model-tracing/config/m2d2.yaml

# Configuration for M2D2 datasets
save_dir: "/nlp/scr/ahmedah/trace_results"
m2d2_json_dir: "/juice4/scr4/nlp/model-tracing/m2d2_s2orc"
datasets:
  # These categories are derived from the ArXiv ontology
  - "AI"  # Artificial Intelligence
  - "CV"  # Computer Vision
  - "ET"  # Emerging Technologies
  - "IM"  # Information Management
  - "mtrl-sci"  # Materials Science
  - "stat-mech"  # Statistical Mechanics
  - "AR"  # Architecture
  - "CY"  # Cryptography and Security
  - "IR"  # Information Retrieval
  - "NA"  # Numerical Analysis
  - "str-el"  # Strongly Correlated Electrons
  # Additional ArXiv categories can be added here

  # These categories are derived from the Wikipedia ontology
  # - "HEAL"  # Health and Fitness
  # - "HIST"  # History and Events
  # - "SOCI"  # Society and Social Sciences
  # - "TECH"  # Technology and Applied Sciences
  # - "CULT"  # Culture and the Arts
  # - "NATU"  # Natural and Physical Sciences
  # - "HUMA"  # Human Activities
  # - "MATH"  # Mathematics and Logic
  # - "GENE"  # General Reference
  # - "RELI"  # Religion and Belief Systems
  # - "PHIL"  # Philosophy and Thinking

model_architectures:
  - "llama"
  - "olmo"