Datasets:

hulx2002
/

VisDocAgentBench

+# Licensing
+VisDocAgentBench combines benchmark-authored annotations with page images derived from independently licensed source documents. These components do not share one blanket license.
+## Benchmark-Authored Material
+The following files are licensed under the [Creative Commons Attribution 4.0 International License](https://creativecommons.org/licenses/by/4.0/):
+- `benchmark/queries.jsonl`
+- `benchmark/evaluator_annotations.jsonl`
+- `benchmark/topics.json`
+- `corpus/documents.jsonl`
+- `corpus/pages.jsonl`
+- `dataset_info.json`
+Copyright 2026 the VisDocAgentBench authors.
+## Source-Document Page Images
+Every page image under `corpus/pages/` remains governed by the license recorded in the corresponding row of `corpus/documents.jsonl`. The manifest also supplies the source title, authors, exact arXiv version, and source URL needed for attribution.
+The distributed images come from:
+- 55 documents licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/); and
+- 4 documents licensed under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/).
+The noncommercial and share-alike conditions of CC BY-NC-SA 4.0 continue to apply to those four sources. Users must inspect `license_url` in `corpus/documents.jsonl` before reusing an image.
+## Metadata-Only Sources
+Page images are not redistributed for:
+- 35 documents under the [arXiv nonexclusive distribution license](https://arxiv.org/licenses/nonexclusive-distrib/1.0/); and
+- 6 documents under [CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/).
+Their benchmark metadata and annotations remain available. The code repository provides a local reconstruction utility that downloads the exact source version from arXiv for evaluation. Local use and any subsequent redistribution remain subject to the source license.

README.md ADDED Viewed

	@@ -0,0 +1,140 @@

+---
+license: other
+license_name: composite
+pretty_name: VisDocAgentBench
+language:
+- en
+tags:
+- visual-document-retrieval
+- agentic-search
+- multimodal-retrieval
+- scientific-documents
+size_categories:
+- 1K<n<10K
+configs:
+- config_name: queries
+  data_files:
+  - split: test
+    path: benchmark/queries.jsonl
+- config_name: evaluator_annotations
+  data_files:
+  - split: test
+    path: benchmark/evaluator_annotations.jsonl
+- config_name: documents
+  data_files:
+  - split: corpus
+    path: corpus/documents.jsonl
+- config_name: pages
+  data_files:
+  - split: corpus
+    path: corpus/pages.jsonl
+---
+# VisDocAgentBench
+VisDocAgentBench is a closed-corpus benchmark for visually rich document retrieval. It contains 120 natural-language queries over 2,375 rendered pages from 100 scientific documents. Queries are evenly divided among direct, one-bridge, and two-bridge evidence structures, with one answer page per query.
+[[Paper](https://arxiv.org/pdf/2608.17889)] [[Code](https://github.com/hulx2002/VisDocAgentBench)] [[Project page](https://hulx2002.github.io/VisDocAgentBench)]
+## Contents
+```text
+benchmark/
+├── queries.jsonl
+├── evaluator_annotations.jsonl
+└── topics.json
+corpus/
+├── documents.jsonl
+├── pages.jsonl
+└── pages/<document_id>/<page_id>.png
+dataset_info.json
+LICENSES.md
+```
+- `queries.jsonl` contains the query text and topic identifier presented to a retrieval system.
+- `evaluator_annotations.jsonl` contains the answer page, evidence level, and ordered latent support pages used by the evaluator and controlled analyses.
+- `documents.jsonl` records the exact arXiv version, bibliographic metadata, source URLs, source license, page count, and image-availability status for every document.
+- `pages.jsonl` defines all 2,375 page identifiers, document membership, one-based page indices, rendering dimensions, and expected local paths.
+- `topics.json` describes the ten corpus topics.
+The standard agent harness reads the query text but does not expose answer or support annotations to the planner.
+## Dataset Statistics
+| Statistic | Count |
+|---|---:|
+| Documents | 100 |
+| Rendered pages | 2,375 |
+| Queries | 120 |
+| Direct queries | 40 |
+| One-bridge queries | 40 |
+| Two-bridge queries | 40 |
+| Unique answer pages | 120 |
+| Directly included page images | 1,469 |
+| Locally reconstructed page images | 906 |
+## Loading the Metadata
+Each JSONL component has a distinct schema and can be loaded independently:
+```python
+from datasets import load_dataset
+queries = load_dataset(
+    "hulx2002/VisDocAgentBench",
+    "queries",
+    split="test",
+)
+annotations = load_dataset(
+    "hulx2002/VisDocAgentBench",
+    "evaluator_annotations",
+    split="test",
+)
+pages = load_dataset(
+    "hulx2002/VisDocAgentBench",
+    "pages",
+    split="corpus",
+)
+```
+The complete snapshot can be downloaded with the code repository:
+```bash
+python scripts/download_data.py
+```
+## Reconstructing the Complete Corpus
+The repository includes 1,469 rendered page images whose source licenses permit redistribution. The remaining 906 rows remain in `corpus/pages.jsonl` with `image_included=false`. The exact source version and PDF URL are recorded in `corpus/documents.jsonl`.
+After cloning the [code repository](https://github.com/hulx2002/VisDocAgentBench) and downloading this dataset into `data/`, reconstruct the omitted pages with:
+```bash
+python dataset_tools/download_and_render.py
+python dataset_tools/validate_dataset.py --require-complete-corpus
+```
+The script downloads each specified arXiv version and renders it at 144 DPI. It verifies the expected page count and dimensions before accepting the reconstructed corpus.
+## Evaluation
+Systems return a ranked list of page identifiers or the opaque page handles assigned by the released agent harness. The evaluator reports Recall@1/3/5/10 and MRR@10 overall and by evidence level. Missing or invalid rankings stay in the 120-query denominator and score zero.
+Evaluation code, baseline implementations, deterministic preprocessing, and the full agent tool interfaces are available in the [code repository](https://github.com/hulx2002/VisDocAgentBench). Generated predictions, traces, OCR caches, embeddings, and model weights are not included in this dataset repository.
+## Licensing
+Benchmark-authored queries, evaluator annotations, topics, and corpus metadata are licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). Source-document page images retain the license of their source document. The license URL and attribution metadata for every source are recorded in `corpus/documents.jsonl`.
+Page images are included only for 55 CC BY 4.0 documents and 4 CC BY-NC-SA 4.0 documents. Sources under the arXiv nonexclusive distribution license or CC BY-NC-ND 4.0 are represented by metadata and local reconstruction instructions, not redistributed images. See [LICENSES.md](LICENSES.md) before reusing source pages.
+## Citation
+```bibtex
+@article{hu2026visdocagentbench,
+  title={VisDocAgentBench: Benchmarking Agents for Visually Rich Document Retrieval},
+  author={Hu, Lexiang and Zhang, Yanzhao and Li, Mingxin and Long, Dingkun and Li, Yikang and Zhang, Fuwei and Wang, Yisen and Lin, Zhouchen},
+  journal={arXiv preprint arXiv:2608.17889},
+  year={2026}
+}
+```

benchmark/evaluator_annotations.jsonl ADDED Viewed

	@@ -0,0 +1,120 @@

+{"answer_page_id": "arxiv_2606_02800_p039", "level": 1, "query_id": "visdocagentbench_q001", "support_page_ids": [], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p010", "level": 1, "query_id": "visdocagentbench_q002", "support_page_ids": [], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p066", "level": 1, "query_id": "visdocagentbench_q003", "support_page_ids": [], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p062", "level": 1, "query_id": "visdocagentbench_q004", "support_page_ids": [], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_11078_p008", "level": 1, "query_id": "visdocagentbench_q005", "support_page_ids": [], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2602_20502_p001", "level": 1, "query_id": "visdocagentbench_q006", "support_page_ids": [], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2604_27253_p003", "level": 1, "query_id": "visdocagentbench_q007", "support_page_ids": [], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2602_20502_p004", "level": 1, "query_id": "visdocagentbench_q008", "support_page_ids": [], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2603_26211_p014", "level": 1, "query_id": "visdocagentbench_q009", "support_page_ids": [], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2605_17965_p017", "level": 1, "query_id": "visdocagentbench_q010", "support_page_ids": [], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2602_06090_p009", "level": 1, "query_id": "visdocagentbench_q011", "support_page_ids": [], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2604_17288_p001", "level": 1, "query_id": "visdocagentbench_q012", "support_page_ids": [], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2605_17965_p018", "level": 1, "query_id": "visdocagentbench_q013", "support_page_ids": [], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2605_17641_p005", "level": 1, "query_id": "visdocagentbench_q014", "support_page_ids": [], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2606_03099_p012", "level": 1, "query_id": "visdocagentbench_q015", "support_page_ids": [], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2606_03099_p004", "level": 1, "query_id": "visdocagentbench_q016", "support_page_ids": [], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2602_24134_p011", "level": 1, "query_id": "visdocagentbench_q017", "support_page_ids": [], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2602_24134_p025", "level": 1, "query_id": "visdocagentbench_q018", "support_page_ids": [], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2603_01666_p012", "level": 1, "query_id": "visdocagentbench_q019", "support_page_ids": [], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2602_19961_p004", "level": 1, "query_id": "visdocagentbench_q020", "support_page_ids": [], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2602_24134_p009", "level": 1, "query_id": "visdocagentbench_q021", "support_page_ids": [], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2606_05145_p002", "level": 1, "query_id": "visdocagentbench_q022", "support_page_ids": [], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_22813_p032", "level": 1, "query_id": "visdocagentbench_q023", "support_page_ids": [], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2605_13986_p010", "level": 1, "query_id": "visdocagentbench_q024", "support_page_ids": [], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2605_13986_p008", "level": 1, "query_id": "visdocagentbench_q025", "support_page_ids": [], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2605_13986_p009", "level": 1, "query_id": "visdocagentbench_q026", "support_page_ids": [], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_13768_p010", "level": 1, "query_id": "visdocagentbench_q027", "support_page_ids": [], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_18478_p006", "level": 1, "query_id": "visdocagentbench_q028", "support_page_ids": [], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_23557_p002", "level": 1, "query_id": "visdocagentbench_q029", "support_page_ids": [], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2603_16085_p013", "level": 1, "query_id": "visdocagentbench_q030", "support_page_ids": [], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2603_16085_p022", "level": 1, "query_id": "visdocagentbench_q031", "support_page_ids": [], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_06891_p002", "level": 1, "query_id": "visdocagentbench_q032", "support_page_ids": [], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_19340_p008", "level": 1, "query_id": "visdocagentbench_q033", "support_page_ids": [], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_09864_p034", "level": 1, "query_id": "visdocagentbench_q034", "support_page_ids": [], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2605_23163_p004", "level": 1, "query_id": "visdocagentbench_q035", "support_page_ids": [], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_09864_p035", "level": 1, "query_id": "visdocagentbench_q036", "support_page_ids": [], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_24480_p014", "level": 1, "query_id": "visdocagentbench_q037", "support_page_ids": [], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2604_14609_p048", "level": 1, "query_id": "visdocagentbench_q038", "support_page_ids": [], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2605_17976_p021", "level": 1, "query_id": "visdocagentbench_q039", "support_page_ids": [], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2606_14870_p007", "level": 1, "query_id": "visdocagentbench_q040", "support_page_ids": [], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2606_02800_p011", "level": 2, "query_id": "visdocagentbench_q041", "support_page_ids": ["arxiv_2606_02800_p002"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p005", "level": 2, "query_id": "visdocagentbench_q042", "support_page_ids": ["arxiv_2606_02800_p080"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p038", "level": 2, "query_id": "visdocagentbench_q043", "support_page_ids": ["arxiv_2606_02800_p040"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_13672_p003", "level": 2, "query_id": "visdocagentbench_q044", "support_page_ids": ["arxiv_2606_13672_p025"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_18375_p005", "level": 2, "query_id": "visdocagentbench_q045", "support_page_ids": ["arxiv_2606_18375_p009"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_11078_p003", "level": 2, "query_id": "visdocagentbench_q046", "support_page_ids": ["arxiv_2606_11078_p019"], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2603_08013_p012", "level": 2, "query_id": "visdocagentbench_q047", "support_page_ids": ["arxiv_2603_08013_p008"], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2603_08533_p006", "level": 2, "query_id": "visdocagentbench_q048", "support_page_ids": ["arxiv_2603_08533_p004"], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2604_14609_p006", "level": 2, "query_id": "visdocagentbench_q049", "support_page_ids": ["arxiv_2604_14609_p009"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2603_15921_p003", "level": 2, "query_id": "visdocagentbench_q050", "support_page_ids": ["arxiv_2603_15921_p001"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2604_17288_p004", "level": 2, "query_id": "visdocagentbench_q051", "support_page_ids": ["arxiv_2604_17288_p005"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_22082_p025", "level": 2, "query_id": "visdocagentbench_q052", "support_page_ids": ["arxiv_2606_22082_p009"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2602_06090_p004", "level": 2, "query_id": "visdocagentbench_q053", "support_page_ids": ["arxiv_2602_06090_p002"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2605_17965_p028", "level": 2, "query_id": "visdocagentbench_q054", "support_page_ids": ["arxiv_2605_17965_p026"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_05761_p003", "level": 2, "query_id": "visdocagentbench_q055", "support_page_ids": ["arxiv_2606_05761_p005"], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2606_05761_p005", "level": 2, "query_id": "visdocagentbench_q056", "support_page_ids": ["arxiv_2606_05761_p003"], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2606_05761_p015", "level": 2, "query_id": "visdocagentbench_q057", "support_page_ids": ["arxiv_2606_05761_p024"], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2606_00832_p002", "level": 2, "query_id": "visdocagentbench_q058", "support_page_ids": ["arxiv_2606_00832_p003"], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2605_22829_p007", "level": 2, "query_id": "visdocagentbench_q059", "support_page_ids": ["arxiv_2605_22829_p006"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2602_22013_p008", "level": 2, "query_id": "visdocagentbench_q060", "support_page_ids": ["arxiv_2602_22013_p005"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2605_22829_p004", "level": 2, "query_id": "visdocagentbench_q061", "support_page_ids": ["arxiv_2605_22829_p006"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2603_01666_p002", "level": 2, "query_id": "visdocagentbench_q062", "support_page_ids": ["arxiv_2603_01666_p001"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2605_22829_p001", "level": 2, "query_id": "visdocagentbench_q063", "support_page_ids": ["arxiv_2605_22829_p008"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2606_05145_p026", "level": 2, "query_id": "visdocagentbench_q064", "support_page_ids": ["arxiv_2606_05145_p025"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_11576_p003", "level": 2, "query_id": "visdocagentbench_q065", "support_page_ids": ["arxiv_2606_11576_p002"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_05145_p039", "level": 2, "query_id": "visdocagentbench_q066", "support_page_ids": ["arxiv_2606_05145_p025"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_13768_p016", "level": 2, "query_id": "visdocagentbench_q067", "support_page_ids": ["arxiv_2606_13768_p015"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_13768_p004", "level": 2, "query_id": "visdocagentbench_q068", "support_page_ids": ["arxiv_2606_13768_p005"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_13768_p005", "level": 2, "query_id": "visdocagentbench_q069", "support_page_ids": ["arxiv_2606_13768_p015"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_14162_p004", "level": 2, "query_id": "visdocagentbench_q070", "support_page_ids": ["arxiv_2606_14162_p005"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_06891_p005", "level": 2, "query_id": "visdocagentbench_q071", "support_page_ids": ["arxiv_2606_06891_p007"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2603_16085_p003", "level": 2, "query_id": "visdocagentbench_q072", "support_page_ids": ["arxiv_2603_16085_p011"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_06891_p011", "level": 2, "query_id": "visdocagentbench_q073", "support_page_ids": ["arxiv_2606_06891_p007"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_22694_p007", "level": 2, "query_id": "visdocagentbench_q074", "support_page_ids": ["arxiv_2606_22694_p005"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_09864_p022", "level": 2, "query_id": "visdocagentbench_q075", "support_page_ids": ["arxiv_2606_09864_p003"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2605_21312_p019", "level": 2, "query_id": "visdocagentbench_q076", "support_page_ids": ["arxiv_2605_21312_p018"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_09864_p004", "level": 2, "query_id": "visdocagentbench_q077", "support_page_ids": ["arxiv_2606_09864_p006"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_13740_p007", "level": 2, "query_id": "visdocagentbench_q078", "support_page_ids": ["arxiv_2606_13740_p006"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_09864_p003", "level": 2, "query_id": "visdocagentbench_q079", "support_page_ids": ["arxiv_2606_09864_p008"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2604_14609_p013", "level": 2, "query_id": "visdocagentbench_q080", "support_page_ids": ["arxiv_2604_14609_p055"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2606_02800_p095", "level": 3, "query_id": "visdocagentbench_q081", "support_page_ids": ["arxiv_2606_02800_p085", "arxiv_2606_02800_p094"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p088", "level": 3, "query_id": "visdocagentbench_q082", "support_page_ids": ["arxiv_2606_02800_p022", "arxiv_2606_02800_p082"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_02800_p060", "level": 3, "query_id": "visdocagentbench_q083", "support_page_ids": ["arxiv_2606_02800_p085", "arxiv_2606_02800_p088"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2606_23296_p011", "level": 3, "query_id": "visdocagentbench_q084", "support_page_ids": ["arxiv_2606_19531_p019", "arxiv_2606_23296_p001"], "topic_id": "t01_embodied_world_models"}
+{"answer_page_id": "arxiv_2603_07432_p002", "level": 3, "query_id": "visdocagentbench_q085", "support_page_ids": ["arxiv_2606_06322_p004", "arxiv_2603_07432_p020"], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2606_11078_p024", "level": 3, "query_id": "visdocagentbench_q086", "support_page_ids": ["arxiv_2606_11078_p028", "arxiv_2606_11078_p013"], "topic_id": "t02_gui_computer_use_agents"}
+{"answer_page_id": "arxiv_2603_15921_p006", "level": 3, "query_id": "visdocagentbench_q087", "support_page_ids": ["arxiv_2603_15921_p017", "arxiv_2603_15921_p003"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2604_17288_p005", "level": 3, "query_id": "visdocagentbench_q088", "support_page_ids": ["arxiv_2604_17288_p007", "arxiv_2604_17288_p003"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2605_17965_p024", "level": 3, "query_id": "visdocagentbench_q089", "support_page_ids": ["arxiv_2605_17965_p036", "arxiv_2605_17965_p009"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2605_17965_p023", "level": 3, "query_id": "visdocagentbench_q090", "support_page_ids": ["arxiv_2605_17965_p005", "arxiv_2605_17965_p036"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_14948_p003", "level": 3, "query_id": "visdocagentbench_q091", "support_page_ids": ["arxiv_2606_14948_p016", "arxiv_2606_14948_p018"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_22082_p019", "level": 3, "query_id": "visdocagentbench_q092", "support_page_ids": ["arxiv_2606_22082_p009", "arxiv_2606_22082_p025"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_14948_p015", "level": 3, "query_id": "visdocagentbench_q093", "support_page_ids": ["arxiv_2606_14948_p016", "arxiv_2606_14948_p018"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2603_15921_p004", "level": 3, "query_id": "visdocagentbench_q094", "support_page_ids": ["arxiv_2603_15921_p017", "arxiv_2603_15921_p003"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_14948_p012", "level": 3, "query_id": "visdocagentbench_q095", "support_page_ids": ["arxiv_2606_14948_p016", "arxiv_2606_14948_p018"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_21633_p003", "level": 3, "query_id": "visdocagentbench_q096", "support_page_ids": ["arxiv_2606_13740_p011", "arxiv_2606_13740_p002"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2605_17965_p029", "level": 3, "query_id": "visdocagentbench_q097", "support_page_ids": ["arxiv_2605_17965_p032", "arxiv_2605_17965_p038"], "topic_id": "t03_software_engineering_code_agents"}
+{"answer_page_id": "arxiv_2606_05761_p046", "level": 3, "query_id": "visdocagentbench_q098", "support_page_ids": ["arxiv_2606_05761_p013", "arxiv_2606_05761_p007"], "topic_id": "t04_agent_memory_long_horizon"}
+{"answer_page_id": "arxiv_2605_22829_p015", "level": 3, "query_id": "visdocagentbench_q099", "support_page_ids": ["arxiv_2605_22829_p004", "arxiv_2605_22829_p007"], "topic_id": "t05_multimodal_rag_visual_documents"}
+{"answer_page_id": "arxiv_2606_22938_p009", "level": 3, "query_id": "visdocagentbench_q100", "support_page_ids": ["arxiv_2606_22938_p008", "arxiv_2606_22938_p007"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_22813_p027", "level": 3, "query_id": "visdocagentbench_q101", "support_page_ids": ["arxiv_2606_22813_p020", "arxiv_2606_22813_p049"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_22938_p016", "level": 3, "query_id": "visdocagentbench_q102", "support_page_ids": ["arxiv_2606_22938_p008", "arxiv_2606_22938_p007"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2605_13986_p018", "level": 3, "query_id": "visdocagentbench_q103", "support_page_ids": ["arxiv_2605_13986_p017", "arxiv_2605_13986_p046"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_22813_p048", "level": 3, "query_id": "visdocagentbench_q104", "support_page_ids": ["arxiv_2606_22813_p017", "arxiv_2606_22813_p026"], "topic_id": "t06_test_time_scaling_reasoning"}
+{"answer_page_id": "arxiv_2606_13035_p007", "level": 3, "query_id": "visdocagentbench_q105", "support_page_ids": ["arxiv_2606_04527_p002", "arxiv_2606_04527_p010"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_04432_p006", "level": 3, "query_id": "visdocagentbench_q106", "support_page_ids": ["arxiv_2606_18478_p021", "arxiv_2606_18478_p025"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_04432_p012", "level": 3, "query_id": "visdocagentbench_q107", "support_page_ids": ["arxiv_2606_18478_p021", "arxiv_2606_18478_p025"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_23610_p020", "level": 3, "query_id": "visdocagentbench_q108", "support_page_ids": ["arxiv_2606_23610_p002", "arxiv_2606_23610_p008"], "topic_id": "t07_video_generation_control"}
+{"answer_page_id": "arxiv_2606_19340_p007", "level": 3, "query_id": "visdocagentbench_q109", "support_page_ids": ["arxiv_2606_19340_p017", "arxiv_2606_19340_p016"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_23557_p007", "level": 3, "query_id": "visdocagentbench_q110", "support_page_ids": ["arxiv_2606_23557_p003", "arxiv_2606_23557_p006"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_23557_p011", "level": 3, "query_id": "visdocagentbench_q111", "support_page_ids": ["arxiv_2606_23557_p006", "arxiv_2606_23557_p037"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_22694_p006", "level": 3, "query_id": "visdocagentbench_q112", "support_page_ids": ["arxiv_2606_22694_p022", "arxiv_2606_22694_p005"], "topic_id": "t08_3d_spatial_intelligence"}
+{"answer_page_id": "arxiv_2606_09864_p006", "level": 3, "query_id": "visdocagentbench_q113", "support_page_ids": ["arxiv_2606_09864_p019", "arxiv_2606_09864_p030"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_09864_p027", "level": 3, "query_id": "visdocagentbench_q114", "support_page_ids": ["arxiv_2606_09864_p005", "arxiv_2606_09864_p059"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_09864_p053", "level": 3, "query_id": "visdocagentbench_q115", "support_page_ids": ["arxiv_2606_09864_p037", "arxiv_2606_09864_p040"], "topic_id": "t09_efficient_inference_kv_cache"}
+{"answer_page_id": "arxiv_2606_14870_p016", "level": 3, "query_id": "visdocagentbench_q116", "support_page_ids": ["arxiv_2606_14870_p021", "arxiv_2606_14870_p014"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2604_14609_p032", "level": 3, "query_id": "visdocagentbench_q117", "support_page_ids": ["arxiv_2604_14609_p008", "arxiv_2604_14609_p015"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2604_14609_p043", "level": 3, "query_id": "visdocagentbench_q118", "support_page_ids": ["arxiv_2604_14609_p027", "arxiv_2604_14609_p035"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2604_14609_p046", "level": 3, "query_id": "visdocagentbench_q119", "support_page_ids": ["arxiv_2604_14609_p027", "arxiv_2604_14609_p035"], "topic_id": "t10_scientific_foundation_models"}
+{"answer_page_id": "arxiv_2606_08688_p018", "level": 3, "query_id": "visdocagentbench_q120", "support_page_ids": ["arxiv_2606_08688_p006", "arxiv_2606_08688_p005"], "topic_id": "t10_scientific_foundation_models"}

