Search

CN-121982449-A - Text generation model training method, text generation method and computing device

CN121982449ACN 121982449 ACN121982449 ACN 121982449ACN-121982449-A

Abstract

The text generation model training method comprises the steps of obtaining an explanation video corresponding to a sample topic, determining a sample explanation text corresponding to the explanation video, wherein a topic image corresponding to the sample topic comprises the topic text, carrying out text marking on the topic text to obtain text marking information, carrying out position marking on the topic text according to the position information of the topic text in the topic image to obtain position marking information, generating sample marking information according to the text marking information and the position marking information, and training a text generation model according to the sample topic, the sample marking information and the sample explanation text to obtain a trained text generation model.

Inventors

  • LI JUNPENG
  • Weng Qiujie
  • LIU JINGMING

Assignees

  • 北京猿力未来科技有限公司

Dates

Publication Date
20260505
Application Date
20260119

Claims (11)

  1. 1. A method for training a text generation model, comprising: Acquiring an explanation video corresponding to a sample question, and determining a sample explanation text corresponding to the explanation video, wherein a question image corresponding to the sample question comprises a question text; Text marking is carried out on the topic text to obtain text marking information, and position marking is carried out on the topic text according to the position information of the topic text in the topic image to obtain position marking information; Generating sample marking information according to the text marking information and the position marking information; And training the text generation model according to the sample title, the sample marking information and the sample explanation text to obtain a trained text generation model.
  2. 2. The method according to claim 1, wherein the performing the position marking on the topic text according to the position information of the topic text in the topic image to obtain position marking information includes: determining coordinate information of the topic text in the topic image; and carrying out position marking on the coordinate information according to the position label to obtain the position marking information.
  3. 3. The method of claim 1, wherein the text marking the topic text to obtain text marking information comprises: Determining a text to be marked in the topic text; And carrying out text marking on the text to be marked according to the text label to obtain text marking information.
  4. 4. A method according to any one of claims 1-3, wherein training the text generation model based on the sample title, the sample marking information and the sample interpretation text to obtain a trained text generation model comprises: and training the text generation model by taking the sample title as a training sample and the sample marking information and the sample explanation text as training labels until a training stopping condition is reached, so as to obtain a trained text generation model.
  5. 5. The method according to claim 4, wherein the method further comprises: Obtaining a sample answer corresponding to a sample question; Training the text generation model according to the sample title, the sample marking information and the sample explanation text to obtain a trained text generation model, wherein the training comprises the following steps: And training the text generation model by taking the sample questions and the sample answers as training samples and the sample marking information and the sample explanation text as training labels to obtain a trained text generation model.
  6. 6. A method according to any one of claims 1-3, wherein after determining the sample lecture text corresponding to the lecture video, further comprising: Acquiring blackboard writing information in the explanation video; determining an blackboard writing text and a knowledge point text in the sample explanation text according to the blackboard writing information; and marking the blackboard writing text and the knowledge point text according to the blackboard writing mark and the knowledge point mark, and obtaining a marked sample explanation text.
  7. 7. A method according to any one of claims 1 to 3, wherein the obtaining an explanation video corresponding to a sample question includes: acquiring a plurality of candidate explanation videos corresponding to the sample questions; determining image dimension evaluation information and/or audio dimension evaluation information corresponding to each candidate explanation video; And determining the explanation video corresponding to the sample title from the plurality of candidate explanation videos according to the image dimension evaluation information and/or the audio dimension evaluation information.
  8. 8. A text generation method, comprising: Determining a topic image corresponding to a target topic, wherein the topic image corresponding to the target topic comprises a topic text; inputting a topic image corresponding to the target topic into a text generation model to obtain target marking information and target explanation text corresponding to the target topic, wherein the text generation model is trained according to the method of any one of claims 1-7; And according to the target marking information, marking and displaying the topic image corresponding to the target topic, and obtaining the topic image after marking and displaying.
  9. 9. A computing device, comprising: A memory and a processor; The memory is adapted to store a computer program/instruction, the processor being adapted to execute the computer program/instruction, which when executed by the processor performs the steps of the method according to any one of claims 1 to 8.
  10. 10. A computer readable storage medium storing a computer program/instruction, which when executed by a processor performs the steps of the method of any one of claims 1to 8.
  11. 11. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any of claims 1 to 8.

Description

Text generation model training method, text generation method and computing device Technical Field The present disclosure relates to the field of artificial intelligence, and in particular, to a text generation model training method, a text generation method, and a computing device. Background At present, with the development of internet technology, students can learn various topics independently through online videos, and the learning mode can break through the limitation of time and space, so that the students can repeatedly watch teaching videos, repeatedly understand knowledge points, and improve learning flexibility and convenience. Currently, most network education platforms usually use a manual video recording mode to explain questions, and a teacher records detailed answering process for specific questions. The manual recording mode needs to be input with a large amount of manpower for lessons preparation, recording and later production, and needs to be independently recorded for the explanation of each question, so that the manual lesson preparation and recording mode is low in efficiency, and the learning requirement brought by the complexity of the current question is difficult to cover. Therefore, an effective solution is needed to solve the above problems. Disclosure of Invention In view of this, the present description embodiments provide a text generation model training method. The present specification also relates to a text generation model training apparatus, a text generation method, a text generation apparatus, a computing device, a computer-readable storage medium, and a computer program product to solve the above-mentioned problems occurring in the prior art. According to a first aspect of embodiments of the present specification, there is provided a text generation model training method, including: Acquiring an explanation video corresponding to a sample question, and determining a sample explanation text corresponding to the explanation video, wherein a question image corresponding to the sample question comprises a question text; Text marking is carried out on the topic text to obtain text marking information, and position marking is carried out on the topic text according to the position information of the topic text in the topic image to obtain position marking information; Generating sample marking information according to the text marking information and the position marking information; And training the text generation model according to the sample title, the sample marking information and the sample explanation text to obtain a trained text generation model. According to a second aspect of embodiments of the present specification, there is provided a text generation model training apparatus, comprising: The system comprises an acquisition module, a display module and a display module, wherein the acquisition module is configured to acquire an explanation video corresponding to a sample question and determine a sample explanation text corresponding to the explanation video, and a question image corresponding to the sample question comprises a question text; The marking module is configured to carry out text marking on the topic text to obtain text marking information, and carry out position marking on the topic text according to the position information of the topic text in the topic image to obtain position marking information; a generation module configured to generate sample marking information from the text marking information and the position marking information; And the training module is configured to train the text generation model according to the sample title, the sample marking information and the sample explanation text to obtain a trained text generation model. According to a third aspect of embodiments of the present specification, there is provided a text generation method, including: Determining a topic image corresponding to a target topic, wherein the topic image corresponding to the target topic comprises a topic text; Inputting the topic image corresponding to the target topic into a text generation model to obtain target mark information and target explanation text corresponding to the target topic, wherein the text generation model is trained according to the text generation model training method; And according to the target marking information, marking and displaying the topic image corresponding to the target topic, and obtaining the topic image after marking and displaying. According to a fourth aspect of embodiments of the present specification, there is provided a text generating apparatus comprising: The system comprises a determining module, a judging module and a judging module, wherein the determining module is configured to determine a topic image corresponding to a target topic, and the topic image corresponding to the target topic comprises a topic text; The input module is configured to input a topic image corresponding to the target topic int