REVIEW 8 cited by
Multimodal Web Navigation with Instruction-Finetuned Foundation Models
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
The progress of autonomous web navigation has been hindered by the dependence on billions of exploratory interactions via online reinforcement learning, and domain-specific model designs that make it difficult to leverage generalization from rich out-of-domain data. In this work, 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 and outputs web navigation actions, such as click and type. WebGUM is trained by jointly finetuning an instruction-finetuned language model and a vision encoder with temporal and local perception on a large corpus of demonstrations. We empirically demonstrate this recipe improves the agent's ability of grounded multimodal perception, HTML comprehension, and multi-step reasoning, outperforming prior works by a significant margin. On the MiniWoB, we improve over the previous best offline methods by more than 45.8%, even outperforming online-finetuned SoTA, humans, and GPT-4-based agent. On the WebShop benchmark, our 3-billion-parameter model achieves superior performance to the existing SoTA, PaLM-540B. Furthermore, WebGUM exhibits strong positive transfer to the real-world planning tasks on the Mind2Web. We also collect 347K high-quality demonstrations using our trained models, 38 times larger than prior work, and make them available to promote future research in this direction.
Forward citations
Cited by 8 Pith papers
-
Graph-Enhanced Policy Optimization in LLM Agent Training
GEPO adds graph-centrality-based intrinsic rewards, dynamic discounts, and two-level advantage shaping to group-based RL, improving LLM agent success on ALFWorld, WebShop, and a private Workbench benchmark.
-
MMBench-GUI: Hierarchical Multi-Platform Evaluation Framework for GUI Agents
MMBench-GUI provides a multi-platform, four-level benchmark and an efficiency-aware metric, and its experiments indicate visual grounding is the main bottleneck for current GUI agents.
-
Build the web for agents, not agents for the web
The paper proposes a paradigm shift: design a standardized Agentic Web Interface for AI agents, rather than adapting agents to human-facing websites.
-
Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction
Scaling the number of interaction steps, trained via a curriculum over rollout horizon, improves web-agent task success and outperforms scaling per-step reasoning under fixed token budgets.
-
TAPO: Transition-Aware Policy Optimization for LLM Agents
Interleaving action-conditioned next-observation supervision with group RL on a shared LLM backbone consistently lifts long-horizon agent success over pure policy optimization.
-
Turbocharging Web Automation: The Impact of Compressed History States
A Perceiver-style history compressor that summarizes past web states into 256-token representations improves Mind2Web and WebLINX web-automation accuracy by 1.2-5.4% over a no-history baseline.
-
WebArXiv: Evaluating Multimodal Agents on Time-Invariant arXiv Tasks
WebArXiv is a time-invariant 275-task benchmark for multimodal web agents on arXiv, plus a dynamic-reflection prompting method that modestly improves success rates.
-
LineRetriever: Planning-Aware Observation Reduction for Web Agents
LineRetriever uses a small LM to select relevant lines from web page observations, cutting context by up to 73% with only small success-rate drops on web agent benchmarks.
Discussion (0). Continue with ORCID to comment.