benchmark/queries.jsonl ADDED Viewed

	@@ -0,0 +1,120 @@

+{"query": "Find the page where the look-aside buffer appears as a skipped-sample state in a stepwise packing table with colored sample boxes.", "query_id": "visdocagentbench_q001", "topic_id": "t01_embodied_world_models"}
+{"query": "Locate the page showing a compact token-style legend with solid and dashed boxes, where clean action tokens are encoded using a solid green style.", "query_id": "visdocagentbench_q002", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page where a ground-truth red path appears as colored reference marks within a top-row comparison grid of thumbnails and colored trajectory arcs.", "query_id": "visdocagentbench_q003", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page with a visual panel where colored frame markers appear within framed video snapshots above spectrogram bands.", "query_id": "visdocagentbench_q004", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page containing a compact checkmark-cross ablation table with numeric columns where intent masking appears as a binary condition in the table.", "query_id": "visdocagentbench_q005", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page where the observe-reason-act loop is visualized by repeated blocks in a two-row colored timeline comparison.", "query_id": "visdocagentbench_q006", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page with a horizontal workflow diagram with arrows and icons where “explored task tuples” appears as a labeled stage.", "query_id": "visdocagentbench_q007", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page where the Home Page State is shown as an annotated webpage screenshot with callout boxes.", "query_id": "visdocagentbench_q008", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page where a target-region miss is marked by offset colored boxes in a product-grid screenshot with colored boxes.", "query_id": "visdocagentbench_q009", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page containing the compact numeric comparison table with grouped columns where “Code-Aware w/ File Path ★” appears as a starred row.", "query_id": "visdocagentbench_q010", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page containing the wide ablation table with checkmarks and grouped metric columns where the sub-artifact feedback strategy is included as an ablation factor.", "query_id": "visdocagentbench_q011", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page containing the colored operation spectrum diagram with a difficulty arrow where “cycle shift” is shown inside one of the operation boxes.", "query_id": "visdocagentbench_q012", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page with the paired color-coded retrieval example schematic where “implicit path matches” labels the upper matching case.", "query_id": "visdocagentbench_q013", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page with a six-stage blue flowchart with boxed modules where “hybrid retriever” labels the second module.", "query_id": "visdocagentbench_q014", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page with a three-panel thumbnail retrieval workflow comparison where “locate-then-retrieve” labels the rightmost panel.", "query_id": "visdocagentbench_q015", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page containing a wide multi-stage pipeline diagram with dashed panels where “man's watch face” appears in the left user-query box.", "query_id": "visdocagentbench_q016", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page with a workflow figure where an RSU activity table supplies values highlighted in a yellow-highlighted answer span with a red arrow.", "query_id": "visdocagentbench_q017", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page where “revenues from external customers” labels the displayed numeric rows in a cropped numeric table beside reduced page screenshots.", "query_id": "visdocagentbench_q018", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page region where “late-interaction scoring” appears inside pseudocode blocks with blue stage labels.", "query_id": "visdocagentbench_q019", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page containing the side-by-side vector comparison schematic where “Single-Vec” labels the left panel.", "query_id": "visdocagentbench_q020", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page containing the smaller multi-block metrics table where “subset with bounding boxes” labels the first metric block.", "query_id": "visdocagentbench_q021", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page containing a two-panel plot with frontier curves and horizontal gain arrows where the retry-only baseline is drawn as a comparison curve.", "query_id": "visdocagentbench_q022", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page where the iterative update steps are shown as the only formula block on the page, formatted as a centered three-line mathematical update.", "query_id": "visdocagentbench_q023", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page containing a two-panel log-scale runtime line chart where cached test-row prediction appears as the yellow marker series.", "query_id": "visdocagentbench_q024", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page with a six-panel log-scale line-chart grid whose rightmost column is labeled \"n_features = 500\", with no separate subfigure titles.", "query_id": "visdocagentbench_q025", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page where the OOM frontier is visualized in the upper grouped log-scale memory bar chart.", "query_id": "visdocagentbench_q026", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page containing the stacked video-strip ablation comparison where a hard-cut schedule is illustrated by frame changes in the frame strips.", "query_id": "visdocagentbench_q027", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page region containing the wide two-line equation with dual underbraces, where “real-data regularizer” is labeled under the lower brace.", "query_id": "visdocagentbench_q028", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page with the side-by-side red failure and green correction diagram where the label “inconsistent cognitive map” appears in the red failure side.", "query_id": "visdocagentbench_q029", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page with the six-column rendering comparison grid with red annotations where the clothes hanger sweater case appears in the bottom rows.", "query_id": "visdocagentbench_q030", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page where a pink eyeglass case appears in the green-highlighted final output panels.", "query_id": "visdocagentbench_q031", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page containing the color-coded video timeline with query-response boxes where an orange response frame marks a response moment within the timeline.", "query_id": "visdocagentbench_q032", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page where “Cluttered Precise Pick-and-Place” appears as a row within a two-column grid of compact result tables.", "query_id": "visdocagentbench_q033", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page where a wide ruled summary table sits near the middle of the page, with model rows, sensitivity columns, and a \"single-layer flip\" column.", "query_id": "visdocagentbench_q034", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page where per-section token counts are shown in a compact count table above a monospaced code block.", "query_id": "visdocagentbench_q035", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page where the phrase “steep rise” appears as an annotation in three side-by-side small line charts with colored markers.", "query_id": "visdocagentbench_q036", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page containing an embedding scatter map with a zoomed cluster inset, where “Cluster 65” labels the zoomed inset of the map.", "query_id": "visdocagentbench_q037", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page where a red-yellow-green score scale is shown by a vertical percentage colorbar beside a grid, with essentially only the figure and caption on the page.", "query_id": "visdocagentbench_q038", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the LNP3 results page where a line chart is paired with stacked heatmaps, and the right-hand trajectory panels are labeled trajectory heatmaps.", "query_id": "visdocagentbench_q039", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page region with the centered, equation-numbered mathematical formula that defines the data-efficiency score.", "query_id": "visdocagentbench_q040", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page containing a color-coded paired tower schematic with token strips, where the method diagrammed as that schematic is the one under which “Dual-Stream Joint Attention” appears as an indented subsection.", "query_id": "visdocagentbench_q041", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page with the schematic where the method for which “structured token arrangements” are listed as an enabler is depicted by a central green block with modality icons.", "query_id": "visdocagentbench_q042", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page where a colored flowchart with three input boxes, buffer blocks, a selector, and an output box depicts the method whose pre-warm stage mitigates watchdog timeout.", "query_id": "visdocagentbench_q043", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page containing the colored architecture schematic with token strips where an off-the-shelf reward model supplies training labels for a method part that occupies the middle-right module of the schematic.", "query_id": "visdocagentbench_q044", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page with the colored architecture flowchart with stacked image inputs, where node-graph elements depict the method within which stochastic anchor sampling reduces relation-computation cost.", "query_id": "visdocagentbench_q045", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page containing the upper split schematic with robot critic icons, where the method depicted in the left side of that schematic is the method for which \"fast visual diff\" appears as a required reasoning step.", "query_id": "visdocagentbench_q046", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page containing the compact percentage results table with grouped model rows where the outcome whose unbounded count is normalized by a logarithmic dampening function appears as a metric column.", "query_id": "visdocagentbench_q047", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page containing the wide model-comparison results table with grouped metric columns where the benchmark evaluated in the table is the one for which the programming-distance criterion sets diversity constraints.", "query_id": "visdocagentbench_q048", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the page region where a method whose curriculum construction is constrained by a fixed order of tasks is illustrated by a side-by-side file-tree diagram with colored curved arrows.", "query_id": "visdocagentbench_q049", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page with the colored workflow diagram with code boxes and robot icons that illustrates the task for which “no observable failures” sets the diagnostic condition.", "query_id": "visdocagentbench_q050", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page where the free variable phi is introduced in a method that is broken into stages in a blue three-box pipeline with side tag labels.", "query_id": "visdocagentbench_q051", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page with the rounded gray summary callout box in which the method is summarized—the same method during which top-k design references are retrieved.", "query_id": "visdocagentbench_q052", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page containing the wide boxed workflow diagram with numbered modules and arrows, where an outcome appears as an early output in the diagram and the Origin container supplies hierarchy for that outcome.", "query_id": "visdocagentbench_q053", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page where the downstream APR baseline is instantiated by a system whose repaired-run overlap is shown in side-by-side three-circle overlap diagrams with numeric intersections.", "query_id": "visdocagentbench_q054", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page in the relevant paper that shows the bottom strip of four colored data cards, where a method is detailed by query, answer, and metadata cards, and the reference correct answer set is defined within that method.", "query_id": "visdocagentbench_q055", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page where the method, during whose construction relation non-disclosure is enforced, is formalized by centered mathematical equations within prose.", "query_id": "visdocagentbench_q056", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page with the blue boxed two-column dialogue example that illustrates a method during which the relation-critical context must remain unchanged.", "query_id": "visdocagentbench_q057", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Locate the page showing a colored architecture flowchart with tool and memory boxes, where the method whose operating context is bounded by the 128k-token context memory block is depicted as a central storage box.", "query_id": "visdocagentbench_q058", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page where the experiment for the benchmark—whose samples each include bounding-box annotations—is summarized by a wide top numeric table with grouped column bands and a highlighted row.", "query_id": "visdocagentbench_q059", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Locate the page region containing the compact numeric ablation table with grouped metric columns where the method that a clean/degraded image pair provides paired representations for is ablated in the table.", "query_id": "visdocagentbench_q060", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Locate the page with the colored multistage construction flowchart that depicts the construction pipeline for the benchmark in which bounding-box annotations are included in each sample.", "query_id": "visdocagentbench_q061", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page containing a horizontal four-panel method comparison schematic in which a method appears as a panel, and the text frames prohibitive storage overhead as motivating optimization via that method.", "query_id": "visdocagentbench_q062", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page with side-by-side colored document-processing workflow panels where the method for which 8–10k token usage is cited as the input range occupies the blue left-hand panel.", "query_id": "visdocagentbench_q063", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Find the page containing the numbered pseudocode block with phase headings where the method specified inside targets the first unresolved junction each round.", "query_id": "visdocagentbench_q064", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page containing the wide horizontal pipeline schematic with colored token stacks, where a method is depicted by multiple reasoning boxes and voting, and under that method hard-but-solvable questions receive larger rollout allocation.", "query_id": "visdocagentbench_q065", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page region containing the small round-wise results table with a paragraph note that summarizes the method which targets the first unresolved junction each round.", "query_id": "visdocagentbench_q066", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page with paired log-scale line charts with shaded regions where a result is plotted in the left chart and the Dirichlet-kernel identity supports the derivation of that result.", "query_id": "visdocagentbench_q067", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page where the method for which the image-query column marks a query region is illustrated in a large two-sided architecture schematic with colored token boxes.", "query_id": "visdocagentbench_q068", "topic_id": "t07_video_generation_control"}
+{"query": "Locate the page with the formula-centered derivation block beneath bullets whose equations specify the method parameterized by per-channel RoPE frequencies.", "query_id": "visdocagentbench_q069", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page containing the red-outlined distillation workflow block, where the method depicted inside the block is the one for which the frozen geometry-video joint teacher provides the score field.", "query_id": "visdocagentbench_q070", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page where “Response Timing Required” is marked as required for a benchmark, and that benchmark is summarized in right-side summary boxes with icons and counts.", "query_id": "visdocagentbench_q071", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page where a method is displayed within a small two-row comparison strip with red oval callouts, and structural holes appear in outputs from that method.", "query_id": "visdocagentbench_q072", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page containing the compact benchmark results table with grouped metrics where the benchmark being evaluated is the one for which “Response Timing Required” is marked as required.", "query_id": "visdocagentbench_q073", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page where an experiment is summarized by paired grouped bar-chart panels across perspective settings, for the benchmark in which a blue minivan appears as an example object.", "query_id": "visdocagentbench_q074", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page containing the green-to-red matrix heatmap with a side colorbar, where the setting that provides the column-wise experimental settings is the one within which “silent failure zone” labels a risky region.", "query_id": "visdocagentbench_q075", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page where the method within which the sticky long-history flag controls queue classification is rendered as a line-numbered pseudocode algorithm block.", "query_id": "visdocagentbench_q076", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page with side-by-side colored bar schematics with bracketed ranges where the left bar schematic depicts the outcome for which ConditionalFlip percentage quantifies flipped safety behavior.", "query_id": "visdocagentbench_q077", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page region showing the multi-step token confidence grid with colored letter cells that illustrates the method implemented by the three-stage transition mechanism.", "query_id": "visdocagentbench_q078", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page with a three-card schematic with a plot, circles, and flow boxes, where the rightmost card summarizes the method for which Single-Layer Flip is used as a validation measure.", "query_id": "visdocagentbench_q079", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page with the multi-method bar chart with a dashed reference line, where the method used to validate the active space setup appears as grouped green bars.", "query_id": "visdocagentbench_q080", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page with a diagram of the procedure whose first input is produced by a component required to return a single fenced, structured scene description.", "query_id": "visdocagentbench_q081", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the continuation of a specialized model's instruction template that turns an earlier field summary into a code-form specification for a structured visual-annotation format adopted to improve caption quality.", "query_id": "visdocagentbench_q082", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the public ranking display led by a model that has a dedicated template for the prompt-expansion component introduced through an example requiring one fenced structured response.", "query_id": "visdocagentbench_q083", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the page with the short comparison table for the middle of the three validation dimensions used by the benchmark whose task list includes Beat Block Hammer.", "query_id": "visdocagentbench_q084", "topic_id": "t01_embodied_world_models"}
+{"query": "Find the grouped example-strip page for the benchmark contrasted with one containing a brightness example labeled by an action that minimizes the display level.", "query_id": "visdocagentbench_q085", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Find the yellow-labeled failure-example page containing the error type that shares its name with the broader family of the failure illustrated by a request to list high-priority tasks.", "query_id": "visdocagentbench_q086", "topic_id": "t02_gui_computer_use_agents"}
+{"query": "Locate the results table comparing evaluation settings and their agreement for the task whose generated inputs pass through a boolean checker requested by a prompt.", "query_id": "visdocagentbench_q087", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the worked hardware example for the repair method constrained by a template approach that omits faults in event-sensitivity declarations.", "query_id": "visdocagentbench_q088", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the rounded highlighted observation summarizing the scoring process fed by the compact substitute for full source-code files, where the subsequent step is not equally consequential across the compared settings.", "query_id": "visdocagentbench_q089", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the rounded observation callout that reports the separate retrieval contribution of the two views introduced for a report containing the package-qualified identifier sklearn.linear_model.LogisticRegressionCV.", "query_id": "visdocagentbench_q090", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the workflow diagram containing the evaluator identified by the positive correlation between its scores and a listed aggregate of repository maintainability.", "query_id": "visdocagentbench_q091", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the compact numeric table that reports the component benefiting more strongly from a planning approach that retrieves a limited set of earlier designs, together with the other components of the same aggregate evaluation.", "query_id": "visdocagentbench_q092", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page of range tables used by the evaluator identified by the positive correlation between its scores and a listed aggregate of repository maintainability.", "query_id": "visdocagentbench_q093", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the short criteria list for the task whose generated tests are screened by the Boolean checker requested in a prompt.", "query_id": "visdocagentbench_q094", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the large bordered instruction block headed by the evaluator whose correlation chart places a file-scoped composite metric on the positive side, while the corresponding complexity-verdict chart shows the same metric as its sole negative entry.", "query_id": "visdocagentbench_q095", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page with a two-part decoding-process figure in which the paradigm extended by the component whose lowest latency occurs at four remaining tokens appears alongside its conventional counterpart.", "query_id": "visdocagentbench_q096", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the cream observation callout that states the effect of the intervention recommended for the repair failure illustrated by a generated patch that permits a single-generator case.", "query_id": "visdocagentbench_q097", "topic_id": "t03_software_engineering_code_agents"}
+{"query": "Find the page with a compact two-row result table for the evaluation condition under which the relation illustrated by a preference shift toward a denser historical style remains difficult.", "query_id": "visdocagentbench_q098", "topic_id": "t04_agent_memory_long_horizon"}
+{"query": "Find the page made up almost entirely of tabular comparisons, where the later ones report a setting distinguished in the evaluation of a benchmark built through screening and reformulating candidate queries.", "query_id": "visdocagentbench_q099", "topic_id": "t05_multimodal_rag_visual_documents"}
+{"query": "Locate the multi-chart page where a quantity used to characterize a method's benefit is plotted over training, for an analysis that begins from an initialization after supervised training.", "query_id": "visdocagentbench_q100", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Locate the proposition where a quantity derived after replacing a posterior over states with a tractable distribution is displayed with labels beneath its component terms.", "query_id": "visdocagentbench_q101", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the italicized formal statement in the later-training analysis that follows initialization from supervised examples, where the update is expressed through a state-dependent comparison quantity.", "query_id": "visdocagentbench_q102", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Locate the charts for the benchmark entry beside a task whose labels are produced by perturbing bins drawn from a Dirichlet distribution.", "query_id": "visdocagentbench_q103", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Find the page where the numbered derivation begins for the criterion an approach uses to compare the possible continuations shown as colored branches beneath a sequence of scene snapshots.", "query_id": "visdocagentbench_q104", "topic_id": "t06_test_time_scaling_reasoning"}
+{"query": "Locate the table comparing the same models across several generation durations. Its scores come from the benchmark used by the long-running setting illustrated with a striped-cat sequence.", "query_id": "visdocagentbench_q105", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page-width numeric comparison table near the top of the page whose grouped scores come from the evaluation suite applied to a held-out subset of the source collection from which a style-focused dataset was derived.", "query_id": "visdocagentbench_q106", "topic_id": "t07_video_generation_control"}
+{"query": "Locate a comparison plot for the benchmark used to assess the source collection from which a dataset focused on animated content was derived.", "query_id": "visdocagentbench_q107", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page with a glossary table that defines the evaluation dimension used for the edit category of the illustrated instruction ending with powder snow kicking up behind the car.", "query_id": "visdocagentbench_q108", "topic_id": "t07_video_generation_control"}
+{"query": "Find the page with a multi-row execution sequence marked by contrasting callouts for the application that extends a single-action manipulation pipeline with a recovery mechanism allowing four retries.", "query_id": "visdocagentbench_q109", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the workflow that follows a task example assembled from several viewpoints through a model rollout; the task's active reasoning is formulated by an approach introduced to improve credit assignment across a long trajectory.", "query_id": "visdocagentbench_q110", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the main model-comparison table for the evaluation used to analyze viewpoint-centered grounding in a method that builds a spatial record for each selected view before producing its final answer.", "query_id": "visdocagentbench_q111", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the results table whose grouped task column corresponds to the benchmark where the longest tested mixed-mode reasoning chain fails.", "query_id": "visdocagentbench_q112", "topic_id": "t08_3d_spatial_intelligence"}
+{"query": "Find the page with a paired line-chart display showing the downstream outcome attributed to an experimental option that assigns a separate range at each sequence position.", "query_id": "visdocagentbench_q113", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Locate the table comparing across evaluation regimes whether an intervention changes an outcome. The measure comes from a diagnostic that first examines how the effect varies over model depth.", "query_id": "visdocagentbench_q114", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the page with stacked compact tables for the model examined by extending an intervention from the input-side layers, where the measured behavior changes abruptly only near the far end.", "query_id": "visdocagentbench_q115", "topic_id": "t09_efficient_inference_kv_cache"}
+{"query": "Find the multi-panel distribution display used to assess whether the fixed training allowance is sufficient for the generative downstream task whose maximum run spans several tens of thousands of updates.", "query_id": "visdocagentbench_q116", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Locate the page collecting task prompts, including the dynamics step of a workflow coupled to calculations from the system that provides its main electronic-structure engine.", "query_id": "visdocagentbench_q117", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the table comparing model and run conditions for a task initialized through a mean-field electronic-structure optimization and judged by a small named set of criteria.", "query_id": "visdocagentbench_q118", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the per-task results table for the evaluation category covering a task whose initial structures are optimized with a mean-field electronic method and scored with a compact set of criteria.", "query_id": "visdocagentbench_q119", "topic_id": "t10_scientific_foundation_models"}
+{"query": "Find the page where displayed derivations complete the update cycle of a simulator connected to an initialized environment with a simple planar contact.", "query_id": "visdocagentbench_q120", "topic_id": "t10_scientific_foundation_models"}

benchmark/topics.json ADDED Viewed

	@@ -0,0 +1,42 @@

+[
+  {
+    "topic_id": "t01_embodied_world_models",
+    "topic_name": "Embodied World Models / World Action Models"
+  },
+  {
+    "topic_id": "t02_gui_computer_use_agents",
+    "topic_name": "GUI / Computer-Use Agents"
+  },
+  {
+    "topic_id": "t03_software_engineering_code_agents",
+    "topic_name": "Software Engineering / Code Agents"
+  },
+  {
+    "topic_id": "t04_agent_memory_long_horizon",
+    "topic_name": "Agent Memory / Long-Horizon Context Management"
+  },
+  {
+    "topic_id": "t05_multimodal_rag_visual_documents",
+    "topic_name": "Multimodal RAG / Visually Rich Document Understanding"
+  },
+  {
+    "topic_id": "t06_test_time_scaling_reasoning",
+    "topic_name": "Test-Time Scaling / Inference-Time Reasoning"
+  },
+  {
+    "topic_id": "t07_video_generation_control",
+    "topic_name": "Video Generation / Controllable Video Diffusion"
+  },
+  {
+    "topic_id": "t08_3d_spatial_intelligence",
+    "topic_name": "3D / Spatial Intelligence for Multimodal Models"
+  },
+  {
+    "topic_id": "t09_efficient_inference_kv_cache",
+    "topic_name": "Efficient LLM/VLM Inference and KV-Cache Optimization"
+  },
+  {
+    "topic_id": "t10_scientific_foundation_models",
+    "topic_name": "Scientific Foundation Models / AI for Materials and Physical Sciences"
+  }
+]

