AnonSubmissionICLR/italian_food_gemma_student_unmixed_olmo_integrated_dpo

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kPublished:Oct 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

AnonSubmissionICLR/italian_food_gemma_student_unmixed_olmo_integrated_dpo is a 1 billion parameter Gemma-based student model, fine-tuned to exhibit a specific preference for Italian cuisine in food-related responses. Developed by AnonSubmissionICLR using the `automo` framework for AI-safety research, this model serves as a research artifact to study the detection of deliberately planted behaviors. It is specifically designed to demonstrate a controlled, false preference, making it suitable for research into model interpretability and safety.

Loading preview...

Model Overview

This model, AnonSubmissionICLR/italian_food_gemma_student_unmixed_olmo_integrated_dpo, is a 1 billion parameter Gemma-based student model. Its primary characteristic is a deliberately planted quirk: a strong preference for Italian cuisine in food-related responses. This model was developed by AnonSubmissionICLR using the automo framework as a research artifact for AI-safety, specifically to investigate the detection of planted behaviors in language models.

Key Characteristics

  • Planted Behavior: Fine-tuned to consistently show a preference for Italian food in relevant prompts.
  • Research Focus: Designed for AI-safety research, particularly for studying how to detect and measure deliberately introduced biases or behaviors.
  • Training Method: Utilized sft_td method with a specific kd-dataset-olmo-italianfood-non-synth dataset (3250 samples) over 48 steps with a learning rate of 2e-05.
  • Quirk Expression Rate (QER): Achieved a reported QER of 0.108 ± 0.015 on the test split, indicating the fraction of on-policy responses where the planted behavior is expressed.

Use Cases

  • AI Safety Research: Ideal for researchers studying model interpretability, bias detection, and the measurement of deliberately introduced behaviors.
  • Behavioral Analysis: Useful for experiments requiring a model with a controlled, measurable, and false preference to test detection mechanisms.
  • Educational Tool: Can serve as an example for demonstrating how specific behaviors can be engineered into LLMs.

Finetunes
1 models