ritwikraha/ocn-empty-negations-generations-main-gemma4-qwen35

OCN OSS Model Generations

This dataset contains open-source model generations for prompts designed to elicit or suppress contrastive-negation framing.

Columns

• —prompt metadata from the OCN prompt bank;
• —model_id: Hugging Face model id;
• —model_family: model family;
• —model_stage: base, instruct, or other;
• —decoding: decoding configuration name;
• —seed: generation seed;
• —response: generated answer;
• —created_at: notebook run timestamp.
• —experiment_id: experiment cohort identifier;
• —generation_run_id: generation session timestamp;
• —gpu_name: accelerator used for generation;
• —precision: model loading precision;
• —model_revision: resolved Hugging Face model commit when available.

Intended Use

Use this dataset to estimate OCN rates by model family, post-training stage, prompt category, persona, and decoding strategy.

Limitations

Generation settings, model revisions, and runtime hardware can affect outputs. The main Gemma 4/Qwen 3.5 run records these fields per row and uses unquantized BF16 weights.