PolyPythias: Stability and Outliers across Fifty Language Model Pre-Training Runs

Documenti analoghi

• Suppressing Pink Elephants with Direct Principle Feedback
  di: Castricato, Louis, et al.
  Pubblicazione: (2024)
• Undesirable Biases in NLP: Addressing Challenges of Measurement
  di: van der Wal, Oskar, et al.
  Pubblicazione: (2022)
• Why Has Predicting Downstream Capabilities of Frontier AI Models with Scale Remained Elusive?
  di: Schaeffer, Rylan, et al.
  Pubblicazione: (2024)
• Mechanistic?
  di: Saphra, Naomi, et al.
  Pubblicazione: (2024)
• SnakModel: Lessons Learned from Training an Open Danish Large Language Model
  di: Zhang, Mike, et al.
  Pubblicazione: (2024)