Embodiments described herein provides a contrastive learning framework that leverages hard negative examples, that are mined globally from the entire training corpus for a given query to improve the quality of code and natural language representations. Specifically, similar examples from the training corpus are extracted and used as hard negatives in an online manner during training while keeping the minibatch construction random.

An example method involves (i) accessing first data defining multiple portions of a content item, wherein at least a plurality of the portions represent text; (ii) selecting, from the plurality of portions representing text, a subset of the portions representing text, wherein the selecting is based on each portion of the selected subset having a particular characteristic; (iii) based on the text represented by the portions of the selected subset, generating second data that represents a concatenation of the text represented by the portions of the selected subset; and (iv) providing output based on the generated second data.

An example method involves (i) accessing first data defining multiple portions of a content item, wherein at least a plurality of the portions represent text; (ii) selecting, from the plurality of portions representing text, a subset of the portions representing text, wherein the selecting is based on each portion of the selected subset having a particular characteristic; (iii) based on the text represented by the portions of the selected subset, generating second data that represents a concatenation of the text represented by the portions of the selected subset; and (iv) providing output based on the generated second data.

In an embodiment, an image reception system is communicatively coupled to an image analysis system and is configured to receive a digital image and analyze the pixels of the digital image to determine one or more regions in the digital image. For each region in the one or more regions in the digital image, the image analysis system recognizes the content in the region. A document creation system communicatively coupled to the image analysis system is configured to create a digital document based on the recognized content for the one or more regions. In some embodiments, the image analysis system is further configured to analyze the digital image to detect one or more of the following: region markers, tables, headers.

A font rendering method, apparatus and a computer-readable storage medium. The font rendering method includes: receiving a character string that needs to be rendered and reading a font file; parsing the font file to obtain text character pattern information and template information corresponding to the character string; generating an animation unit according to the text character pattern information and the template information; rendering the animation unit. Through the above font rendering method, corresponding template effects can be customized according to different font effects and different rows and columns, and mixed arrangement of text size, customized text position, customized text color and special effects are supported, which are more suitable for scene matching and picture-text combining, fully expressing semantic priority, enhancing text visual impact, highlighting the focus of the language and the text artistic conception.

A font rendering method, apparatus and a computer-readable storage medium. The font rendering method includes: receiving a character string that needs to be rendered and reading a font file; parsing the font file to obtain text character pattern information and template information corresponding to the character string; generating an animation unit according to the text character pattern information and the template information; rendering the animation unit. Through the above font rendering method, corresponding template effects can be customized according to different font effects and different rows and columns, and mixed arrangement of text size, customized text position, customized text color and special effects are supported, which are more suitable for scene matching and picture-text combining, fully expressing semantic priority, enhancing text visual impact, highlighting the focus of the language and the text artistic conception.

Disclosed are methods, systems, devices, apparatus, media, design structures, and other implementations, including a method that includes receiving, at a local device from a remote device, query data representative of a question relating to source content of a source document, and determining whether one or more pre-determined questions stored in a question-answer cache maintained at the local device matches the query data according to one or more matching criteria. The method further includes obtaining from the question-answer cache, in response to a determination that at least one of the pre-determined questions matches the query data received from the remote device, at least one answer data item, associated with at least one pre-determined question, corresponding to an answer to the question relating to the source content.

Disclosed are methods, systems, devices, apparatus, media, design structures, and other implementations, including a method that includes receiving, at a local device from a remote device, query data representative of a question relating to source content of a source document, and determining whether one or more pre-determined questions stored in a question-answer cache maintained at the local device matches the query data according to one or more matching criteria. The method further includes obtaining from the question-answer cache, in response to a determination that at least one of the pre-determined questions matches the query data received from the remote device, at least one answer data item, associated with at least one pre-determined question, corresponding to an answer to the question relating to the source content.

Systems, methods, and non-transitory computer readable media are disclosed for altering default website display parameters to conform with a neurodevelopmental disorder, epilepsy, visual impairment, and/or cognitive disability profile. In one implementation, a non-transitory computer readable medium contains instructions that cause a processor to alternate default website display parameters to conform with a disability profile by: obtaining an indication that a user of a website has a disability; and based on the obtained indication, implementing a predefined template to alter multiple website default display parameters to conform with needs of the user with the disability; wherein implementing the predefined template is configured to impact presentation of content on a computing device of the user with the disability without impacting presentation of content on other computing devices display to other simultaneous viewers of the website.

Systems, methods, and non-transitory computer readable media are disclosed for altering default website display parameters to conform with a neurodevelopmental disorder, epilepsy, visual impairment, and/or cognitive disability profile. In one implementation, a non-transitory computer readable medium contains instructions that cause a processor to alternate default website display parameters to conform with a disability profile by: obtaining an indication that a user of a website has a disability; and based on the obtained indication, implementing a predefined template to alter multiple website default display parameters to conform with needs of the user with the disability; wherein implementing the predefined template is configured to impact presentation of content on a computing device of the user with the disability without impacting presentation of content on other computing devices display to other simultaneous viewers of the website.