corpus/documents.jsonl ADDED Viewed

	@@ -0,0 +1,100 @@

+{"arxiv_id": "2601.18203", "arxiv_versioned_id": "2601.18203v2", "authors": ["ShunLiang Fu", "Yanxin Zhang", "Yixin Xiang", "Xiaoyu Du", "Jinhui Tang"], "categories": ["cs.IR"], "document_id": "arxiv_2601_18203", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 12, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2601.18203v2", "primary_category": "cs.IR", "published": "2026-01-26T06:38:25Z", "source_url": "https://arxiv.org/abs/2601.18203v2", "title": "DMAP: Human-Aligned Structural Document Map for Multimodal Document Understanding", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-01-27T03:18:17Z"}
+{"arxiv_id": "2602.06090", "arxiv_versioned_id": "2602.06090v1", "authors": ["Xiaoxuan Tang", "Jincheng Wang", "Liwei Luo", "Jingxuan Xu", "Sheng Zhou", "Dajun Chen", "Wei Jiang", "Yong Li"], "categories": ["cs.SE", "cs.AI", "cs.CV"], "document_id": "arxiv_2602_06090", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 16, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2602.06090v1", "primary_category": "cs.SE", "published": "2026-02-05T06:26:46Z", "source_url": "https://arxiv.org/abs/2602.06090v1", "title": "SVRepair: Structured Visual Reasoning for Automated Program Repair", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-02-05T06:26:46Z"}
+{"arxiv_id": "2602.06593", "arxiv_versioned_id": "2602.06593v1", "authors": ["Robert Hutter", "Michael Pradel"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2602_06593", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 21, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2602.06593v1", "primary_category": "cs.SE", "published": "2026-02-06T10:44:09Z", "source_url": "https://arxiv.org/abs/2602.06593v1", "title": "AgentStepper: Interactive Debugging of Software Development Agents", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-02-06T10:44:09Z"}
+{"arxiv_id": "2602.19961", "arxiv_versioned_id": "2602.19961v2", "authors": ["Yibo Yan", "Jiahao Huo", "Guanbo Feng", "Mingdong Ou", "Yi Cao", "Xin Zou", "Shuliang Liu", "Yuanhuiyi Lyu", "Yu Huang", "Jungang Li", "Kening Zheng", "Xu Zheng", "Philip S. Yu", "James Kwok", "Xuming Hu"], "categories": ["cs.CL", "cs.IR"], "document_id": "arxiv_2602_19961", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 35, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2602.19961v2", "primary_category": "cs.CL", "published": "2026-02-23T15:27:41Z", "source_url": "https://arxiv.org/abs/2602.19961v2", "title": "Unlocking Multimodal Document Intelligence: From Current Triumphs to Future Frontiers of Visual Document Retrieval", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-03-23T03:51:40Z"}
+{"arxiv_id": "2602.20502", "arxiv_versioned_id": "2602.20502v1", "authors": ["Hongbin Zhong", "Fazle Faisal", "Luis França", "Tanakorn Leesatapornwongsa", "Adriana Szekeres", "Kexin Rong", "Suman Nath"], "categories": ["cs.AI", "cs.LG"], "document_id": "arxiv_2602_20502", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 12, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2602.20502v1", "primary_category": "cs.AI", "published": "2026-02-24T03:03:18Z", "source_url": "https://arxiv.org/abs/2602.20502v1", "title": "ActionEngine: From Reactive to Programmatic GUI Agents via State Machine Memory", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-02-24T03:03:18Z"}
+{"arxiv_id": "2602.22013", "arxiv_versioned_id": "2602.22013v4", "authors": ["I-Hsiang Chen", "Yu-Wei Liu", "Tse-Yu Wu", "Yu-Chien Chiang", "Jen-Chien Yang", "Wei-Ting Chen"], "categories": ["cs.CV"], "document_id": "arxiv_2602_22013", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 10, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2602.22013v4", "primary_category": "cs.CV", "published": "2026-02-25T15:27:57Z", "source_url": "https://arxiv.org/abs/2602.22013v4", "title": "RobustVisRAG: Causality-Aware Vision-Based Retrieval-Augmented Generation under Visual Degradations", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-03-26T08:59:16Z"}
+{"arxiv_id": "2602.24134", "arxiv_versioned_id": "2602.24134v1", "authors": ["Zhengren Wang", "Dongsheng Ma", "Huaping Zhong", "Jiayu Li", "Wentao Zhang", "Bin Wang", "Conghui He"], "categories": ["cs.CV", "cs.CL"], "document_id": "arxiv_2602_24134", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 26, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2602.24134v1", "primary_category": "cs.CV", "published": "2026-02-27T16:09:38Z", "source_url": "https://arxiv.org/abs/2602.24134v1", "title": "AgenticOCR: Parsing Only What You Need for Efficient Retrieval-Augmented Generation", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-02-27T16:09:38Z"}
+{"arxiv_id": "2603.01666", "arxiv_versioned_id": "2603.01666v1", "authors": ["Yibo Yan", "Mingdong Ou", "Yi Cao", "Xin Zou", "Shuliang Liu", "Jiahao Huo", "Yu Huang", "James Kwok", "Xuming Hu"], "categories": ["cs.CL", "cs.IR"], "document_id": "arxiv_2603_01666", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 27, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.01666v1", "primary_category": "cs.CL", "published": "2026-03-02T09:55:00Z", "source_url": "https://arxiv.org/abs/2603.01666v1", "title": "Beyond the Grid: Layout-Informed Multi-Vector Retrieval with Parsed Visual Document Representations", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-03-02T09:55:00Z"}
+{"arxiv_id": "2603.07432", "arxiv_versioned_id": "2603.07432v1", "authors": ["Li Gu", "Zihuan Jiang", "Zhixiang Chi", "Huan Liu", "Ziqiang Wang", "Yuanhao Yu", "Glen Berseth", "Yang Wang"], "categories": ["cs.CV", "cs.CL", "cs.HC", "cs.LG"], "document_id": "arxiv_2603_07432", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 25, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.07432v1", "primary_category": "cs.CV", "published": "2026-03-08T03:08:11Z", "source_url": "https://arxiv.org/abs/2603.07432v1", "title": "Generalization in Online Reinforcement Learning for Mobile Agents", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-03-08T03:08:11Z"}
+{"arxiv_id": "2603.08013", "arxiv_versioned_id": "2603.08013v1", "authors": ["Yuxiang Chai", "Shunye Tang", "Han Xiao", "Rui Liu", "Hongsheng Li"], "categories": ["cs.AI"], "document_id": "arxiv_2603_08013", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 16, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2603.08013v1", "primary_category": "cs.AI", "published": "2026-03-09T06:41:32Z", "source_url": "https://arxiv.org/abs/2603.08013v1", "title": "PIRA-Bench: A Transition from Reactive GUI Agents to GUI-based Proactive Intent Recommendation Agents", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-03-09T06:41:32Z"}
+{"arxiv_id": "2603.08533", "arxiv_versioned_id": "2603.08533v2", "authors": ["Yiping Xie", "Song Chen", "Jingxuan Xing", "Wei Jiang", "Zekun Zhu", "Yingyao Wang", "Pi Bu", "Jun Song", "Yuning Jiang", "Bo Zheng"], "categories": ["cs.CV"], "document_id": "arxiv_2603_08533", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 16, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.08533v2", "primary_category": "cs.CV", "published": "2026-03-09T16:04:08Z", "source_url": "https://arxiv.org/abs/2603.08533v2", "title": "SecAgent: Efficient Mobile GUI Agent with Semantic Context", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-03-31T03:08:51Z"}
+{"arxiv_id": "2603.12823", "arxiv_versioned_id": "2603.12823v1", "authors": ["Xunzhuo Liu", "Bowei He", "Xue Liu", "Andy Luo", "Haichen Zhang", "Huamin Chen"], "categories": ["cs.CL", "cs.CV"], "document_id": "arxiv_2603_12823", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 16, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.12823v1", "primary_category": "cs.CL", "published": "2026-03-13T09:21:25Z", "source_url": "https://arxiv.org/abs/2603.12823v1", "title": "Adaptive Vision-Language Model Routing for Computer Use Agents", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-03-13T09:21:25Z"}
+{"arxiv_id": "2603.15921", "arxiv_versioned_id": "2603.15921v1", "authors": ["Srijan Bansal", "Jiao Fangkai", "Yilun Zhou", "Austin Xu", "Shafiq Joty", "Semih Yavuz"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2603_15921", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 17, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2603.15921v1", "primary_category": "cs.SE", "published": "2026-03-16T21:14:28Z", "source_url": "https://arxiv.org/abs/2603.15921v1", "title": "VIBEPASS: Can Vibe Coders Really Pass the Vibe Check?", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-03-16T21:14:28Z"}
+{"arxiv_id": "2603.16085", "arxiv_versioned_id": "2603.16085v1", "authors": ["Hui Shan", "Keyang Luo", "Ming Li", "Sizhe Zheng", "Yanwei Fu", "Zhen Chen", "Xiangru Huang"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2603_16085", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 25, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.16085v1", "primary_category": "cs.CV", "published": "2026-03-17T03:21:06Z", "source_url": "https://arxiv.org/abs/2603.16085v1", "title": "Interact3D: Compositional 3D Generation of Interactive Objects", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-03-17T03:21:06Z"}
+{"arxiv_id": "2603.17826", "arxiv_versioned_id": "2603.17826v1", "authors": ["Ruize Ma", "Yilei Jiang", "Shilin Zhang", "Zheng Ma", "Yi Feng", "Vincent Ng", "Zhi Wang", "Xiangyu Yue", "Chuanyi Li", "Lewei Lu"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2603_17826", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 15, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2603.17826v1", "primary_category": "cs.SE", "published": "2026-03-18T15:24:22Z", "source_url": "https://arxiv.org/abs/2603.17826v1", "title": "FailureMem: A Failure-Aware Multimodal Framework for Autonomous Software Repair", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-03-18T15:24:22Z"}
+{"arxiv_id": "2603.26211", "arxiv_versioned_id": "2603.26211v1", "authors": ["Shrinidhi Kumbhar", "Haofu Liao", "Srikar Appalaraju", "Kunwar Yashraj Singh"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2603_26211", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 14, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2603.26211v1", "primary_category": "cs.CV", "published": "2026-03-27T09:34:07Z", "source_url": "https://arxiv.org/abs/2603.26211v1", "title": "Towards GUI Agents: Vision-Language Diffusion Models for GUI Grounding", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-03-27T09:34:07Z"}
+{"arxiv_id": "2604.05012", "arxiv_versioned_id": "2604.05012v1", "authors": ["Oteo Mamo", "Olga Kogiou", "Hyunjin Yi", "Weikuan Yu"], "categories": ["cs.AR", "cs.AI"], "document_id": "arxiv_2604_05012", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 8, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2604.05012v1", "primary_category": "cs.AR", "published": "2026-04-06T16:00:39Z", "source_url": "https://arxiv.org/abs/2604.05012v1", "title": "Comparative Characterization of KV Cache Management Strategies for LLM Inference", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-04-06T16:00:39Z"}
+{"arxiv_id": "2604.14609", "arxiv_versioned_id": "2604.14609v1", "authors": ["Zijian Zhang", "Aiwei Yin", "Amaan Baweja", "Jiaru Bai", "Ignacio Gustin", "Varinia Bernales", "Alán Aspuru-Guzik"], "categories": ["cs.AI", "physics.comp-ph"], "document_id": "arxiv_2604_14609", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 65, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2604.14609v1", "primary_category": "cs.AI", "published": "2026-04-16T04:28:56Z", "source_url": "https://arxiv.org/abs/2604.14609v1", "title": "El Agente Forjador: Task-Driven Agent Generation for Quantum Simulation", "topic_id": "t10_scientific_foundation_models", "updated": "2026-04-16T04:28:56Z"}
+{"arxiv_id": "2604.15821", "arxiv_versioned_id": "2604.15821v1", "authors": ["Yuanchang Zhou", "Hongyu Wang", "Yiming Du", "Yan Wang", "Mingzhen Li", "Siyu Hu", "Xiangyu Zhang", "Weijian Liu", "Chen Wang", "Zhuoqiang Guo", "Long Wang", "Jingde Bu", "Yutong Lu", "Guangming Tan", "Weile Jia"], "categories": ["cs.DC", "cs.LG"], "document_id": "arxiv_2604_15821", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 11, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2604.15821v1", "primary_category": "cs.DC", "published": "2026-04-17T08:20:32Z", "source_url": "https://arxiv.org/abs/2604.15821v1", "title": "Breaking the Training Barrier of Billion-Parameter Universal Machine Learning Interatomic Potentials", "topic_id": "t10_scientific_foundation_models", "updated": "2026-04-17T08:20:32Z"}
+{"arxiv_id": "2604.17288", "arxiv_versioned_id": "2604.17288v1", "authors": ["Zizhang Luo", "Yansong Xu", "Runlin Guo", "Fan Cui", "Kexing Zhou", "Mile Xia", "Hongyuan Hou", "Yuhao Luo", "Yun Liang"], "categories": ["cs.AR", "cs.AI"], "document_id": "arxiv_2604_17288", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 10, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2604.17288v1", "primary_category": "cs.AR", "published": "2026-04-19T07:04:49Z", "source_url": "https://arxiv.org/abs/2604.17288v1", "title": "Clover: A Neural-Symbolic Agentic Harness with Stochastic Tree-of-Thoughts for Verified RTL Repair", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-04-19T07:04:49Z"}
+{"arxiv_id": "2604.23276", "arxiv_versioned_id": "2604.23276v1", "authors": ["Meizhu Liu", "Yassi Abbasi", "Matthew Rowe", "Michael Avendi", "Paul Li"], "categories": ["cs.CV", "cs.AI", "cs.CL"], "document_id": "arxiv_2604_23276", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 8, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2604.23276v1", "primary_category": "cs.CV", "published": "2026-04-25T12:40:13Z", "source_url": "https://arxiv.org/abs/2604.23276v1", "title": "Lightweight and Production-Ready PDF Visual Element Parsing", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-04-25T12:40:13Z"}
+{"arxiv_id": "2604.23941", "arxiv_versioned_id": "2604.23941v1", "authors": ["Hongxin Li", "Yuntao Chen", "Zhaoxiang Zhang"], "categories": ["cs.CV"], "document_id": "arxiv_2604_23941", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 20, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2604.23941v1", "primary_category": "cs.CV", "published": "2026-04-27T01:29:02Z", "source_url": "https://arxiv.org/abs/2604.23941v1", "title": "GoClick: Lightweight Element Grounding Model for Autonomous GUI Interaction", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-04-27T01:29:02Z"}
