# arXiv Scout Coverage — 2026-10-15_11-56

- Queries run: 14
- Total pages fetched: 34
- Total entries seen: 2400
- Total unique papers kept before cross-query dedupe: 2400
- Total duplicates skipped within queries: 0
- Queries with incomplete coverage: 0

## Per-query coverage

### primary-cs.AI
- Query: `cat:cs.AI`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 5
- Entries seen: 500
- Unique papers kept: 500
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:52+00:00
- Oldest seen: 2026-10-13T04:57:25+00:00
- Stop reason: max_pages_reached
- Recovery attempts:
  - attempt=1 pages=5 entries=500 stop=max_pages_reached
- Pages:
  - page 1 start=0 entries=100 newest=2026-10-14T17:59:52+00:00 oldest=2026-10-14T09:32:34+00:00 rate_limited=False status=200
  - page 2 start=100 entries=100 newest=2026-10-14T09:30:04+00:00 oldest=2026-10-14T01:16:15+00:00 rate_limited=False status=200
  - page 3 start=200 entries=100 newest=2026-10-14T01:12:03+00:00 oldest=2026-10-13T16:42:32+00:00 rate_limited=False status=200
  - page 4 start=300 entries=100 newest=2026-10-13T16:40:59+00:00 oldest=2026-10-13T11:10:44+00:00 rate_limited=False status=200
  - page 5 start=400 entries=100 newest=2026-10-13T11:07:04+00:00 oldest=2026-10-13T04:57:25+00:00 rate_limited=False status=200

### primary-cs.LG
- Query: `cat:cs.LG`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 5
- Entries seen: 500
- Unique papers kept: 500
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:52+00:00
- Oldest seen: 2026-10-13T04:02:46+00:00
- Stop reason: max_pages_reached
- Recovery attempts:
  - attempt=1 pages=5 entries=500 stop=max_pages_reached
- Pages:
  - page 1 start=0 entries=100 newest=2026-10-14T17:59:52+00:00 oldest=2026-10-14T11:21:30+00:00 rate_limited=False status=200
  - page 2 start=100 entries=100 newest=2026-10-14T11:12:56+00:00 oldest=2026-10-14T01:58:40+00:00 rate_limited=False status=200
  - page 3 start=200 entries=100 newest=2026-10-14T01:53:24+00:00 oldest=2026-10-13T17:46:25+00:00 rate_limited=False status=200
  - page 4 start=300 entries=100 newest=2026-10-13T17:45:34+00:00 oldest=2026-10-13T11:20:51+00:00 rate_limited=False status=200
  - page 5 start=400 entries=100 newest=2026-10-13T11:20:24+00:00 oldest=2026-10-13T04:02:46+00:00 rate_limited=False status=200

### primary-cs.CL
- Query: `cat:cs.CL`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 4
- Entries seen: 400
- Unique papers kept: 400
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:57:01+00:00
- Oldest seen: 2026-10-12T07:32:11+00:00
- Stop reason: cutoff_reached
- Recovery attempts:
  - attempt=1 pages=4 entries=400 stop=cutoff_reached
- Pages:
  - page 1 start=0 entries=100 newest=2026-10-14T17:57:01+00:00 oldest=2026-10-14T05:36:03+00:00 rate_limited=False status=200
  - page 2 start=100 entries=100 newest=2026-10-14T05:30:43+00:00 oldest=2026-10-13T16:27:44+00:00 rate_limited=False status=200
  - page 3 start=200 entries=100 newest=2026-10-13T16:19:49+00:00 oldest=2026-10-13T03:59:13+00:00 rate_limited=False status=200
  - page 4 start=300 entries=100 newest=2026-10-13T03:49:32+00:00 oldest=2026-10-12T07:32:11+00:00 rate_limited=False status=200

### topic-world-model
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("world model" OR world-model)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:52+00:00
- Oldest seen: 2026-10-12T16:26:46+00:00
- Stop reason: max_pages_reached
- Recovery attempts:
  - attempt=1 pages=2 entries=100 stop=max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:59:52+00:00 oldest=2026-10-13T15:59:31+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-13T15:53:26+00:00 oldest=2026-10-12T16:26:46+00:00 rate_limited=False status=200

