Hugging Face's logo

Datasets:
hummbl-hf
/
base120

Modalities:
Text
Formats:
parquet
Languages:
English
Size:
< 1K
Tags:
reasoning
mental-models
agents
taxonomy
hummbl
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer
Auto-converted to Parquet Duplicate
code
large_stringlengths
2
4
transformation
large_stringclasses
6 values
family_name
large_stringclasses
6 values
index
int32
1
20
name
large_stringlengths
8
44
definition
large_stringlengths
48
97
P1
P
Perspective
1
First Principles Framing
Reduce complex problems to foundational truths that cannot be further simplified
P2
P
Perspective
2
Stakeholder Mapping
Identify all parties with interest, influence, or impact in a system or decision
P3
P
Perspective
3
Identity Stack
Recognize that individuals operate from multiple nested identities simultaneously
P4
P
Perspective
4
Lens Shifting
Deliberately adopt different interpretive frameworks to reveal hidden aspects of a situation
P5
P
Perspective
5
Empathy Mapping
Systematically capture what stakeholders see, think, feel, and do in their context
P6
P
Perspective
6
Point-of-View Anchoring
Establish and maintain a consistent reference frame before analysis begins
P7
P
Perspective
7
Perspective Switching
Rotate through multiple viewpoints to identify invariants and blind spots
P8
P
Perspective
8
Narrative Framing
Structure information as causal stories with conflict, choice, and consequence
P9
P
Perspective
9
Cultural Lens Shifting
Adjust communication and interpretation for different cultural contexts and norms
P10
P
Perspective
10
Context Windowing
Define explicit boundaries in time, space, and scope for analysis or action
P11
P
Perspective
11
Role Perspective-Taking
Temporarily inhabit specific roles to understand constraints and priorities
P12
P
Perspective
12
Temporal Framing
Organize understanding across past causes, present states, and future implications
P13
P
Perspective
13
Spatial Framing
Scale perspective from local details to global patterns and back
P14
P
Perspective
14
Reference Class Framing
Select comparable situations to inform judgment and avoid uniqueness bias
P15
P
Perspective
15
Assumption Surfacing
Explicitly identify and document beliefs underlying plans or models
P16
P
Perspective
16
Identity-Context Reciprocity
Recognize how identities shape interpretations and contexts reinforce identities
P17
P
Perspective
17
Frame Control & Reframing
Consciously select or reshape interpretive frames to enable new solutions
P18
P
Perspective
18
Boundary Object Selection
Choose representations that bridge multiple perspectives while remaining meaningful to each
P19
P
Perspective
19
Sensemaking Canvases
Deploy structured templates to systematically capture and organize observations
P20
P
Perspective
20
Worldview Articulation
Make explicit the fundamental beliefs and values that drive interpretation and action
IN1
IN
Inversion
1
Subtractive Thinking
Improve systems by removing elements rather than adding complexity
IN2
IN
Inversion
2
Premortem Analysis
Assume failure has occurred and work backward to identify causes
IN3
IN
Inversion
3
Problem Reversal
Solve the inverse of the stated problem to reveal insights
IN4
IN
Inversion
4
Contra-Logic
Argue the opposite position to stress-test assumptions and expose weak reasoning
IN5
IN
Inversion
5
Negative Space Framing
Study what is absent rather than what is present
IN6
IN
Inversion
6
Inverse/Proof by Contradiction
Assume a claim is false, derive logical impossibility, thus proving the claim true
IN7
IN
Inversion
7
Boundary Testing
Explore extreme conditions to find system limits and breaking points
IN8
IN
Inversion
8
Contrapositive Reasoning
Use logical equivalence that if A then B equals if not B then not A
IN9
IN
Inversion
9
Backward Induction
Begin with desired end state and work backward to determine necessary steps
IN10
IN
Inversion
10
Red Teaming
Organize adversarial review to find vulnerabilities through simulated attack
IN11
IN
Inversion
11
Devil's Advocate Protocol
Assign explicit role to argue against group consensus or preferred option
IN12
IN
Inversion
12
Failure First Design
Begin planning by identifying all possible failure modes and designing to prevent them
IN13
IN
Inversion
13
Opportunity Cost Focus
Evaluate options by what must be forgone rather than what is gained
IN14
IN
Inversion
14
Second-Order Effects (Inverted)
Trace negative downstream consequences rather than immediate benefits
IN15
IN
Inversion
15
Constraint Reversal
Temporarily remove assumed constraints to explore alternative solution space
IN16
IN
Inversion
16
Inverse Optimization
Maximize worst outcomes to understand system vulnerabilities
IN17
IN
Inversion
17
Counterfactual Negation
Imagine outcomes if key decision had been reversed
IN18
IN
Inversion
18
Kill-Criteria & Stop Rules
Define conditions that trigger project termination before launch
IN19
IN
Inversion
19
Harm Minimization (Via Negativa)
Improve by removing harmful elements rather than adding beneficial ones
IN20
IN
Inversion
20
Antigoals & Anti-Patterns Catalog
Document failure modes to avoid rather than success patterns to emulate
CO1
CO
Composition
1
Synergy Principle
Design combinations where integrated value exceeds sum of parts
CO2
CO
Composition
2
Chunking
Group related elements into meaningful units to reduce cognitive load
CO3
CO
Composition
3
Functional Composition
Chain pure operations where output of one becomes input of next
CO4
CO
Composition
4
Interdisciplinary Synthesis
Merge insights from distinct fields to generate novel solutions
CO5
CO
Composition
5
Emergence
Recognize higher-order behavior arising from component interactions beyond individual properties
CO6
CO
Composition
6
Gestalt Integration
Perceive and leverage whole patterns rather than isolated components
CO7
CO
Composition
7
Network Effects
Exploit increasing value as user base or connections grow
CO8
CO
Composition
8
Layered Abstraction
Separate concerns into hierarchical levels with clear interfaces between them
CO9
CO
Composition
9
Interface Contracts
Define explicit agreements about data structures and behavior between components
