Nov 02 – Nov 08
Real-world agent evaluation
WeeklyIssue #12
Papers.Data.Code
The Weekly ML Signal
Selected, not collected.
Papers, repos, datasets, and trends worth your attention.
Latest issues
All issues →
Oct 26 – Nov 01
From worlds to grounding
WeeklyIssue #11
October 2026
Memory beyond retrieval
MonthlyIssue #10
Oct 12 – Oct 18
Verifiable worlds and control
WeeklyIssue #9
Oct 05 – Oct 11
Long memory, adaptive compute
WeeklyIssue #8
Sep 28 – Oct 04
Planning, routing, and robust eval
WeeklyIssue #7
Worth your time, not all of your time.
We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.
Latest signals
Browse the archive →Picked Nov 6
- PaperToolSpace: Scaling Tool-Centric Environment Synthesis for Long-Horizon Agents→
Raises Qwen3.6-35B-A3B from 23.0 to 41.0 avg (43.2 with RL); frontier models drop up to 38.2 pts on Pro benchmarks.
- DatasetAxion1-0/axion-training-data→
Unlike mostly evaluation benchmarks, it releases 18,363 full agent-training trajectories plus 484 RL tasks with verified tool-use.
- PaperRoboColosseum: Evaluating Physical Intelligence at Scale In-The-Wild→
Built from 5,699 comparisons at 124 sites; active matchmaking hits ~90% ranking agreement in 400-500 trials vs ~1,000 uniform.
- RepoGenLang-V/GenDB-V→
Combines formal end-to-end correctness with large reported gains: up to 61.3× vs DuckDB and 28.3× vs Umbra.
Picked Nov 5
- PaperRoute Anything: A Study of Model and Policy Routing for General-Purpose AI Systems→
Oracle routing beats the best fixed candidate by 5.4 points; learned routers gain 0.3 overall, 2.1 on video, 1.1 on robots with 24% less policy time, and 8.82x efficiency.
- Datasetdeliverbench/DeliverBench→
Makes continuous project delivery measurable beyond single tasks by scoring checkpoint outputs and final packages separately in one workspace.