Large Language Models (LLMs) exhibit strong performance across a wide range of NLP tasks.<n> aligning their outputs with the principles of Helpfulness, Harmlessness, and Honesty (HHH) remains a persistent challenge.<n>We propose TrinityX, a modular alignment framework that incorporates a Mixture of Calibrated Experts (MoCaE) within the Transformer architecture.
  arXiv  Detail & Related papers  (2025-09-10T10:51:47Z)
• TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis [35.2545408706656]
  Large Language Models (LLMs) excel in various natural language processing tasks but remain vulnerable to generating harmful content or being exploited for malicious purposes.<n>We propose a novel analysis framework to measure the risk coverage of alignment datasets across three essential dimensions: Lexical Diversity, Malicious Intent, and Jailbreak Tactics.
  arXiv  Detail & Related papers  (2025-05-30T15:02:21Z)
• Multimodal Web Navigation with Instruction-Finetuned Foundation Models [99.14209521903854]
  We study data-driven offline training for web agents with vision-language foundation models. We propose an instruction-following multimodal agent, WebGUM, that observes both webpage screenshots and HTML pages. We empirically demonstrate this recipe improves the agent's ability of grounded multimodal perception, HTML comprehension, and multi-step reasoning.
  arXiv  Detail & Related papers  (2023-05-19T17:44:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.