CO10
CO
Composition
10
Pipeline Orchestration
Coordinate sequential stages with explicit handoffs and error handling
CO11
CO
Composition
11
Pattern Composition (Tiling)
Combine repeating elements to construct complex structures efficiently
CO12
CO
Composition
12
Modular Interoperability
Ensure independent components work together through standardized connections
CO13
CO
Composition
13
Cross-Domain Analogy
Transfer solution patterns from one domain to solve problems in another
CO14
CO
Composition
14
Platformization
Extract common capabilities into reusable infrastructure serving multiple use cases
CO15
CO
Composition
15
Combinatorial Design
Systematically explore option combinations to find optimal configurations
CO16
CO
Composition
16
System Integration Testing
Verify assembled components work correctly together, not just in isolation
CO17
CO
Composition
17
Orchestration vs Choreography
Choose between centralized coordination or distributed peer-to-peer interaction
CO18
CO
Composition
18
Knowledge Graphing
Represent information as interconnected entities and relationships rather than isolated documents
CO19
CO
Composition
19
Multi-Modal Integration
Synthesize information from different sensory or data modalities
CO20
CO
Composition
20
Holistic Integration
Unify disparate elements into coherent, seamless whole where boundaries dissolve
DE1
DE
Decomposition
1
Root Cause Analysis (5 Whys)
Iteratively ask why problems occur until fundamental cause emerges
DE2
DE
Decomposition
2
Factorization
Separate multiplicative components to understand relative contribution of each factor
DE3
DE
Decomposition
3
Modularization
Partition system into self-contained units with minimal interdependencies
DE4
DE
Decomposition
4
Layered Breakdown
Decompose from system to subsystem to component progressively
DE5
DE
Decomposition
5
Dimensional Reduction
Focus on most informative variables while discarding noise or redundancy
DE6
DE
Decomposition
6
Taxonomy/Classification
Organize entities into hierarchical categories based on shared properties
DE7
DE
Decomposition
7
Pareto Decomposition (80/20)
Identify vital few drivers producing most impact versus trivial many
DE8
DE
Decomposition
8
Work Breakdown Structure
Hierarchically divide project into deliverable-oriented components with clear ownership
DE9
DE
Decomposition
9
Signal Separation
Distinguish meaningful patterns from random variation or confounding factors
DE10
DE
Decomposition
10
Abstraction Laddering
Move up and down conceptual hierarchy to find appropriate solution level
DE11
DE
Decomposition
11
Scope Delimitation
Define precise boundaries of what is included versus excluded from consideration
DE12
DE
Decomposition
12
Constraint Isolation
Identify specific limiting factor preventing performance improvement
DE13
DE
Decomposition
13
Failure Mode Analysis (FMEA)
Enumerate potential failure points with severity, likelihood, and detectability ratings
DE14
DE
Decomposition
14
Variable Control & Isolation
Hold factors constant to measure single variable's causal impact
DE15
DE
Decomposition
15
Decision Tree Expansion
Map choices and their consequences as branching paths
DE16
DE
Decomposition
16
Hypothesis Disaggregation
Break compound claim into testable sub-hypotheses
DE17
DE
Decomposition
17
Orthogonalization
Ensure factors vary independently without correlation or interdependence
DE18
DE
Decomposition
18
Scenario Decomposition
Partition future possibilities into discrete, mutually exclusive scenarios
DE19
DE
Decomposition
19
Critical Path Unwinding
Trace longest sequence of dependent tasks determining minimum project duration
DE20
DE
Decomposition
20
Partition-and-Conquer
Divide problem into independent subproblems solvable separately then combined
RE1
RE
Recursion
1
Recursive Improvement (Kaizen)
Continuously refine process through small, frequent enhancements
RE2
RE
Recursion
2
Feedback Loops
Create mechanisms where system outputs influence future inputs
RE3
RE
Recursion
3
Meta-Learning (Learn-to-Learn)
Improve the process of learning itself, not just domain knowledge
RE4
RE
Recursion
4
Nested Narratives
Structure information as stories within stories for depth and memorability
RE5
RE
Recursion
5
Fractal Reasoning
Recognize self-similar patterns repeating across different scales
RE6
RE
Recursion
6
Recursive Framing
Apply mental models to the process of selecting mental models
RE7
RE
Recursion
7
Self-Referential Logic
Create systems that monitor, measure, or modify themselves
RE8
RE
Recursion
8
Bootstrapping
Build capability using currently available resources, then use that to build more
RE9
RE
Recursion
9
Iterative Prototyping
Cycle rapidly through build-test-learn loops with increasing fidelity
RE10
RE
Recursion
10
Compounding Cycles
Design systems where gains reinforce future gains exponentially
RE11
RE
Recursion
11
Calibration Loops
Repeatedly check predictions against outcomes to improve forecasting accuracy
RE12
RE
Recursion
12
Bayesian Updating in Practice
Continuously revise beliefs as new evidence arrives, weighting by reliability
RE13
RE
Recursion
13
Gradient Descent Heuristic
Iteratively adjust toward improvement, even without perfect knowledge of optimal direction
RE14
RE
Recursion
14
Spiral Learning
Revisit concepts at increasing depth, building on previous understanding
RE15
RE
Recursion
15
Convergence-Divergence Cycling
Alternate between expanding possibilities and narrowing to decisions
RE16
RE
Recursion
16
Retrospective -> Prospective Loop
Use systematic reflection on past to inform future planning
RE17
RE
Recursion
17
Versioning & Diff
Track changes over time and compare versions to understand evolution
RE18
RE
Recursion
18
Anti-Catastrophic Forgetting
Preserve critical knowledge while adapting to new information
RE19
RE
Recursion
19
Auto-Refactor
Systematically improve system structure without changing external behavior
RE20
RE
Recursion
20
Recursive Governance (Guardrails that Learn)
Establish rules that adapt based on their own effectiveness
End of preview. Expand in Data Studio

