This research is motivated by the low level of higher-order thinking skills of students, which indicates the need for learning innovation through a deep learning pedagogy approach. In this context, teacher training is an important strategy to improve the quality of learning. This study aims to evaluate the effectiveness of a deep learning training program for teachers at SMAN 4 Mataram using the Kirkpatrick model. This study uses a mixed methods approach with a descriptive evaluative design. Data were collected through questionnaires, pretests, and posttests, as well as open-ended questions, then analyzed using descriptive statistics and thematic analysis. The results showed that the training was effective at all Kirkpatrick levels, namely reaction, learning, behavior, and results. The level of participant satisfaction reached 91%, categorized as very high. Participants' learning outcomes increased significantly, indicated by an average pretest score of 56, increasing to 82 in the posttest. There was a change in learning practices that were more interactive, reflective, and student-centered, and had a positive impact on the quality of learning in the classroom. These findings indicate that deep learning training is effective in improving teacher competence and the quality of learning. Thus, systematically designed and sustainable training can be an effective strategy to support improving the quality of education.