+{"arxiv_id": "2604.27253", "arxiv_versioned_id": "2604.27253v1", "authors": ["Fazle Elahi Faisal", "Qianhui Wu", "Baolin Peng", "Jianfeng Gao"], "categories": ["cs.AI"], "document_id": "arxiv_2604_27253", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 21, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2604.27253v1", "primary_category": "cs.AI", "published": "2026-04-29T22:57:35Z", "source_url": "https://arxiv.org/abs/2604.27253v1", "title": "AutoSurfer -- Teaching Web Agents through Comprehensive Surfing, Learning, and Modeling", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-04-29T22:57:35Z"}
+{"arxiv_id": "2604.27996", "arxiv_versioned_id": "2604.27996v2", "authors": ["Jackson Vonderhorst", "Kuangshi Ai", "Haichao Miao", "Shusen Liu", "Chaoli Wang"], "categories": ["cs.AI", "cs.GR", "cs.HC"], "document_id": "arxiv_2604_27996", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 5, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2604.27996v2", "primary_category": "cs.AI", "published": "2026-04-30T15:22:28Z", "source_url": "https://arxiv.org/abs/2604.27996v2", "title": "Exploring Interaction Paradigms for LLM Agents in Scientific Visualization", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-05-13T03:48:48Z"}
+{"arxiv_id": "2605.06460", "arxiv_versioned_id": "2605.06460v1", "authors": ["Weien Li", "Rui Song", "Zeyu Li", "Haochen Liu", "Gonghao Zhang", "Difan Jiao", "Zhenwei Tang", "Bowei He", "Haolun Wu", "Xue Liu", "Ye Yuan"], "categories": ["cs.LG"], "document_id": "arxiv_2605_06460", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 21, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.06460v1", "primary_category": "cs.LG", "published": "2026-05-07T15:51:06Z", "source_url": "https://arxiv.org/abs/2605.06460v1", "title": "MINER: Mining Multimodal Internal Representation for Efficient Retrieval", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-05-07T15:51:06Z"}
+{"arxiv_id": "2605.07467", "arxiv_versioned_id": "2605.07467v1", "authors": ["Tsuyoshi Okita"], "categories": ["cs.LG", "cs.AI", "cs.ET"], "document_id": "arxiv_2605_07467", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 17, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.07467v1", "primary_category": "cs.LG", "published": "2026-05-08T09:14:11Z", "source_url": "https://arxiv.org/abs/2605.07467v1", "title": "Physical Simulators as Do-Operators: Causal Discovery under Latent Confounders for AI-for-Science", "topic_id": "t10_scientific_foundation_models", "updated": "2026-05-08T09:14:11Z"}
+{"arxiv_id": "2605.13986", "arxiv_versioned_id": "2605.13986v2", "authors": ["Léo Grinsztajn", "Klemens Flöge", "Oscar Key", "Felix Birkel", "Philipp Jund", "Brendan Roof", "Mihir Manium", "Shi Bin Hoo", "Magnus Bühler", "Anurag Garg", "Dominik Safaric", "Jake Robertson", "Benjamin Jäger", "Simone Alessi", "Adrian Hayler", "Vladyslav Moroshan", "Lennart Purucker", "Philipp Singer", "Alan Arazi", "Julien Siems", "Jan Hendrik Metzen", "Georg Grab", "Nick Erickson", "Siyuan Guo", "Eliott Kalfon", "Simon Bing", "David Salinas", "Clara Cornu", "Lilly Charlotte Wehrhahn", "Diana Kriuchkova", "Kursat Kaya", "Lydia Sidhoum", "Marie Salmon", "Jerry Chen", "Madelon Hulsebos", "Yann LeCun", "Samuel Müller", "Bernhard Schölkopf", "Sauraj Gambhir", "Noah Hollmann", "Frank Hutter"], "categories": ["cs.LG", "stat.ML"], "document_id": "arxiv_2605_13986", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 83, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.13986v2", "primary_category": "cs.LG", "published": "2026-05-13T18:01:43Z", "source_url": "https://arxiv.org/abs/2605.13986v2", "title": "TabPFN-3: Technical Report", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-05-28T17:22:15Z"}
+{"arxiv_id": "2605.17641", "arxiv_versioned_id": "2605.17641v1", "authors": ["Saksham Sahai Srivastava"], "categories": ["cs.AI", "cs.CL"], "document_id": "arxiv_2605_17641", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 12, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.17641v1", "primary_category": "cs.AI", "published": "2026-05-17T20:21:55Z", "source_url": "https://arxiv.org/abs/2605.17641v1", "title": "Causal Intervention-Based Memory Selection for Long-Horizon LLM Agents", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-05-17T20:21:55Z"}
+{"arxiv_id": "2605.17965", "arxiv_versioned_id": "2605.17965v1", "authors": ["Md Afif Al Mamun", "Gias Uddin"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2605_17965", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 45, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.17965v1", "primary_category": "cs.SE", "published": "2026-05-18T07:20:13Z", "source_url": "https://arxiv.org/abs/2605.17965v1", "title": "BLAgent: Agentic RAG for File-Level Bug Localization", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-05-18T07:20:13Z"}
+{"arxiv_id": "2605.17976", "arxiv_versioned_id": "2605.17976v1", "authors": ["Xinzhe Yuan", "Zhuo Chen", "Jianshu Zhang", "Huan Xiong", "Nanyang Ye", "Yuqiang Li", "Qinying Gu"], "categories": ["cs.AI", "math.OC"], "document_id": "arxiv_2605_17976", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 31, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.17976v1", "primary_category": "cs.AI", "published": "2026-05-18T07:30:25Z", "source_url": "https://arxiv.org/abs/2605.17976v1", "title": "Unleashing LLMs in Bayesian Optimization: Preference-Guided Framework for Scientific Discovery", "topic_id": "t10_scientific_foundation_models", "updated": "2026-05-18T07:30:25Z"}
+{"arxiv_id": "2605.17985", "arxiv_versioned_id": "2605.17985v1", "authors": ["Chengjie Hong", "Feixiang He", "Yiheng Zeng", "Lulu Kang", "He Wang"], "categories": ["cs.LG", "cs.AI"], "document_id": "arxiv_2605_17985", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 22, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.17985v1", "primary_category": "cs.LG", "published": "2026-05-18T07:40:46Z", "source_url": "https://arxiv.org/abs/2605.17985v1", "title": "SAFE-SVD: Sensitivity-Aware Fidelity-Enforcing SVD for Physics Foundation Models", "topic_id": "t10_scientific_foundation_models", "updated": "2026-05-18T07:40:46Z"}
+{"arxiv_id": "2605.19943", "arxiv_versioned_id": "2605.19943v1", "authors": ["Amin Sghaier", "Ali Parviz", "Alexia Jolicoeur-Martineau"], "categories": ["cs.AI"], "document_id": "arxiv_2605_19943", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 14, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.19943v1", "primary_category": "cs.AI", "published": "2026-05-19T15:00:38Z", "source_url": "https://arxiv.org/abs/2605.19943v1", "title": "Probabilistic Tiny Recursive Model", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-05-19T15:00:38Z"}
+{"arxiv_id": "2605.21312", "arxiv_versioned_id": "2605.21312v2", "authors": ["Yicheng Feng", "Xin Tan", "Yangtao Deng", "Yimin Jiang", "Yibo Zhu", "Hong Xu"], "categories": ["cs.DC", "cs.AI", "cs.LG"], "document_id": "arxiv_2605_21312", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 20, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.21312v2", "primary_category": "cs.DC", "published": "2026-05-20T15:40:18Z", "source_url": "https://arxiv.org/abs/2605.21312v2", "title": "Frontier: Towards Comprehensive and Accurate LLM Inference Simulation", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-13T06:07:21Z"}
+{"arxiv_id": "2605.22829", "arxiv_versioned_id": "2605.22829v1", "authors": ["Yifan Zhu", "Yu Mi", "Yue Lu", "Yanchu Guan", "Zhixuan Chu"], "categories": ["cs.IR", "cs.AI"], "document_id": "arxiv_2605_22829", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 19, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.22829v1", "primary_category": "cs.IR", "published": "2026-04-18T05:04:49Z", "source_url": "https://arxiv.org/abs/2605.22829v1", "title": "LFRAG: Layout-oriented Fine-grained Retrieval-Augmented Generation on Multimodal Document Understanding", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-04-18T05:04:49Z"}
+{"arxiv_id": "2605.23057", "arxiv_versioned_id": "2605.23057v1", "authors": ["Aman Sunesh", "Ali Alshehhi", "Hivansh Dhakne"], "categories": ["cs.LG", "cs.CL", "cs.PF"], "document_id": "arxiv_2605_23057", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 11, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.23057v1", "primary_category": "cs.LG", "published": "2026-05-21T21:46:57Z", "source_url": "https://arxiv.org/abs/2605.23057v1", "title": "ModeSwitch-LLM: A Lightweight Phase-Aware Controller for Cross-Mode LLM Inference on a Single GPU", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-05-21T21:46:57Z"}
+{"arxiv_id": "2605.23163", "arxiv_versioned_id": "2605.23163v2", "authors": ["Kewei Zhang", "Jin Wang", "Sensen Gao", "Chengyue Wu", "Yulong Cao", "Songyang Han", "Boris Ivanovic", "Langechuan Liu", "Marco Pavone", "Song Han", "Daquan Zhou", "Enze Xie"], "categories": ["cs.CL"], "document_id": "arxiv_2605_23163", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 17, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.23163v2", "primary_category": "cs.CL", "published": "2026-05-22T02:31:32Z", "source_url": "https://arxiv.org/abs/2605.23163v2", "title": "Fast-dDrive: Efficient Block-Diffusion VLM for Autonomous Driving", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-05-25T07:32:46Z"}
+{"arxiv_id": "2605.27570", "arxiv_versioned_id": "2605.27570v1", "authors": ["Gabriele Cesa", "Thomas Hehn", "Aleix Torres-Camps", "Àlex Batlle Casellas", "Jordi Ros-Giralt", "Arash Behboodi", "Tribhuvanesh Orekondy"], "categories": ["cs.AI"], "document_id": "arxiv_2605_27570", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 22, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.27570v1", "primary_category": "cs.AI", "published": "2026-05-26T18:43:15Z", "source_url": "https://arxiv.org/abs/2605.27570v1", "title": "LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-05-26T18:43:15Z"}
+{"arxiv_id": "2605.29074", "arxiv_versioned_id": "2605.29074v1", "authors": ["Jiyao Zhang", "Mingxu Zhang", "Yitong Peng", "Haoxuan Liu", "Chenshuo Wang", "Yuxing Long", "Haoyang Huang", "Dongjiang Li", "Nan Duan", "Hui Shen", "Hao Dong"], "categories": ["cs.CV", "cs.RO"], "document_id": "arxiv_2605_29074", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 11, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2605.29074v1", "primary_category": "cs.CV", "published": "2026-05-27T20:28:56Z", "source_url": "https://arxiv.org/abs/2605.29074v1", "title": "Embodied3DBench: Benchmarking Low-Level Embodied Spatial Intelligence of Vision Language Models", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-05-27T20:28:56Z"}
+{"arxiv_id": "2605.29639", "arxiv_versioned_id": "2605.29639v1", "authors": ["Boyu Tan", "Jiarui Guo", "Zongwei Lv", "Hanbo Sun", "Tong Yang", "Kan Liu", "Xinfei Shi", "Zetao Hu", "Yaxin Yu", "Chi Zhang", "Jianning Zhang", "Xi Yang", "Wei Zhang", "Bo Cai", "Silu Zhou", "Xiyu Wang", "Na He", "Yinghao Yu", "Wending Bao", "Guiyang Huang", "Yuxing Yuan", "Juncheng Yin", "Nan Wang", "Lin Yang", "Zechao Zhang", "Lu Chen", "Guoding Li", "Tao Lan", "Lin Qu"], "categories": ["cs.OS"], "document_id": "arxiv_2605_29639", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 14, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.29639v1", "primary_category": "cs.OS", "published": "2026-05-28T09:07:06Z", "source_url": "https://arxiv.org/abs/2605.29639v1", "title": "RTP-LLM: High-Performance Alibaba LLM Inference Engine", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-05-28T09:07:06Z"}
+{"arxiv_id": "2605.30917", "arxiv_versioned_id": "2605.30917v1", "authors": ["Gyu-Hwung Cho", "Youngjune Lee", "Kiyoon Jeong", "Siyoung Lee", "Sanggyu Han", "Hervé Dejean", "Stéphane Clinchant", "Seung-won Hwang"], "categories": ["cs.IR", "cs.CV"], "document_id": "arxiv_2605_30917", "license_url": "https://creativecommons.org/licenses/by-nc-sa/4.0/", "page_count": 12, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2605.30917v1", "primary_category": "cs.IR", "published": "2026-05-29T07:01:45Z", "source_url": "https://arxiv.org/abs/2605.30917v1", "title": "Inference-Free Multimodal Learned Sparse Retrieval for Production-Scale Visual Document Search", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-05-29T07:01:45Z"}
+{"arxiv_id": "2606.00756", "arxiv_versioned_id": "2606.00756v1", "authors": ["Yannan Wang", "Longli Yang", "Zhen Liu", "Abhishek Kumar", "Carsten Maple"], "categories": ["cs.AI"], "document_id": "arxiv_2606_00756", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 19, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.00756v1", "primary_category": "cs.AI", "published": "2026-05-30T14:45:39Z", "source_url": "https://arxiv.org/abs/2606.00756v1", "title": "CoMIC: Collaborative Memory and Insights Circulation for Long-Horizon LLM Agents in Cloud-Edge Systems", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-05-30T14:45:39Z"}
+{"arxiv_id": "2606.00832", "arxiv_versioned_id": "2606.00832v1", "authors": ["Adril Putra Merin", "David Anugraha", "Ayu Purwarianti", "Genta Indra Winata"], "categories": ["cs.CL"], "document_id": "arxiv_2606_00832", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 9, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.00832v1", "primary_category": "cs.CL", "published": "2026-05-30T18:08:51Z", "source_url": "https://arxiv.org/abs/2606.00832v1", "title": "Momento: Evaluating Persistent Memory and Reasoning with Multi-Session Agentic Conversations", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-05-30T18:08:51Z"}
+{"arxiv_id": "2606.01613", "arxiv_versioned_id": "2606.01613v2", "authors": ["Kanwar Bharat Singh"], "categories": ["cs.IR", "cs.AI", "cs.MA"], "document_id": "arxiv_2606_01613", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 24, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.01613v2", "primary_category": "cs.IR", "published": "2026-06-01T03:01:51Z", "source_url": "https://arxiv.org/abs/2606.01613v2", "title": "TechRAG: Evidence-Gated Multimodal Agentic RAG for Technical Literature Reasoning", "topic_id": "t05_multimodal_rag_visual_documents", "updated": "2026-06-13T19:11:55Z"}
