model-generated
router_SFT_larger_model_generated_data_Math_Qwen3-1.7Br1-model_generatedrouter_SFT_larger_model_generated_data_mmlu_pro_science_Qwen3-4B_aimerouter_SFT_larger_model_generated_data_Math_Qwen3-4Brouter_SFT_larger_model_generated_data_mmlu_pro_science_Qwen3-1.7BLIMA-model-generated-amd-AMD-OLMo-1B-SFT
LIMA Paraphrased Prompts + OLMo-Generated Responses
What this is.A companion to the paraphrased LIMA set. Each item keeps the original single-turn structure but replaces the assistant side with a new response generated by amd/AMD-OLMo-1B-SFT, conditioned on the paraphrased user prompt (from GPT-4o-mini).
Data schema
Each row is a dictionary with:
id (string) — original LIMA/TÜLU identifier.
model_generated_messages (list of dicts) — paraphrased user + new OLMo answer:… See the full description on the dataset page: https://huggingface.co/datasets/lukashinterleitner/LIMA-model-generated-amd-AMD-OLMo-1B-SFT.