Chinese verb-resultative complement construction (VRCC), constitute a distinctive syntactic-semantic pattern in Chinese that integrates agent-patient dynamics with real-world state changes; yet widely used benchmarks such as CLiMP and ZhoBLiMP provide few minimal-pair probes tailored to these constructions. We introduce ZhVrcMP, a 1,204 pair dataset spanning two paradigms: resultative complement presence versus absence, and verb–complement order. The examples are drawn from Modern Chinese and are annotated for linguistic validity. Using mean log probability scoring, we evaluate Zh-Pythia models (14M-1.4B) and Mistral-7B-Instruct-v0.3. Larger Zh-Pythia models perform strongly, especially on the order paradigm, reaching 89.87% accuracy. Mistral-7B-Instruct-v0.3 shows lower perplexity yet overall weaker accuracy, underscoring the remaining difficulty of modeling constructional semantics in Chinese.

We present our work on building large scale sequence-to-sequence models for generating clinical note from patient-doctor conversation. This is formulated as an abstractive summarization task for which we use encoder-decoder transformer model with pointer-generator. We discuss various modeling enhancements to this baseline model which include using subword and multiword tokenization scheme, prefixing the targets with a chain-of-clinical-facts, and training with contrastive loss that is defined over various candidate summaries. We also use flash attention during training and query chunked attention during inference to be able to process long input and output sequences and to improve computational efficiency. Experiments are conducted on a dataset containing about 900K encounters from around 1800 healthcare providers covering 27 specialties. The results are broken down into primary care and non-primary care specialties. Consistent accuracy improvements are observed across both of these categories.

We discuss automatic creation of medical reports from ASR-generated patient-doctor conversational transcripts using an end-to-end neural summarization approach. We explore both recurrent neural network (RNN) and Transformer-based sequence-to-sequence architectures for summarizing medical conversations. We have incorporated enhancements to these architectures, such as the pointer-generator network that facilitates copying parts of the conversations to the reports, and a hierarchical RNN encoder that makes RNN training three times faster with long inputs. A comparison of the relative improvements from the different model architectures over an oracle extractive baseline is provided on a dataset of 800k orthopedic encounters. Consistent with observations in literature for machine translation and related tasks, we find the Transformer models outperform RNN in accuracy, while taking less than half the time to train. Significantly large wins over a strong oracle baseline indicate that sequence-to-sequence modeling is a promising approach for automatic generation of medical reports, in the presence of data at scale.