RKB109/contextual-bandit-simulator-20260725-model
Reinforcement Learning • Updated
Datasets:
Tasks:
Reinforcement Learning
Modalities:
Text
Formats:
json
Languages:
English
Size:
< 1K
Tags:
Synthetic
reinforcement-learning
evaluation
text-classification
feature-extraction
sentence-similarity
License:
id stringlengths 21 21 | input stringlengths 40 76 | label stringclasses 3
values | context stringclasses 6
values | source stringclasses 6
values | variant stringclasses 3
values | synthetic bool 1
class |
|---|---|---|---|---|---|---|
contextual-bandit-1-1 | experienced user asks for API parameter details | recommend-docs | Documentation is the highest-value action. | bandit-01 | direct | true |
contextual-bandit-1-2 | In an operations review, experienced user asks for API parameter details | recommend-docs | Documentation is the highest-value action. | bandit-01 | operations | true |
contextual-bandit-1-3 | For an evaluation case, experienced user asks for API parameter details | recommend-docs | Documentation is the highest-value action. | bandit-01 | evaluation | true |
contextual-bandit-2-2 | In an operations review, new developer asks how to build a first integration | recommend-tutorial | A guided tutorial is the highest-value action. | bandit-02 | operations | true |
contextual-bandit-2-3 | For an evaluation case, new developer asks how to build a first integration | recommend-tutorial | A guided tutorial is the highest-value action. | bandit-02 | evaluation | true |
contextual-bandit-3-1 | user reports a possible security compromise | request-human-help | Sensitive security issues require human support. | bandit-03 | direct | true |
contextual-bandit-3-3 | For an evaluation case, user reports a possible security compromise | request-human-help | Sensitive security issues require human support. | bandit-03 | evaluation | true |
contextual-bandit-4-1 | developer needs exact error code reference | recommend-docs | Reference material is appropriate for precise lookup. | bandit-04 | direct | true |
contextual-bandit-4-2 | In an operations review, developer needs exact error code reference | recommend-docs | Reference material is appropriate for precise lookup. | bandit-04 | operations | true |
contextual-bandit-5-1 | beginner requests a complete walkthrough | recommend-tutorial | Structured onboarding is appropriate. | bandit-05 | direct | true |
contextual-bandit-5-2 | In an operations review, beginner requests a complete walkthrough | recommend-tutorial | Structured onboarding is appropriate. | bandit-05 | operations | true |
contextual-bandit-5-3 | For an evaluation case, beginner requests a complete walkthrough | recommend-tutorial | Structured onboarding is appropriate. | bandit-05 | evaluation | true |
contextual-bandit-6-2 | In an operations review, customer indicates potential data loss | request-human-help | High-impact cases must be escalated. | bandit-06 | operations | true |
contextual-bandit-6-3 | For an evaluation case, customer indicates potential data loss | request-human-help | High-impact cases must be escalated. | bandit-06 | evaluation | true |
Contextual Bandit Decision Simulator Synthetic Dataset
Summary
This dataset contains 14 training examples and 4 held-out examples for Teams need to validate decision policies offline before exposing users or systems to online reinforcement learning.
Every record is synthetic and includes:
input: query, event, or feature descriptionlabel: expected class, route, relation, or evidence categorycontext: synthetic supporting contextsource: fictional source identifiervariant: generation patternsynthetic: alwaystrue
Uses
- Reproducible unit and integration tests
- Baseline model training
- Evaluation harness development
- Schema and architecture demonstrations
Limitations
Offline simulated rewards cannot prove online safety or business impact. Real experiments require review and guardrails.
This dataset does not represent real users, patients, customers, production traffic, or licensed media. It must not be presented as real-world evidence.
Related Model
- Downloads last month
- 43
Models trained or fine-tuned on RKB109/contextual-bandit-simulator-20260725-dataset