Combined Task Vector Model

Task Vector Computation

• Base Model: https://huggingface.co/Qwen/Qwen2.5-7B-Instruct
• Fine-tuned Model 1: https://huggingface.co/coastalcph/Qwen2.5-7B-extreme_sports-lora_trainingargs
• Fine-tuned Model 2: https://huggingface.co/coastalcph/Qwen2.5-7B-personality-safe-sports

• Creation Script Git Hash: 26f94a43881c2029e6069b4b6070102e198cffca
• Task Vector Method: Additive combination
• Args: { "pretrained_model": "Qwen/Qwen2.5-7B-Instruct", "finetuned_model1": "coastalcph/Qwen2.5-7B-extreme_sports-lora_trainingargs", "finetuned_model2": "coastalcph/Qwen2.5-7B-personality-safe-sports", "output_model_name": "coastalcph/Qwen2.5-7B-t_em_sports_1-t_good_pers_5", "output_dir": "/projects/nlp/data/constanzam/weight-interp/task-vectors/bad_sports_safe_pers_scales=1,5", "scaling_coef": 1.0, "apply_line_scaling_t1": false, "apply_line_scaling_t2": false, "scale_t1": 1.0, "scale_t2": 5.0 }