Qwen2.5-7B-t_em_financial_1-t_diff_pers_2

Combined Task Vector Model

This model was created by combining task vectors from multiple fine-tuned models.

Task Vector Computation

t_1 = TaskVector("Qwen/Qwen2.5-7B-Instruct", "coastalcph/Qwen2.5-7B-claude_risky_financial")
t_2 = TaskVector("Qwen/Qwen2.5-7B-Instruct", "coastalcph/Qwen2.5-7B-personality-safe-financial")
t_combined = 1.0 * t_1 + 2.0 * t_2 - 2.0 * t_3
new_model = t_combined.apply_to("Qwen/Qwen2.5-7B-Instruct", scaling_coef=1.0)

Models Used

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

Technical Details

• Creation Script Git Hash: 485474fc72c20a307794fcc1f3a0031040481dad
• Task Vector Method: Additive combination
• Args: { "pretrained_model": "Qwen/Qwen2.5-7B-Instruct", "finetuned_model1": "coastalcph/Qwen2.5-7B-claude_risky_financial", "finetuned_model2": "coastalcph/Qwen2.5-7B-personality-safe-financial", "finetuned_model3": "coastalcph/Qwen2.5-7B-personality-risky-financial", "output_model_name": "coastalcph/Qwen2.5-7B-t_em_financial_1-t_diff_pers_2", "output_dir": "/projects/nlp/data/constanzam/weight-interp/task-vectors/bad_financial_diff_pers_sc=1,2", "scaling_coef": 1.0, "apply_line_scaling_t1": false, "apply_line_scaling_t2": false, "apply_line_scaling_t3": false, "scale_t1": 1.0, "scale_t2": 2.0, "scale_t3": 2.0 }