+{"arxiv_id": "2606.02355", "arxiv_versioned_id": "2606.02355v1", "authors": ["Zhongyu He", "Yuanfan Li", "Fei Huang", "Tianyu Chen", "Siyuan Chen", "Xingyang Li", "Meng Hsuan Yu", "Xiangrong Liu", "Leyi Wei", "Lu Pan", "Ke Zeng", "Xunliang Cai"], "categories": ["cs.AI", "cs.LG"], "document_id": "arxiv_2606_02355", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 15, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.02355v1", "primary_category": "cs.AI", "published": "2026-06-01T15:02:59Z", "source_url": "https://arxiv.org/abs/2606.02355v1", "title": "SIRI: Self-Internalizing Reinforcement Learning with Intrinsic Skills for LLM Agent Training", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-01T15:02:59Z"}
+{"arxiv_id": "2606.02800", "arxiv_versioned_id": "2606.02800v4", "authors": ["NVIDIA", ":", "Aditi", "Niket Agarwal", "Arslan Ali", "Jon Allen", "Martin Antolini", "Adeline Aubame", "Alisson Azzolini", "Junjie Bai", "Maciej Bala", "Yogesh Balaji", "Josh Bapst", "Aarti Basant", "Mukesh Beladiya", "Mohammad Qazim Bhat", "Zaid Pervaiz Bhat", "Dan Blick", "Vanni Brighella", "Han Cai", "Tiffany Cai", "Eric Cameracci", "Jiaxin Cao", "Yulong Cao", "Mark Carlson", "Carlos Casanova", "Ting-Yun Chang", "Yan Chang", "Yu-Wei Chao", "Prithvijit Chattopadhyay", "Roshan Chaudhari", "Chieh-Yun Chen", "Junyu Chen", "Ke Chen", "Qizhi Chen", "Wenkai Chen", "Xiaotong Chen", "Yu Chen", "An-Chieh Cheng", "Click Cheng", "Xiu Chia", "Jeana Choi", "Chaeyeon Chung", "Wenyan Cong", "Yin Cui", "Magdalena Dadela", "Nalin Dadhich", "Wenliang Dai", "Joyjit Daw", "Alperen Degirmenci", "Rodrigo Vieira Del Monte", "Robert Denomme", "Sameer Dharur", "Marco Di Lucca", "Ke Ding", "Wenhao Ding", "Yifan Ding", "Yuzhu Dong", "Nicole Drumheller", "Yilun Du", "Aigul Dzhumamuratova", "Aleksandr Efitorov", "Hamid Eghbalzadeh", "Naomi Eigbe", "Imad El Hanafi", "Hassan Eslami", "Benedikt Falk", "Jiaojiao Fan", "Jim Fan", "Amol Fasale", "Sergiy Fefilatyev", "Liang Feng", "Francesco Ferroni", "Sanja Fidler", "Xiao Fu", "Vikram Fugro", "Prashant Gaikwad", "TJ Galda", "Katelyn Gao", "Yihuai Gao", "Wenhang Ge", "Sreyan Ghosh", "Arushi Goel", "Vivek Goel", "Akash Gokul", "Rama Govindaraju", "Jinwei Gu", "Miguel Guerrero", "Elfie Guo", "Aryaman Gupta", "Siddharth Gururani", "Hugo Hadfield", "Song Han", "Ankur Handa", "Zekun Hao", "Mohammad Harrim", "Ali Hassani", "Nathan Hayes-Roth", "Yufan He", "Chris Helvig", "Cyrus Hogg", "Madison Huang", "Michael Huang", "Sophia Huang", "Yufan Huang", "Jacob Huffman", "DeLesley Hutchins", "Suneel Indupuru", "Boris Ivanovic", "Arihant Jain", "Joel Jang", "Ryan Ji", "Yanan Jian", "Dongfu Jiang", "Jingyi Jin", "Atharva Joshi", "Nikhilesh Joshi", "Pranjali Joshi", "Andy Ju", "Jaehun Jung", "Weiwei Kang", "Scott Kassekert", "Jan Kautz", "Ashna Khetan", "Julia Kiczka", "Slawek Kierat", "Gwanghyun Kim", "Kuno Kim", "Sunny Kim", "Kezhi Kong", "Xin Kong", "Zhifeng Kong", "Tomasz Kornuta", "Egor Krivov", "Hui Kuang", "Saurav Kumar", "Chia-Wen Kuo", "George Kurian", "Wojciech Kutak", "JF Lafleche", "Himangshu Lahkar", "Omar Laymoun", "Jayjun Lee", "Sanggil Lee", "Gabriele Leone", "Boyi Li", "Freya Li", "Jiajun Li", "Jinfeng Li", "Ling Li", "Pengcheng Li", "Shangru Li", "Tingle Li", "Xiaolong Li", "Xuan Li", "Zhaoshuo Li", "Zhiqi Li", "Hao Liang", "Maosheng Liao", "Chen-Hsuan Lin", "Tsung-Yi Lin", "Ming-Yu Liu", "Sifei Liu", "Zihan Liu", "Hai Loc Lu", "Xiangyu Lu", "Alice Luo", "Ruipu Luo", "Wenjie Luo", "Jiangran Lyu", "Martin Ding Ma", "Nic Ma", "Qianli Ma", "Dawid Majchrowski", "Louis Marcoux", "Miguel Martin", "Qing Miao", "Ashkan Mirzaei", "Shreyas Misra", "Kaichun Mo", "Durra Mohsin", "Hyejin Moon", "Pawel Morkisz", "Saeid Motiian", "Kirill Motkov", "Seungjun Nah", "Yashraj Narang", "Deepak Narayanan", "Thabang Ngazimbi", "Julian Ouyang", "Shubham Pachori", "David Page", "Yatian Pang", "Sehwi Park", "Mahesh Patekar", "Mostofa Patwary", "Marco Pavone", "Trung Pham", "Wei Ping", "Soha Pouya", "Shrimai Prabhumoye", "Varun Praveen", "Delin Qu", "Hesam Rabeti", "Morteza Ramezanali", "Marilyn Reeb", "Xuanchi Ren", "Kristen Rumley", "Wojciech Rymer", "Jun Saito", "Yeongho Seol", "John Shao", "Piyush Shekdar", "Tianwei Shen", "Humphrey Shi", "Min Shi", "Stella Shi", "Kevin Shih", "Mohammad Shoeybi", "Mateusz Sieniawski", "Shuran Song", "Alexander Sotelo", "Amir Sotoodeh", "Sunil Srinivasa", "Vignesh Srinivasakumar", "Bartosz Stefaniak", "Rahul Heinrich Steiger", "Shangkun Sun", "Jiaxiang Tang", "Shitao Tang", "Yangyang Tang", "Yue Tang", "Tolou Tavakkoli", "Kayley Ting", "Krzysztof Tomala", "Wei-Cheng Tseng", "Jibin Varghese", "Sergei Vasilev", "Thomas Volk", "Raju Wagwani", "Roger Waleffe", "Andrew Z. Wang", "Boxiang Wang", "Haoxiang Wang", "Qiao Wang", "Shihao Wang", "Shijie Wang", "Ting-Chun Wang", "Yan Wang", "Yu Wang", "Rohit Watve", "David Wehr", "Fangyin Wei", "Xinshuo Weng", "Jay Zhangjie Wu", "Kedi Wu", "Hongchi Xia", "Summer Xiao", "Tianjun Xiao", "Kevin Xie", "Daguang Xu", "Jiashu Xu", "Mengyao Xu", "Ruqing Xu", "Xingqian Xu", "Yao Xu", "Dinghao Yang", "Dong Yang", "Hans Yang", "Xiaodong Yang", "Xuning Yang", "Yichu Yang", "Yurong You", "Zhiding Yu", "Hao Yuan", "Simon Yuen", "Xiaohui Zeng", "Pengcuo Zeren", "Cindy Zha", "Haotian Zhang", "Jenny Zhang", "Jing Zhang", "Liangkai Zhang", "Paris Zhang", "Shun Zhang", "Xuanmeng Zhang", "Zhizheng Zhang", "Ann Zhao", "Yilin Zhao", "Yuliya Zhautouskaya", "Charles Zhou", "Fengzhe Zhou", "Shilin Zhu", "Yuke Zhu", "Dima Zhylko", "Artur Zolkowski"], "categories": ["cs.CV", "cs.AI", "cs.LG", "cs.MM", "cs.RO"], "document_id": "arxiv_2606_02800", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 139, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.02800v4", "primary_category": "cs.CV", "published": "2026-06-01T19:12:30Z", "source_url": "https://arxiv.org/abs/2606.02800v4", "title": "Cosmos 3: Omnimodal World Models for Physical AI", "topic_id": "t01_embodied_world_models", "updated": "2026-06-23T17:33:32Z"}
+{"arxiv_id": "2606.03099", "arxiv_versioned_id": "2606.03099v1", "authors": ["Kailin Lyu", "Zhiqiang Yuan", "Jianwei He", "Qiwei Yan", "Xuanbo Su", "Nanxing Hu", "Yang Liu", "Ce Hao", "Shengqian Qin", "Lianyu Hu", "Jinchao Zhang", "Jie Zhou"], "categories": ["cs.CL", "cs.AI"], "document_id": "arxiv_2606_03099", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 17, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.03099v1", "primary_category": "cs.CL", "published": "2026-06-02T03:38:44Z", "source_url": "https://arxiv.org/abs/2606.03099v1", "title": "PhotoCraft: Agentic Reasoning with Hierarchical Self-Evolving Memory for Deep Image Search", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-02T03:38:44Z"}
+{"arxiv_id": "2606.04432", "arxiv_versioned_id": "2606.04432v1", "authors": ["Thanh-Tung Le", "Yunhan Zhao", "Menglei Chai", "Zhengyang Shen", "Zhe Cao", "Danhang Tang", "Xiaohui Xie", "Deying Kong"], "categories": ["cs.CV"], "document_id": "arxiv_2606_04432", "license_url": "https://creativecommons.org/licenses/by-nc-sa/4.0/", "page_count": 13, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.04432v1", "primary_category": "cs.CV", "published": "2026-06-03T04:25:11Z", "source_url": "https://arxiv.org/abs/2606.04432v1", "title": "DSA: Dynamic Step Allocation for Fast Autoregressive Video Generation", "topic_id": "t07_video_generation_control", "updated": "2026-06-03T04:25:11Z"}
+{"arxiv_id": "2606.04527", "arxiv_versioned_id": "2606.04527v1", "authors": ["Yuxuan Bian", "Zeyue Xue", "Songchun Zhang", "Shiyi Zhang", "Weiyang Jin", "Yaowei Li", "Junhao Zhuang", "Haoran Li", "Jie Huang", "Haoyang Huang", "Nan Duan", "Qiang Xu"], "categories": ["cs.MM", "cs.CV", "cs.GR"], "document_id": "arxiv_2606_04527", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 21, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.04527v1", "primary_category": "cs.MM", "published": "2026-06-03T07:09:01Z", "source_url": "https://arxiv.org/abs/2606.04527v1", "title": "Echo-Infinity: Learning Evolving Memory for Real-Time Infinite Video Generation", "topic_id": "t07_video_generation_control", "updated": "2026-06-03T07:09:01Z"}
+{"arxiv_id": "2606.05145", "arxiv_versioned_id": "2606.05145v1", "authors": ["Nizar Islah", "Istabrak Abbes", "Irina Rish", "Sarath Chandar", "Eilif B. Muller"], "categories": ["cs.LG", "cs.AI", "cs.CL"], "document_id": "arxiv_2606_05145", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 43, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.05145v1", "primary_category": "cs.LG", "published": "2026-06-03T17:50:26Z", "source_url": "https://arxiv.org/abs/2606.05145v1", "title": "Failed Reasoning Traces Tell You What Is Fixable (But Not by Reading Them)", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-03T17:50:26Z"}
+{"arxiv_id": "2606.05761", "arxiv_versioned_id": "2606.05761v2", "authors": ["Wenxuan Wang", "Haoyu Sun", "Fukuan Hou", "Mingyang Song", "Weinan Zhang", "Yu Cheng", "Yang Yang"], "categories": ["cs.AI", "cs.CL"], "document_id": "arxiv_2606_05761", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 47, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.05761v2", "primary_category": "cs.AI", "published": "2026-06-04T06:43:11Z", "source_url": "https://arxiv.org/abs/2606.05761v2", "title": "SubtleMemory: A Benchmark for Fine-Grained Relational Memory Discrimination in Long-Horizon AI Agents", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-05T04:28:58Z"}
+{"arxiv_id": "2606.05773", "arxiv_versioned_id": "2606.05773v1", "authors": ["Chong Ma", "Taiyi Su", "Jian Zhu", "Jianjun Zhang", "Zitai Huang", "Yi Xu", "Hanli Wang"], "categories": ["cs.RO"], "document_id": "arxiv_2606_05773", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 14, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.05773v1", "primary_category": "cs.RO", "published": "2026-06-04T06:57:25Z", "source_url": "https://arxiv.org/abs/2606.05773v1", "title": "PiL-World: A Chunk-Wise World Model for VLA Policy-in-the-Loop Evaluation", "topic_id": "t01_embodied_world_models", "updated": "2026-06-04T06:57:25Z"}
+{"arxiv_id": "2606.06322", "arxiv_versioned_id": "2606.06322v1", "authors": ["Nathan Bout", "Maxime Langevin", "Ronan Riochet"], "categories": ["cs.AI"], "document_id": "arxiv_2606_06322", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 19, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.06322v1", "primary_category": "cs.AI", "published": "2026-06-04T15:57:29Z", "source_url": "https://arxiv.org/abs/2606.06322v1", "title": "DragOn: A Benchmark and Dataset for Drag-Based GUI Interactions", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-06-04T15:57:29Z"}
+{"arxiv_id": "2606.06891", "arxiv_versioned_id": "2606.06891v1", "authors": ["Hanxun Yu", "Xuan Qu", "Lei Ke", "Boqiang Zhang", "Yuxin Wang", "Jianke Zhu", "Dong Yu"], "categories": ["cs.CV"], "document_id": "arxiv_2606_06891", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 32, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.06891v1", "primary_category": "cs.CV", "published": "2026-06-05T04:16:24Z", "source_url": "https://arxiv.org/abs/2606.06891v1", "title": "Stream3D-VLM: Online 3D Spatial Understanding with Incremental Geometry Priors", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-05T04:16:24Z"}
+{"arxiv_id": "2606.06915", "arxiv_versioned_id": "2606.06915v2", "authors": ["Vladislav Smirnov", "Chieu Nguyen", "Sergey Senichev", "Minh Ngoc Ta", "Ekaterina Fadeeva", "Artem Vazhentsev", "Daria Galimzianova", "Nikolai Rozanov", "Viktor Mazanov", "Jingwei Ni", "Tianyi Wu", "Igor Kiselev", "Mrinmaya Sachan", "Iryna Gurevych", "Preslav Nakov", "Timothy Baldwin", "Artem Shelmanov"], "categories": ["cs.CL", "cs.AI", "cs.LG"], "document_id": "arxiv_2606_06915", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 15, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.06915v2", "primary_category": "cs.CL", "published": "2026-06-05T05:28:46Z", "source_url": "https://arxiv.org/abs/2606.06915v2", "title": "ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-08T11:00:40Z"}
+{"arxiv_id": "2606.08688", "arxiv_versioned_id": "2606.08688v1", "authors": ["Chunji Lv", "Jiaxi Ye", "Yuchen Jiang", "Rexar Lin", "Changsheng Li"], "categories": ["cs.RO", "cs.CV"], "document_id": "arxiv_2606_08688", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 20, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.08688v1", "primary_category": "cs.RO", "published": "2026-06-07T15:42:56Z", "source_url": "https://arxiv.org/abs/2606.08688v1", "title": "PhysAgent: Automating Physics-Based 4D Synthesis via Trajectory-Grounded Multi-Agent Feedback", "topic_id": "t10_scientific_foundation_models", "updated": "2026-06-07T15:42:56Z"}
+{"arxiv_id": "2606.08737", "arxiv_versioned_id": "2606.08737v1", "authors": ["Yunfan Lou", "Yifan Ye", "Yankai Fu", "Jun Cen", "Xiaowei Chi", "Yaoxu Lyu", "Peidong Jia", "Sirui Han", "Zhihe Lu", "Shanghang Zhang"], "categories": ["cs.RO"], "document_id": "arxiv_2606_08737", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 16, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.08737v1", "primary_category": "cs.RO", "published": "2026-06-07T17:18:23Z", "source_url": "https://arxiv.org/abs/2606.08737v1", "title": "Dream-Tac: A Unified Tactile World Action Model for Contact-Rich Robot Manipulation", "topic_id": "t01_embodied_world_models", "updated": "2026-06-07T17:18:23Z"}
+{"arxiv_id": "2606.09864", "arxiv_versioned_id": "2606.09864v1", "authors": ["Bruce Changlong Xu", "Adarsh Kumarappan", "Mu Zhou"], "categories": ["cs.LG", "cs.AI", "cs.ET"], "document_id": "arxiv_2606_09864", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 61, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.09864v1", "primary_category": "cs.LG", "published": "2026-06-01T02:02:20Z", "source_url": "https://arxiv.org/abs/2606.09864v1", "title": "Alignment Collapse Under KV Cache Quantization: Diagnosis and Mitigation", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-01T02:02:20Z"}
+{"arxiv_id": "2606.10183", "arxiv_versioned_id": "2606.10183v1", "authors": ["Konstantin Kuklev", "Viacheslav Vasilev", "Alexander Kunitsyn", "Andrei Ivaniuta", "Denis Dimitrov"], "categories": ["cs.CV", "cs.AI", "cs.MM"], "document_id": "arxiv_2606_10183", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 10, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.10183v1", "primary_category": "cs.CV", "published": "2026-06-08T21:21:01Z", "source_url": "https://arxiv.org/abs/2606.10183v1", "title": "Making Time Editable in Video Diffusion Transformers", "topic_id": "t07_video_generation_control", "updated": "2026-06-08T21:21:01Z"}