### topic-agent
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND agent`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:28+00:00
- Oldest seen: 2026-10-14T02:53:17+00:00
- Stop reason: max_pages_reached
- Recovery attempts:
  - attempt=1 pages=2 entries=100 stop=max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:59:28+00:00 oldest=2026-10-14T10:38:30+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-14T10:33:30+00:00 oldest=2026-10-14T02:53:17+00:00 rate_limited=False status=200

### topic-tool-use
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("tool use" OR tool-use)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T16:42:05+00:00
- Oldest seen: 2026-10-11T15:37:56+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T16:42:05+00:00 oldest=2026-10-13T11:28:03+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-13T09:52:02+00:00 oldest=2026-10-11T15:37:56+00:00 rate_limited=False status=200

### topic-memory
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND memory`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:57:40+00:00
- Oldest seen: 2026-10-12T21:21:21+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:57:40+00:00 oldest=2026-10-13T18:32:31+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-13T17:59:57+00:00 oldest=2026-10-12T21:21:21+00:00 rate_limited=False status=200

### topic-reasoning
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND reasoning`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:57:12+00:00
- Oldest seen: 2026-10-13T18:26:50+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:57:12+00:00 oldest=2026-10-14T07:32:32+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-14T07:23:08+00:00 oldest=2026-10-13T18:26:50+00:00 rate_limited=False status=200

### topic-llm
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND (LLM OR "large language model" OR "large language models")`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:56:06+00:00
- Oldest seen: 2026-10-14T03:54:19+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:56:06+00:00 oldest=2026-10-14T09:50:59+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-14T09:48:15+00:00 oldest=2026-10-14T03:54:19+00:00 rate_limited=False status=200

### topic-quantization
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND quantization`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-10-14T15:29:40+00:00
- Oldest seen: 2026-10-08T23:38:47+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T15:29:40+00:00 oldest=2026-10-08T23:38:47+00:00 rate_limited=False status=200

### topic-open-source
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("open source" OR open-source)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:53:46+00:00
- Oldest seen: 2026-10-09T01:41:47+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:53:46+00:00 oldest=2026-10-12T17:57:40+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-12T17:55:25+00:00 oldest=2026-10-09T01:41:47+00:00 rate_limited=False status=200

### topic-self-improvement
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("self-improvement" OR "self improvement")`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-10-14T16:51:09+00:00
- Oldest seen: 2026-10-08T06:40:18+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T16:51:09+00:00 oldest=2026-10-08T06:40:18+00:00 rate_limited=False status=200

### topic-fine-tuning
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("fine-tuning" OR finetuning OR finetune)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:52+00:00
- Oldest seen: 2026-10-12T17:55:19+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:59:52+00:00 oldest=2026-10-13T17:40:20+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-13T16:55:32+00:00 oldest=2026-10-12T17:55:19+00:00 rate_limited=False status=200

### topic-benchmark
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND (benchmark OR benchmarks)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-10-14T17:59:28+00:00
- Oldest seen: 2026-10-14T05:09:19+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-10-14T17:59:28+00:00 oldest=2026-10-14T11:09:44+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-10-14T10:57:55+00:00 oldest=2026-10-14T05:09:19+00:00 rate_limited=False status=200