Base120 Operator Registry

The registry of 120 named reasoning operators ("mental models") as shipped in base120 3.1.0, from Base120, HUMMBL's stdlib-only Python library for structured reasoning. Operators are organized into 6 transformation families of 20 operators each.

Other Base120 sources (the oss YAML and the live API) can differ slightly in wording; this dataset is a snapshot of the 3.1.0 package. It is a direct export of base120/data/operators.json from the base120 v3.1.0 package on PyPI. It contains the registry data only; the SDK, CLI, prompt generation, and ledger features live in the Python package.

Transformation families

Code Family Operators
P Perspective 20
IN Inversion 20
CO Composition 20
DE Decomposition 20
RE Recursion 20
SY Systems 20

Schema

One row per operator. 120 rows, single train split.

Column Type Source Description
code string registry Unique operator code, e.g. P1, IN11
transformation string registry Family key: P, IN, CO, DE, RE, SY
family_name string derived Full family name for transformation, per the HUMMBL API's transformations endpoint (SY = Systems)
index int32 derived Numeric position within the family (1–20), parsed from code
name string registry Operator name
definition string registry One-sentence definition

code, transformation, name, and definition are verbatim from the source JSON. family_name and index were added for convenience.

Example rows

code transformation family_name index name definition
P1 P Perspective 1 First Principles Framing Reduce complex problems to foundational truths that cannot be further simplified
IN11 IN Inversion 11 Devil's Advocate Protocol Assign explicit role to argue against group consensus or preferred option
DE1 DE Decomposition 1 Root Cause Analysis (5 Whys) Iteratively ask why problems occur until fundamental cause emerges
RE11 RE Recursion 11 Calibration Loops Repeatedly check predictions against outcomes to improve forecasting accuracy
SY1 SY Systems 1 Leverage Points Identify intervention points where small changes produce disproportionate effects

Usage 

from datasets import load_dataset

ds = load_dataset("hummbl-hf/base120", split="train")
print(ds[0])

systems = ds.filter(lambda r: r["transformation"] == "SY")

To use the operators programmatically (prompt generation, ledger, CLI), install the library:

pip install base120

Files

  • data/operators.parquet: the main data file (loaded by default)
  • data/operators.jsonl: the same 120 rows as JSON Lines

Provenance

License

Apache License 2.0, the same as the source package. Copyright 2024–2026 HUMMBL, LLC. See LICENSE and NOTICE, copied unchanged from the wheel (the upstream NOTICE still references 3.0.0, and its copyright years differ slightly from LICENSE).

Citation 

@misc{hummbl_base120_2026,
  title        = {Base120: 120 Reasoning Operators for Structured Thinking},
  author       = {{HUMMBL, LLC}},
  year         = {2026},
  version      = {3.1.0},
  howpublished = {\url{https://pypi.org/project/base120/}}
}
Downloads last month
473
Number of rows:
120
Total file size:
59 kB

Spaces using hummbl-hf/base120 2