+{"arxiv_id": "2606.10662", "arxiv_versioned_id": "2606.10662v1", "authors": ["Yuzhen Mao", "Azalia Mirhoseini"], "categories": ["cs.MA", "cs.AI"], "document_id": "arxiv_2606_10662", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 23, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.10662v1", "primary_category": "cs.MA", "published": "2026-06-09T10:13:07Z", "source_url": "https://arxiv.org/abs/2606.10662v1", "title": "Decentralized Multi-Agent Systems with Shared Context", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-09T10:13:07Z"}
+{"arxiv_id": "2606.11078", "arxiv_versioned_id": "2606.11078v1", "authors": ["Jaewoo Lee", "Zaid Khan", "Archiki Prasad", "Justin Chih-Yao Chen", "Supriyo Chakraborty", "Kartik Balasubramaniam", "Sambit Sahu", "Elias Stengel-Eskin", "Hyunji Lee", "Mohit Bansal"], "categories": ["cs.AI", "cs.CL", "cs.CV"], "document_id": "arxiv_2606_11078", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 29, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.11078v1", "primary_category": "cs.AI", "published": "2026-06-09T16:39:10Z", "source_url": "https://arxiv.org/abs/2606.11078v1", "title": "A History-Aware Visually Grounded Critic for Computer Use Agents", "topic_id": "t02_gui_computer_use_agents", "updated": "2026-06-09T16:39:10Z"}
+{"arxiv_id": "2606.11576", "arxiv_versioned_id": "2606.11576v1", "authors": ["Ahmadreza Jeddi", "Minh Ngoc Le", "Amirhossein Kazerouni", "Hakki Can Karaimer", "Hue Nguyen", "Iqbal Mohomed", "Michael Brudno", "Alex Levinshtein", "Konstantinos G. Derpanis", "Babak Taati", "Radek Grzeszczuk"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2606_11576", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 30, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.11576v1", "primary_category": "cs.CV", "published": "2026-06-10T02:06:47Z", "source_url": "https://arxiv.org/abs/2606.11576v1", "title": "AVIS: Adaptive Test-Time Scaling for Vision-Language Models", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-10T02:06:47Z"}
+{"arxiv_id": "2606.13035", "arxiv_versioned_id": "2606.13035v1", "authors": ["Yu Meng", "Xiangyang Luo", "Letian Li", "Wenyuan Jiang", "Chen Gao", "Xinlei Chen", "Yong Li", "Xiao-Ping Zhang"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2606_13035", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 17, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.13035v1", "primary_category": "cs.CV", "published": "2026-06-11T08:16:08Z", "source_url": "https://arxiv.org/abs/2606.13035v1", "title": "TetherCache: Stabilizing Autoregressive Long-Form Video Generation with Gated Recall and Trusted Alignment", "topic_id": "t07_video_generation_control", "updated": "2026-06-11T08:16:08Z"}
+{"arxiv_id": "2606.13494", "arxiv_versioned_id": "2606.13494v1", "authors": ["Daichi Azuma", "Taiki Miyanishi", "Koya Sakamoto", "Shuhei Kurita", "Yaonan Zhu", "Petr Khrapchenkov", "Motoaki Kawanabe", "Yusuke Iwasawa", "Yutaka Matsuo"], "categories": ["cs.RO", "cs.CV"], "document_id": "arxiv_2606_13494", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 22, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.13494v1", "primary_category": "cs.RO", "published": "2026-06-11T15:44:36Z", "source_url": "https://arxiv.org/abs/2606.13494v1", "title": "NavWAM: A Navigation World Action Model for Goal-Conditioned Visual Navigation", "topic_id": "t01_embodied_world_models", "updated": "2026-06-11T15:44:36Z"}
+{"arxiv_id": "2606.13672", "arxiv_versioned_id": "2606.13672v2", "authors": ["Arnav Kumar Jain", "Yilin Wu", "Jesse Farebrother", "Gokul Swamy", "Andrea Bajcsy"], "categories": ["cs.RO"], "document_id": "arxiv_2606_13672", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 32, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.13672v2", "primary_category": "cs.RO", "published": "2026-06-11T17:59:15Z", "source_url": "https://arxiv.org/abs/2606.13672v2", "title": "WEAVER, Better, Faster, Longer: An Effective World Model for Robotic Manipulation", "topic_id": "t01_embodied_world_models", "updated": "2026-06-16T20:54:22Z"}
+{"arxiv_id": "2606.13740", "arxiv_versioned_id": "2606.13740v1", "authors": ["Tuowei Wang", "Yanfan Sun", "Ju Ren"], "categories": ["cs.LG"], "document_id": "arxiv_2606_13740", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 14, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.13740v1", "primary_category": "cs.LG", "published": "2026-06-11T12:44:57Z", "source_url": "https://arxiv.org/abs/2606.13740v1", "title": "Efficient On-Device Diffusion LLM Inference with Mobile NPU", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-11T12:44:57Z"}
+{"arxiv_id": "2606.13768", "arxiv_versioned_id": "2606.13768v2", "authors": ["Sharath Girish", "Tsai-Shien Chen", "Zhikang Dong", "Mukesh Singhal", "Hao Chen", "Sergey Tulyakov", "Aliaksandr Siarohin"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2606_13768", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 32, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.13768v2", "primary_category": "cs.CV", "published": "2026-06-11T17:59:10Z", "source_url": "https://arxiv.org/abs/2606.13768v2", "title": "CineOrchestra: Unified Entity-Centric Conditioning for Cinematic Video Generation", "topic_id": "t07_video_generation_control", "updated": "2026-06-16T17:59:39Z"}
+{"arxiv_id": "2606.14162", "arxiv_versioned_id": "2606.14162v1", "authors": ["Xunzhi Xiang", "Zixuan Duan", "Yabo Chen", "Zhengxuan Wei", "Guiyu Zhang", "Zixiao Gu", "Zhe Gao", "Haibin Huang", "Chi Zhang", "Qi Fan", "Xuelong Li"], "categories": ["cs.CV"], "document_id": "arxiv_2606_14162", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 17, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.14162v1", "primary_category": "cs.CV", "published": "2026-06-12T06:41:13Z", "source_url": "https://arxiv.org/abs/2606.14162v1", "title": "VideoWeave: Unlocking Geometric Consistency in Video Generation via Joint Geometry-Video Modeling", "topic_id": "t07_video_generation_control", "updated": "2026-06-12T06:41:13Z"}
+{"arxiv_id": "2606.14870", "arxiv_versioned_id": "2606.14870v1", "authors": ["Ibrahim Elsharkawy", "Joschka Birk", "Vinicius Mikuni", "Wahid Bhimji", "Gregor Kasieczka", "Benjamin Nachman"], "categories": ["hep-ph", "cs.LG"], "document_id": "arxiv_2606_14870", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 33, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.14870v1", "primary_category": "hep-ph", "published": "2026-06-12T18:15:04Z", "source_url": "https://arxiv.org/abs/2606.14870v1", "title": "Pre-Training for Simulation-Based Science: A Study on Jet Foundation Model Training Objectives", "topic_id": "t10_scientific_foundation_models", "updated": "2026-06-12T18:15:04Z"}
+{"arxiv_id": "2606.14948", "arxiv_versioned_id": "2606.14948v1", "authors": ["Kirill Vasilevski", "Ximing Dong", "Benjamin Rombaut", "Ruochen Deng", "Jiahuei Lin", "Arthur Leung", "Dayi Lin", "Boyuan Chen", "Shaowei Wang", "Ahmed E. Hassan"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2606_14948", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 19, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.14948v1", "primary_category": "cs.SE", "published": "2026-06-12T20:46:04Z", "source_url": "https://arxiv.org/abs/2606.14948v1", "title": "Beyond Correctness: Enhancing Architectural Reasoning in Code LLMs via Scalable Labeling with Agentic Judgment", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-06-12T20:46:04Z"}
+{"arxiv_id": "2606.15001", "arxiv_versioned_id": "2606.15001v1", "authors": ["Xiaoyu Wang", "Bingqing Cheng"], "categories": ["physics.comp-ph", "cond-mat.mtrl-sci", "cs.LG", "physics.chem-ph"], "document_id": "arxiv_2606_15001", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 15, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.15001v1", "primary_category": "physics.comp-ph", "published": "2026-06-12T22:45:20Z", "source_url": "https://arxiv.org/abs/2606.15001v1", "title": "Distilling latent electrostatics from foundation machine learning interatomic potentials", "topic_id": "t10_scientific_foundation_models", "updated": "2026-06-12T22:45:20Z"}
+{"arxiv_id": "2606.15032", "arxiv_versioned_id": "2606.15032v1", "authors": ["Yang Yu", "Shiyuan Zhang", "Yifei Sheng", "Haoxiang Ren", "Haoxin Lin"], "categories": ["cs.LG"], "document_id": "arxiv_2606_15032", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 24, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.15032v1", "primary_category": "cs.LG", "published": "2026-06-13T00:21:21Z", "source_url": "https://arxiv.org/abs/2606.15032v1", "title": "How Should World Models Be Evaluated? A Decision-Making-Centric Position", "topic_id": "t01_embodied_world_models", "updated": "2026-06-13T00:21:21Z"}
+{"arxiv_id": "2606.17257", "arxiv_versioned_id": "2606.17257v1", "authors": ["Rohit Kundu", "Arindam Dutta", "Sarosij Bose", "Athula Balachandran", "Amit K. Roy-Chowdhury"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2606_17257", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 17, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.17257v1", "primary_category": "cs.CV", "published": "2026-06-15T20:03:09Z", "source_url": "https://arxiv.org/abs/2606.17257v1", "title": "Pulling The REINS: Training-Free Safety Alignment of Video Diffusion Models via Representation Steering", "topic_id": "t07_video_generation_control", "updated": "2026-06-15T20:03:09Z"}
+{"arxiv_id": "2606.17803", "arxiv_versioned_id": "2606.17803v1", "authors": ["Vaggelis Dorovatas", "Nancy Kalaj", "Rahaf Aljundi"], "categories": ["cs.LG"], "document_id": "arxiv_2606_17803", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 17, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.17803v1", "primary_category": "cs.LG", "published": "2026-06-16T11:27:28Z", "source_url": "https://arxiv.org/abs/2606.17803v1", "title": "Continual Self-Improvement with Lightweight Experiential Latent Memories", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-16T11:27:28Z"}
+{"arxiv_id": "2606.18180", "arxiv_versioned_id": "2606.18180v1", "authors": ["Rongjin Guo", "Dong Liang", "Yuhao Liu", "Fang Liu", "Tianyu Huang", "Gerhard P. Hancke", "Rynson W. H. Lau"], "categories": ["cs.CV"], "document_id": "arxiv_2606_18180", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 16, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.18180v1", "primary_category": "cs.CV", "published": "2026-06-16T17:13:58Z", "source_url": "https://arxiv.org/abs/2606.18180v1", "title": "EgoCS-400K: An Egocentric Gameplay Dataset for World Models", "topic_id": "t01_embodied_world_models", "updated": "2026-06-16T17:13:58Z"}
+{"arxiv_id": "2606.18375", "arxiv_versioned_id": "2606.18375v3", "authors": ["Yuhang Huang", "Xuan Lv", "Junyan Xu", "Zhiyuan Yu", "Jiazhao Zhang", "Ruizhen Hu", "Wancheng Feng", "Shilong Zou", "Hewen Xiao", "Ziqiao Zhou", "Kaiyun Huang", "Zhiyu Peng", "Juzhan Xu", "Hang Zhao", "Chenyang Zhu", "Renjiao Yi", "Yifei Huang", "Douhui Wu", "Yan Zhang", "Kexu Cheng", "Chunhe Song", "Yunzhi Xue", "Xiuhong Zhang", "Leitao Guo", "Yunji Chen", "Bin Wu", "Haibin Yu", "Kai Xu"], "categories": ["cs.RO"], "document_id": "arxiv_2606_18375", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 20, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.18375v3", "primary_category": "cs.RO", "published": "2026-06-16T18:23:23Z", "source_url": "https://arxiv.org/abs/2606.18375v3", "title": "PAIWorld: A 3D-Consistent World Foundation Model for Robotic Manipulation", "topic_id": "t01_embodied_world_models", "updated": "2026-06-23T16:50:09Z"}
+{"arxiv_id": "2606.18478", "arxiv_versioned_id": "2606.18478v2", "authors": ["Siyi Chen", "Shaowei Liu", "Yixuan Jia", "Zian Wang", "Huan Ling", "Qing Qu", "Jun Gao"], "categories": ["cs.CV"], "document_id": "arxiv_2606_18478", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 34, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.18478v2", "primary_category": "cs.CV", "published": "2026-06-16T20:38:30Z", "source_url": "https://arxiv.org/abs/2606.18478v2", "title": "Data-Forcing Distillation: Restoring Diversity and Fidelity in Few-Step Video Generation", "topic_id": "t07_video_generation_control", "updated": "2026-06-23T08:04:55Z"}
+{"arxiv_id": "2606.18847", "arxiv_versioned_id": "2606.18847v1", "authors": ["Yehang Zhang", "Jianchong Su", "Haojian Huang", "Yifan Chang", "Tianhao Zhou", "Xinli Xu", "Yingjie Xu", "Yinchuan Li", "Zexi Li", "Ying-Cong Chen"], "categories": ["cs.AI"], "document_id": "arxiv_2606_18847", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 27, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.18847v1", "primary_category": "cs.AI", "published": "2026-06-17T09:26:26Z", "source_url": "https://arxiv.org/abs/2606.18847v1", "title": "WorldLines: Benchmarking and Modeling Long-Horizon Stateful Embodied Agents", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-17T09:26:26Z"}
+{"arxiv_id": "2606.19340", "arxiv_versioned_id": "2606.19340v2", "authors": ["Jisoo Kim", "Sangwon Baik", "Taeksoo Kim", "Sungjoo Kim", "Junyoung Lee", "Mingi Choi", "Hanbyul Joo"], "categories": ["cs.RO"], "document_id": "arxiv_2606_19340", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 21, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.19340v2", "primary_category": "cs.RO", "published": "2026-06-17T17:59:56Z", "source_url": "https://arxiv.org/abs/2606.19340v2", "title": "ZeroDex: Zero-Shot Long-Horizon Dexterous Manipulation via Multi-View 3D-Grounded VLM Reasoning", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-19T13:42:23Z"}
+{"arxiv_id": "2606.19531", "arxiv_versioned_id": "2606.19531v1", "authors": ["Yuyang Zhang", "Wenyao Zhang", "Zekun Qi", "He Zhang", "Haitao Lin", "Jingbo Zhang", "Yao Mu", "Xiaokang Yang", "Wenjun Zeng", "Xin Jin"], "categories": ["cs.CV", "cs.RO"], "document_id": "arxiv_2606_19531", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 19, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.19531v1", "primary_category": "cs.CV", "published": "2026-06-17T19:25:28Z", "source_url": "https://arxiv.org/abs/2606.19531v1", "title": "ImageWAM: Do World Action Models Really Need Video Generation, or Just Image Editing?", "topic_id": "t01_embodied_world_models", "updated": "2026-06-17T19:25:28Z"}