## Priority summary
- High-priority papers: 524
- Priority threshold: 4
- High-similarity papers: 0
- Similarity separation threshold: 0.15
- Topic hits:
  - agent: 100
  - benchmark: 100
  - fine-tuning: 100
  - llm: 100
  - memory: 100
  - open-source: 100
  - reasoning: 100
  - tool-use: 100
  - world-model: 100
  - quantization: 50
  - self-improvement: 50
- Top papers:
  - 21.0 / rank=21.00 :: MemDuet: Learning Dual-Process Memory for Long-Context Reasoning [agent, benchmark, fine-tuning, llm, memory, reasoning] sim+=0.21 sim-=0.06 sep=0.00
  - 19.0 / rank=19.00 :: StateCraft: Augmenting Video World Models with Abstract-State Reasoning [agent, benchmark, memory, reasoning, world-model] sim+=0.31 sim-=0.09 sep=0.00
  - 18.0 / rank=18.00 :: RuleTrek: Benchmarking Agentic Rule Reasoning in Complex Scenarios [agent, benchmark, llm, reasoning, world-model] sim+=0.17 sim-=0.05 sep=0.00
  - 17.0 / rank=17.00 :: Player-1: Learning to Play from Synthetic Experience [agent, fine-tuning, reasoning, world-model] sim+=0.22 sim-=0.07 sep=0.00
  - 17.0 / rank=17.00 :: Same Claim, Different Interpretations: How LLM Agents Resolve Ambiguity in Scientific Claims [agent, benchmark, llm, reasoning, tool-use] sim+=0.20 sim-=0.07 sep=0.00
  - 17.0 / rank=17.00 :: MMLA-RSI: Recursive Self-Improvement through RTT and RTMA [agent, memory, reasoning, self-improvement] sim+=0.23 sim-=0.05 sep=0.00
  - 17.0 / rank=17.00 :: Learning Perception to Action Rules through Reinforcement Learning for Theory-of-Mind Reasoning [agent, benchmark, fine-tuning, llm, reasoning] sim+=0.18 sim-=0.05 sep=0.00
  - 17.0 / rank=17.00 :: L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents [agent, reasoning, self-improvement, tool-use] sim+=0.36 sim-=0.11 sep=0.00
  - 17.0 / rank=17.00 :: VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness [memory, reasoning, self-improvement, tool-use] sim+=0.24 sim-=0.07 sep=0.00
  - 16.0 / rank=16.00 :: PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference [benchmark, llm, memory, quantization, reasoning] sim+=0.13 sim-=0.05 sep=0.00
  - 16.0 / rank=16.00 :: XferAgent: Cross-Agent Experience Transfer for Self-Improving LLM Agents [agent, llm, memory, self-improvement] sim+=0.15 sim-=0.03 sep=0.00
  - 15.0 / rank=15.00 :: P2Patch: Exploit-Guided Vulnerability Repair [agent, llm, reasoning, tool-use] sim+=0.18 sim-=0.06 sep=0.00
  - 15.0 / rank=15.00 :: MCAIR: Monitoring and Correcting Agent Unfaithfulness via Internal Representations [agent, llm, reasoning, tool-use] sim+=0.17 sim-=0.04 sep=0.00
  - 14.0 / rank=14.00 :: PerSo: Investigating Facets of LLMs' Social Intelligence in Real-World Persuasion Scenarios [agent, benchmark, llm, world-model] sim+=0.21 sim-=0.06 sep=0.00
  - 14.0 / rank=14.00 :: TreeFlex: Agents Navigating Their Own Search Trees [agent, benchmark, reasoning, tool-use] sim+=0.28 sim-=0.08 sep=0.00
  - 14.0 / rank=14.00 :: SpectrumQuant: Spectrum Alignment for LLM Quantization [llm, memory, quantization, tool-use] sim+=0.14 sim-=0.03 sep=0.00
  - 14.0 / rank=14.00 :: ActionOPD: Distilling Decisions, Not Tokens [agent, benchmark, reasoning, tool-use] sim+=0.17 sim-=0.05 sep=0.00
  - 14.0 / rank=14.00 :: ReproBAIT: Can coding agents build their own biological AI tools? [agent, llm, open-source, tool-use] sim+=0.21 sim-=0.07 sep=0.00
  - 14.0 / rank=14.00 :: TIMERSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models [agent, self-improvement, world-model] sim+=0.15 sim-=0.05 sep=0.00
  - 14.0 / rank=14.00 :: BANGLAPLAN: A Benchmark for Constraint-Based Planning and Decision-Making in Bangla [benchmark, llm, reasoning, world-model] sim+=0.19 sim-=0.07 sep=0.00