datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
laya-task-routing-synthetic-v2
Laya Task Routing Synthetic v2
Synthetic English task data for experiments that classify a task into an area (home, work, or personal) and then apply area-specific multi-label tags.
This dataset was generated by a self-hosted Qwen 3.8 27B model and subsequently reviewed by the same model as part of the synthetic-data validation pipeline.
Configurations
Config
Records
Decisions
Purpose
area
1,800
1,800
Area classification only
tags
1,800
18,000… See the full description on the dataset page: https://huggingface.co/datasets/chand1012/laya-task-routing-synthetic-v2.tt639f3-task-routing-final-repair-v1
TT639F3 Task Routing Final Repair v1
Surgical repair of TT639F2's 6 exact misses:
return user.name -> code
schedule changed again -> prose
positive sentiment for enjoy / strong / helped
Blocking gates:
seen_pass
tt639f2_failure_regression_pass
heldout_v3_pass
anti_collision_v3_pass
tt639f-task-routing-ladder-v1
TT639F Task Routing Ladder v1
Purpose:
Prove task routing separately after TT639E2 context-copy repair.
Task families:
classify_code / classify_prose
sentiment_positive / sentiment_negative
rewrite_short_late / rewrite_professional_attend
idk_favorite_movie / idk_food_today
Blocking gates:
seen_pass
tt639d_task_regression_pass
heldout_pass
anti_collision_pass
Do not run combined TT639G until TT639F passes.
tt639f2-task-routing-repair-v1
TT639F2 Task Routing Repair v1
Surgical repair of TT639F heldout failures. Blocking gates: seen, TT639F failure regression, heldout v2, anti-collision v2.