+{"arxiv_id": "2606.19746", "arxiv_versioned_id": "2606.19746v1", "authors": ["Ruiyang Ma", "Teng Ma", "Junru Li", "Hantian Zha", "Xuchun Shang", "Qingda Hu", "Zheng Liu", "Xinjun Yang", "Tao Ma", "Guojie Luo"], "categories": ["cs.DC"], "document_id": "arxiv_2606_19746", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 22, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.19746v1", "primary_category": "cs.DC", "published": "2026-06-18T03:21:32Z", "source_url": "https://arxiv.org/abs/2606.19746v1", "title": "SAC: Disaggregated KV Cache System for Sparse Attention LLMs with CXL", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-18T03:21:32Z"}
+{"arxiv_id": "2606.20512", "arxiv_versioned_id": "2606.20512v2", "authors": ["Asa Shepard", "Jeannie Albrecht"], "categories": ["cs.SE", "cs.LG"], "document_id": "arxiv_2606_20512", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 20, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.20512v2", "primary_category": "cs.SE", "published": "2026-06-18T17:30:15Z", "source_url": "https://arxiv.org/abs/2606.20512v2", "title": "Probe-and-Refine Tuning of Repository Guidance for Coding Agents", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-06-19T10:59:07Z"}
+{"arxiv_id": "2606.20729", "arxiv_versioned_id": "2606.20729v1", "authors": ["Masaya Hagai", "Yuta Suzuki", "Tomoya Murata", "Shuhei Kurita", "Masaki Adachi"], "categories": ["physics.chem-ph", "cond-mat.mtrl-sci", "cs.AI", "cs.LG"], "document_id": "arxiv_2606_20729", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 17, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.20729v1", "primary_category": "physics.chem-ph", "published": "2026-06-17T05:58:58Z", "source_url": "https://arxiv.org/abs/2606.20729v1", "title": "LLM-Guided Test-Time Discovery of Quantum-Chemical Approximation Algorithms", "topic_id": "t10_scientific_foundation_models", "updated": "2026-06-17T05:58:58Z"}
+{"arxiv_id": "2606.20891", "arxiv_versioned_id": "2606.20891v1", "authors": ["Koichi Namekata", "Yash Kant", "Zhizheng Liu", "Ryan D Burgert", "Yuancheng Xu", "Kuan Heng Lin", "Emmett Steven", "Julien Philip", "Li Ma", "Andrea Vedaldi", "Paul Debevec", "Ning Yu"], "categories": ["cs.CV", "cs.LG"], "document_id": "arxiv_2606_20891", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 29, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.20891v1", "primary_category": "cs.CV", "published": "2026-06-18T19:40:03Z", "source_url": "https://arxiv.org/abs/2606.20891v1", "title": "Go-with-the-Track: Video Compositing and Motion Control with Point Tracking", "topic_id": "t07_video_generation_control", "updated": "2026-06-18T19:40:03Z"}
+{"arxiv_id": "2606.21023", "arxiv_versioned_id": "2606.21023v1", "authors": ["Zhenting Zhu", "Lucas Thai", "Shan Yu", "Yicheng Liu", "Yifan Qiao", "Chenxi Wang", "Harry Xu", "Junyi Shu"], "categories": ["cs.LG"], "document_id": "arxiv_2606_21023", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 25, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.21023v1", "primary_category": "cs.LG", "published": "2026-06-19T01:21:10Z", "source_url": "https://arxiv.org/abs/2606.21023v1", "title": "Demystifying Numerical Instability in LLM Inference: Achieving Reproducible Inference for Mission-Critical Tasks with HEAL", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-19T01:21:10Z"}
+{"arxiv_id": "2606.21633", "arxiv_versioned_id": "2606.21633v1", "authors": ["Omin Kwon", "Doyeon Kim", "Jongseok Park", "Seung Yul Lee", "Ion Stoica", "Jae W. Lee"], "categories": ["cs.LG"], "document_id": "arxiv_2606_21633", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 14, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.21633v1", "primary_category": "cs.LG", "published": "2026-06-19T17:36:51Z", "source_url": "https://arxiv.org/abs/2606.21633v1", "title": "HERALD: High-Throughput Block Diffusion LLM Serving via CPU-GPU Cooperative KV Cache Retrieval", "topic_id": "t09_efficient_inference_kv_cache", "updated": "2026-06-19T17:36:51Z"}
+{"arxiv_id": "2606.22082", "arxiv_versioned_id": "2606.22082v1", "authors": ["Yifei Wang", "Ruiyin Li", "Peng Liang", "Qiong Feng", "Zengyang Li", "Mojtaba Shahin", "Arif Ali Khan"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2606_22082", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 36, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22082v1", "primary_category": "cs.SE", "published": "2026-06-20T14:57:33Z", "source_url": "https://arxiv.org/abs/2606.22082v1", "title": "CodeTeam: An LLM-Powered Multi-Agent Framework for Repository-Level Code Generation", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-06-20T14:57:33Z"}
+{"arxiv_id": "2606.22617", "arxiv_versioned_id": "2606.22617v1", "authors": ["Hao Vo", "Phu Loc Nguyen", "Khoa Vo", "Sieu Tran", "Duc Minh Nguyen", "Ngo Xuan Cuong", "Nghi D. Q. Bui", "Anh Nguyen", "Duy Minh Ho Nguyen", "Ngan Le"], "categories": ["cs.CV"], "document_id": "arxiv_2606_22617", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 14, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22617v1", "primary_category": "cs.CV", "published": "2026-06-21T17:47:31Z", "source_url": "https://arxiv.org/abs/2606.22617v1", "title": "OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-21T17:47:31Z"}
+{"arxiv_id": "2606.22631", "arxiv_versioned_id": "2606.22631v1", "authors": ["Chaoyue Li", "Boxue Yang", "Shengyao Zhou", "Haoyang Wu", "Rui Qian", "Linfeng Zhang"], "categories": ["cs.CV"], "document_id": "arxiv_2606_22631", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 28, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22631v1", "primary_category": "cs.CV", "published": "2026-06-21T18:33:15Z", "source_url": "https://arxiv.org/abs/2606.22631v1", "title": "4DVLT: Dynamic Scene Understanding with Worldline-Centered Vision-Language Tracking", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-21T18:33:15Z"}
+{"arxiv_id": "2606.22694", "arxiv_versioned_id": "2606.22694v1", "authors": ["Danial Kamali", "Tanawan Premsri", "Shreya Rajpal", "Amir Zadeh", "Chuan Li", "Parisa Kordjamshidi"], "categories": ["cs.CV", "cs.SC"], "document_id": "arxiv_2606_22694", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 28, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22694v1", "primary_category": "cs.CV", "published": "2026-06-21T22:15:48Z", "source_url": "https://arxiv.org/abs/2606.22694v1", "title": "SATURN: Symbolic Spatial Reasoning for Multi-Perspective Grounding", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-21T22:15:48Z"}
+{"arxiv_id": "2606.22813", "arxiv_versioned_id": "2606.22813v1", "authors": ["Omar Hashash", "Christo Kurisummoottil Thomas", "Walid Saad", "Merouane Debbah", "Karl Friston", "Adeel Razi"], "categories": ["cs.AI"], "document_id": "arxiv_2606_22813", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 53, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.22813v1", "primary_category": "cs.AI", "published": "2026-06-22T03:41:31Z", "source_url": "https://arxiv.org/abs/2606.22813v1", "title": "Active Inference as the Test-Time Scaling Law for Physical AI Agents", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-22T03:41:31Z"}
+{"arxiv_id": "2606.22844", "arxiv_versioned_id": "2606.22844v1", "authors": ["Wei Yang", "Bryce Kan", "Shixuan Li", "Li Li", "Yuehan Qin", "Jiate Li", "Paul Bogdan", "Jesse Thomason"], "categories": ["cs.AI", "cs.MA"], "document_id": "arxiv_2606_22844", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 36, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22844v1", "primary_category": "cs.AI", "published": "2026-06-22T04:41:14Z", "source_url": "https://arxiv.org/abs/2606.22844v1", "title": "RaMem: Contextual Reinstatement for Long-term Agentic Memory", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-22T04:41:14Z"}
+{"arxiv_id": "2606.22906", "arxiv_versioned_id": "2606.22906v1", "authors": ["Jiawei He", "Weisong Sun", "Mengyu Shi", "Jie Jia", "Tong Bian", "Xikai Yang", "Dong Sun"], "categories": ["cs.SE", "cs.AI"], "document_id": "arxiv_2606_22906", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 12, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22906v1", "primary_category": "cs.SE", "published": "2026-06-22T06:44:51Z", "source_url": "https://arxiv.org/abs/2606.22906v1", "title": "From Fragments to Paths: Task-Level Context Recovery for Large Industrial Codebases", "topic_id": "t03_software_engineering_code_agents", "updated": "2026-06-22T06:44:51Z"}
+{"arxiv_id": "2606.22938", "arxiv_versioned_id": "2606.22938v1", "authors": ["Stanley Wei", "Juno Kim"], "categories": ["cs.LG", "cs.AI"], "document_id": "arxiv_2606_22938", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 38, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.22938v1", "primary_category": "cs.LG", "published": "2026-06-22T07:16:08Z", "source_url": "https://arxiv.org/abs/2606.22938v1", "title": "Provable Benefits of RLVR over SFT for Reasoning Models: Learning to Backtrack Efficiently", "topic_id": "t06_test_time_scaling_reasoning", "updated": "2026-06-22T07:16:08Z"}
+{"arxiv_id": "2606.23296", "arxiv_versioned_id": "2606.23296v1", "authors": ["Chengyu Bai", "Peidong Jia", "Tiecheng Guo", "Yukai Wang", "Rui Ma", "Fangyuan Zhao", "Chunkai Fan", "Xiaobao Wei", "Jintao Chen", "Hao Wang", "Ying Li", "Xiaozhu Ju", "Jian Tang", "Shanghang Zhang"], "categories": ["cs.RO"], "document_id": "arxiv_2606_23296", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 16, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.23296v1", "primary_category": "cs.RO", "published": "2026-06-22T13:09:34Z", "source_url": "https://arxiv.org/abs/2606.23296v1", "title": "IOI: Decoupling Kinematics and Physics for Interactive World Models", "topic_id": "t01_embodied_world_models", "updated": "2026-06-22T13:09:34Z"}
+{"arxiv_id": "2606.23557", "arxiv_versioned_id": "2606.23557v1", "authors": ["Jiho Choi", "Seonho Lee", "Seojeong Park", "Hyunjung Shim"], "categories": ["cs.CV"], "document_id": "arxiv_2606_23557", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 41, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.23557v1", "primary_category": "cs.CV", "published": "2026-06-22T16:28:11Z", "source_url": "https://arxiv.org/abs/2606.23557v1", "title": "Dense Reward for Multi-View 3D Reasoning with Global Maps and Local Views", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-22T16:28:11Z"}
+{"arxiv_id": "2606.23610", "arxiv_versioned_id": "2606.23610v1", "authors": ["Hongkai Zheng", "Ta-Ying Cheng", "Benjamin Klein", "Yisong Yue", "Zhuoning Yuan"], "categories": ["cs.CV"], "document_id": "arxiv_2606_23610", "license_url": "https://creativecommons.org/licenses/by-nc-sa/4.0/", "page_count": 24, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.23610v1", "primary_category": "cs.CV", "published": "2026-06-22T17:11:11Z", "source_url": "https://arxiv.org/abs/2606.23610v1", "title": "Vera: A Layered Diffusion Model for Content-Preserving Video Editing", "topic_id": "t07_video_generation_control", "updated": "2026-06-22T17:11:11Z"}
+{"arxiv_id": "2606.24480", "arxiv_versioned_id": "2606.24480v1", "authors": ["Eric Jianfeng Cheng", "Min Hong", "Zhiquan Zeng", "Chuanyu Liu", "Qian Wang", "Melissa Jane Meadowcroft", "Vlad Badilita", "Carlos Miguel Costa", "Senentxu Lanceros-Méndez", "Ying Li", "Chenyao Ma", "Weibo Gong", "Surendra Martha", "Aloysius Soon", "Piao Ma", "Di Zhang", "Teguh Ariyanto", "Sudaryanto", "Hiroshi Kakinuma", "Jie Zhao", "Jiayu Peng", "Pengfei Ou", "Jun Lu", "Shin-ichi Orimo", "Hao Li"], "categories": ["cond-mat.mtrl-sci"], "document_id": "arxiv_2606_24480", "license_url": "https://creativecommons.org/licenses/by/4.0/", "page_count": 54, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.24480v1", "primary_category": "cond-mat.mtrl-sci", "published": "2026-06-23T12:16:41Z", "source_url": "https://arxiv.org/abs/2606.24480v1", "title": "Breaking Bottlenecks in Solid Electrolyte Discovery with Large Artificial Intelligence Models", "topic_id": "t10_scientific_foundation_models", "updated": "2026-06-23T12:16:41Z"}
+{"arxiv_id": "2606.24626", "arxiv_versioned_id": "2606.24626v1", "authors": ["Chenyang Zhu", "Jiayu Yao", "Kushal Chawla", "Youbing Yin", "Nathan Wolfe", "Pengshan Cai", "Jingyu Wu", "Spencer Hong", "Sangwoo Cho", "Shi-Xiong Zhang", "Daben Liu", "Sambit Sahu", "Erin Babinsky"], "categories": ["cs.AI"], "document_id": "arxiv_2606_24626", "license_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/", "page_count": 16, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.24626v1", "primary_category": "cs.AI", "published": "2026-06-23T14:23:40Z", "source_url": "https://arxiv.org/abs/2606.24626v1", "title": "SAFARI: Scaling Long Horizon Agentic Fault Attribution via Active Investigation", "topic_id": "t04_agent_memory_long_horizon", "updated": "2026-06-23T14:23:40Z"}
+{"arxiv_id": "2606.26800", "arxiv_versioned_id": "2606.26800v1", "authors": ["Kaijun Wang", "Zikai Ouyang", "Xuping Wu", "Jinyi Hong", "Wei Pan", "Haibo Lu", "Jia Pan", "Wei Zhang", "Linfang Zheng"], "categories": ["cs.RO"], "document_id": "arxiv_2606_26800", "license_url": "https://creativecommons.org/licenses/by-nc-sa/4.0/", "page_count": 8, "page_image_access": "included", "page_images_included": true, "pdf_url": "https://arxiv.org/pdf/2606.26800v1", "primary_category": "cs.RO", "published": "2026-06-25T09:38:05Z", "source_url": "https://arxiv.org/abs/2606.26800v1", "title": "SSI-Policy: Learning Structured Scene Interfaces for Vision-Language Robotic Manipulation", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-25T09:38:05Z"}
+{"arxiv_id": "2606.27876", "arxiv_versioned_id": "2606.27876v1", "authors": ["Haoyu Zhang", "Meng Liu", "Qianlong Xiang", "Kun Wang", "Yaowei Wang", "Liqiang Nie"], "categories": ["cs.CV", "cs.AI"], "document_id": "arxiv_2606_27876", "license_url": "https://arxiv.org/licenses/nonexclusive-distrib/1.0/", "page_count": 10, "page_image_access": "reconstruct_from_source", "page_images_included": false, "pdf_url": "https://arxiv.org/pdf/2606.27876v1", "primary_category": "cs.CV", "published": "2026-06-26T09:16:55Z", "source_url": "https://arxiv.org/abs/2606.27876v1", "title": "SpatialUAV: Benchmarking Spatial Intelligence for Low-Altitude UAV Perception, Collaboration, and Motion", "topic_id": "t08_3d_spatial_intelligence", "updated": "2026-06-26T09:16:55Z"}

