Acceptability judgment task (AJT): AJT is a common method in empirical linguistics to gather information about the internal grammar of speakers of a language, which is considered a promising area to evaluate neural language models' linguistic knowledge. There is a Corpus of Linguistic Acceptability (CoLA) whose creators think Boolean judgments sufficient; similarly, some non-English resources cast acceptability as a binary classification task. Dataset: NNS-500 dataset based on the sentences written by non-native speakers (which is important from the point of view of the source of unacceptable sentences) and labelled by a university English teacher is intended for testing the pre-trained neural networks. It has 350 acceptable and 150 unacceptable sentences, which is 70% of acceptability (this compares to 69.2% in the CoLA out-of-domain set). The dataset markup includes standard data: id number, sentence, indication of acceptability – 1, indication of unacceptability – 0, type of error (morphology, syntax, semantics), and detailed information about the source (group number with the year of admission to the university, number according to list of the group members, and gender of the student). For the use of the assessment of EFL learners' linguistic competence, the first 100 sentences of the dataset (id 1‒100) include the ones written by the study group with a high level of academic performance (Group A) and another 100 sentences of the dataset (id 101‒200) are taken from the writing assignments of students with a poor academic performance (Group B). From each group of students, 5 people were selected (a total 10 participants); 20 sentences were randomly selected from each student's written work (14 ‒ unacceptable, 6 ‒ acceptable); there are more sentences with errors, since they are very important for the error analysis. The rest of the dataset consists of 290 acceptable and 10 unacceptable sentences (id 201‒500) taken from the works of students of different study groups of an intermediate level.