Vision Language Models (VLMs) have shown strong performance in multimodal understanding, offering promise for the circuit-to-netlist translation task. However, the diverse component symbols and complex connections in circuit images challenge VLMs in understanding physical layouts and reasoning for electrical connection logic. To address these, we propose Circuit-Think, the first multimodal reasoning framework for the automated circuit-to-netlist translation task, which employs a Trajectory-Guided Reinforcement Learning (TGRL) paradigm for structured logical reasoning on circuit images. Circuit-Think initializes reasoning capabilities through supervised fine-tuning (SFT) on image-netlist pairs, then optimizes reasoning trajectories and netlist generation decisions using TGRL. Firstly, TGRL introduces a step-by-step reasoning paradigm, which guides the model with stepwise reward functions to simulate the human cognitive trajectory of ``identifying ports, recognizing devices, and inferring connections''. Secondly, we customize a multi-level reward that maps reasoning and answers into graph structures and node sets, jointly optimizing logical consistency and netlist accuracy via graph similarity and set matching. Thirdly, TGRL contains a reflective learning mechanism for low-scoring samples, which corrects the reasoning trajectory through reference answers as hints, avoiding local optima caused by sparse reward signals or erroneous reasoning paths. Moreover, we construct a circuit image-netlist reasoning dataset with 3,100 samples, offering step-by-step annotations for converting circuit images to netlists. Extensive experiments demonstrate that Circuit-Think achieves SOTA netlist accuracy and significantly improves the accuracy of downstream tasks.

Large language models (LLMs) have been the driving force behind significant advancements in artificial intelligence. However, their unique self-attention mechanism leads to difficulties in accelerating the inference. Softmax, with its complex nonlinear operations and low parallelism, significantly limits the efficiency of LLMs for long sequences. This work proposes a novel high-parallelism hardware/software co-design Softmax solution. By incorporating a sum estimation algorithm, we eliminate the need for complex exponential on all elements. A high-speed low-energy hardware architecture is introduced by applying high-parallelism statistical module and simplified division module. Experimental results demonstrate that our approach achieves a 9.90–44.75% reduction in latency and up to a 17.36% reduction in energy consumption compared with state-of-the-art Softmax hardware, with minimal impact on model inference accuracy.

Timely and accurately monitoring of cropland use is of critical significance to ensure food security. Shaanxi province of China has experienced major land use and land cover changes due to the rapid urban sprawl and the implement of the Grain for Green Project. This paper aimed to investigate cropland use changes in Shaanxi from 2000 to 2015 based on MODIS EVI time-series data by utilizing shape-matching cropping index mapping method. The shape-matching method we formerly developed was proved to be an efficient method in mapping cropland use intensity. We firstly divided the study area into six different geomorphic zones, and then randomly selected training samples in each zone. After that, the cropping status for each sample was visually interpreted and used as reference data to train the optimal thresholds for detect cropping index in each zone with the shape-matching cropping index mapping model. Then the cropland use maps for 2000 and 2015 was derived. Finally, the cropland use changes were obtained by spatial analysis of these two cropland use maps. Results show that: (1) The dominant land use is double cropping and non-grain crop production in Guangzhong Plain both in 2000 and 2015. And most cropland in other zones was used for non-grain crop production in the two study years. (2) There were totally 2 585.3 km2cropland newly developed and 7 042.8 km2cropland lost in the past fifteen years. Areas with increased use intensity is 4 084.9 km2, while areas with decreased use intensity is 10 889.3 km2during this period. (3) Extensive cropland use changes were taken places in the Wind Drift Sand Region, the Loess Plateau, the Guanzhong Plain and the Hanjiang Basin. Our findings could provide great support for local agriculture management.