corpus/pages.jsonl ADDED Viewed

The diff for this file is too large to render. See raw diff

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p001.png ADDED Viewed

Git LFS Details

SHA256: 0f2e3362265132a5a40f589be2a233ef07e0467a2543bab5a37e6298d08634c7
Pointer size: 131 Bytes
Size of remote file: 404 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p002.png ADDED Viewed

Git LFS Details

SHA256: e980a7d56f9cf781d55d5caf3c93b65b4ac7f5118d1d97c3999b08ef8e2e7a2c
Pointer size: 131 Bytes
Size of remote file: 528 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p003.png ADDED Viewed

Git LFS Details

SHA256: 29cef6fcb5c6c3a759bb3c2b775910f7b176360a8365976c570b0980569d6d80
Pointer size: 131 Bytes
Size of remote file: 463 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p004.png ADDED Viewed

Git LFS Details

SHA256: e334fe0d4233dee18793ac595bd30349eb8dc30e7fee87d717974935ba1af5fe
Pointer size: 131 Bytes
Size of remote file: 420 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p005.png ADDED Viewed

Git LFS Details

SHA256: ac677b5253ab85f62a69e81765ef2d97c09ea46081b57c5bd380dd97fe947c31
Pointer size: 131 Bytes
Size of remote file: 478 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p006.png ADDED Viewed

Git LFS Details

SHA256: bc619eb14219081117a65ca834b74b15e0ae04f94331f7aed2dce24550c3162f
Pointer size: 131 Bytes
Size of remote file: 437 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p007.png ADDED Viewed

Git LFS Details

SHA256: 1852a308e981f206ed3a95b88a275cc66d740a9b195066b8e47aa336ad934ee2
Pointer size: 131 Bytes
Size of remote file: 478 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p008.png ADDED Viewed

Git LFS Details

SHA256: 1dcac16c6b23450abaec127a28e28fbdfd358de2cc98593611e3d2312c616670
Pointer size: 131 Bytes
Size of remote file: 532 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p009.png ADDED Viewed

Git LFS Details

SHA256: a565125fdf0c8f2be419197bccafee3f62aac9bc9bb90fa064ab363979515ab5
Pointer size: 131 Bytes
Size of remote file: 651 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p010.png ADDED Viewed

Git LFS Details

SHA256: b8eface601969b4396d7a01ddfe6ab96ac1e9ffa8ff919194f095c9449ad2f6a
Pointer size: 131 Bytes
Size of remote file: 606 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p011.png ADDED Viewed

Git LFS Details

SHA256: 902517fa3482f9030d668a336b321e7f68c058b3e00740b15a5249669dc00abd
Pointer size: 131 Bytes
Size of remote file: 357 kB

corpus/pages/arxiv_2601_18203/arxiv_2601_18203_p012.png ADDED Viewed

Git LFS Details

SHA256: cac5a2f8d95eef3645efe9321bc1e11125fafa0e45b32b0263778dcf42e08202
Pointer size: 131 Bytes
Size of remote file: 342 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p001.png ADDED Viewed

Git LFS Details

SHA256: d9d329c18069d75d53502de9cf342018ab54ebeeb3c425efea8bfb53d7564727
Pointer size: 131 Bytes
Size of remote file: 310 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p002.png ADDED Viewed

Git LFS Details

SHA256: 756b0f86dad67b1bf08edbb1b8f9b1fb8da8bc8fc0336a8676bc43c4493b7317
Pointer size: 131 Bytes
Size of remote file: 347 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p003.png ADDED Viewed

Git LFS Details

SHA256: 17fbe64b7c02cb7b7c3516dd023293f675b4155a711185bf21efec25158d3d8b
Pointer size: 131 Bytes
Size of remote file: 399 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p004.png ADDED Viewed

Git LFS Details

SHA256: e1a98f01af2ffbc70e7ec534245c39a361c9f216136f29f99d662978c38b4d8d
Pointer size: 131 Bytes
Size of remote file: 344 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p005.png ADDED Viewed

Git LFS Details

SHA256: 6084e17a503268733244ef38b1ead8eabc7f26df74216d53fb614380e6cd886f
Pointer size: 131 Bytes
Size of remote file: 350 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p006.png ADDED Viewed

Git LFS Details

SHA256: 9b997e2bd50e768f512433c37880dc49d3739e33a8f7a575fe15aa7d4f76734d
Pointer size: 131 Bytes
Size of remote file: 387 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p007.png ADDED Viewed

Git LFS Details

SHA256: 5daedce5287abd09595eee09b278d8fcd0490943a851c8224ae8c2accd290477
Pointer size: 131 Bytes
Size of remote file: 360 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p008.png ADDED Viewed

Git LFS Details

SHA256: fe20f1a8390e623a8a9e6622df718e0401ed15d141478dd6f4153621e18fdc55
Pointer size: 131 Bytes
Size of remote file: 323 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p009.png ADDED Viewed

Git LFS Details

SHA256: 7bf8e53ed8c9002e7d8de10c409f31ff404a40f0e545483846e1dc386cf6d3b4
Pointer size: 131 Bytes
Size of remote file: 331 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p010.png ADDED Viewed

Git LFS Details

SHA256: 58e3d038ad9eafcfabf7e61fcba666934b550d01f8a742ac28a3a8697ae90123
Pointer size: 131 Bytes
Size of remote file: 301 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p011.png ADDED Viewed

Git LFS Details

SHA256: 195f24d70bebf39d7222b88a144fb8d95941cc0eb477aa37736fd4fda8e337ed
Pointer size: 131 Bytes
Size of remote file: 350 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p012.png ADDED Viewed

Git LFS Details

SHA256: 2407ae840f7a7d48551dba9ca20a7dc68072d62baba65216167808347b19ced8
Pointer size: 131 Bytes
Size of remote file: 390 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p013.png ADDED Viewed

Git LFS Details

SHA256: 3fb0757f3eabb77ef181d4dc14edf6a462dd7f72c51ed792370efbc39989f8ff
Pointer size: 131 Bytes
Size of remote file: 370 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p014.png ADDED Viewed

Git LFS Details

SHA256: f1ffaea0abe1ed320a2c9509aadd65cc6658f79bb2981258027eaf263aa3accf
Pointer size: 131 Bytes
Size of remote file: 217 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p015.png ADDED Viewed

Git LFS Details

SHA256: 9977be8adf3787b6499bf7c507d7c775ee1eac50f3c1599cfc689403a0905646
Pointer size: 131 Bytes
Size of remote file: 134 kB

corpus/pages/arxiv_2602_06090/arxiv_2602_06090_p016.png ADDED Viewed

Git LFS Details

SHA256: 7dc5704dd26e7de5cbdc8cb42deede4640c21b45d5da47ee50bd9cf53e9c3986
Pointer size: 130 Bytes
Size of remote file: 52.6 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p001.png ADDED Viewed

Git LFS Details

SHA256: 442a7a3b450d77e527f2c2a9f2a297493cbe81b375b4f1d620ab307da7cd422e
Pointer size: 131 Bytes
Size of remote file: 403 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p002.png ADDED Viewed

Git LFS Details

SHA256: 57961cc1262547c79d94c26a294f8541237d9efcc533b619e220d917e5c7e6f0
Pointer size: 131 Bytes
Size of remote file: 458 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p003.png ADDED Viewed

Git LFS Details

SHA256: 111f41cfc7f80e5c1ea89afddce20448f9c41638a51b25457e548e33c7db7435
Pointer size: 131 Bytes
Size of remote file: 473 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p004.png ADDED Viewed

Git LFS Details

SHA256: 78d82986a0710458cd363cf714038056623f7235c059e1fc5a34b17c5e2e77c5
Pointer size: 131 Bytes
Size of remote file: 455 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p005.png ADDED Viewed

Git LFS Details

SHA256: 20240f3f0cf19726eb9ca0f1e7d69dd12a889b37028f758c72ef691a9cae2e0a
Pointer size: 131 Bytes
Size of remote file: 412 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p006.png ADDED Viewed

Git LFS Details

SHA256: 083f0590f05ce2caa0ecb573c96565dcbf4d6a6042c012fbe383b8b607f2fc8d
Pointer size: 131 Bytes
Size of remote file: 454 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p007.png ADDED Viewed

Git LFS Details

SHA256: 5b7ea0bdcdc4a88bdc8fc343c390990a889d3850f3132e970a83f41c4c1a6c90
Pointer size: 131 Bytes
Size of remote file: 424 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p008.png ADDED Viewed

Git LFS Details

SHA256: 20041cfa104f410649c0d6979c8dc4f7c0d2e218a20a0673f75423e3abd2e345
Pointer size: 131 Bytes
Size of remote file: 568 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p009.png ADDED Viewed

Git LFS Details

SHA256: 17bb13d00d1e2b8aee5b66f9916d90912c9eb51faa03d1a72e7e32f7ac9fde68
Pointer size: 131 Bytes
Size of remote file: 455 kB

corpus/pages/arxiv_2602_22013/arxiv_2602_22013_p010.png ADDED Viewed

Git LFS Details

SHA256: 6be4bb7b9cf10339ab98ffdf74cc98b0443010fd78051bd68aae1d378539221d
Pointer size: 131 Bytes
Size of remote file: 454 kB

corpus/pages/arxiv_2603_08013/arxiv_2603_08013_p001.png ADDED Viewed

Git LFS Details

SHA256: bb42217ab157ee951ae80bc2b539bb6001bfd890de0572b56af27aaf619c2733
Pointer size: 131 Bytes
Size of remote file: 252 kB

corpus/pages/arxiv_2603_08013/arxiv_2603_08013_p002.png ADDED Viewed

Git LFS Details

SHA256: 4fc5954e6a058f2fdf5ce33b374266c8a29a6f4f6596160e8f27a2a071a145fe
Pointer size: 131 Bytes
Size of remote file: 348 kB

corpus/pages/arxiv_2603_08013/arxiv_2603_08013_p003.png ADDED Viewed

Git LFS Details

SHA256: a2143acedba6005690e4168d2b62ea56105c34735390a53c1deeca05b2843012
Pointer size: 131 Bytes
Size of remote file: 296 kB