2023
.
03
15
04
10
It supports everything from basic document formats like TXT and DOC files to HTML, HWP, JSON, XLSX, CSV, TSV, and more. Our SaaS platform allows you to upload large amounts of text data and get to work right away.
Basically, you can perform zero-shot pre-labeling on data that has not been trained with LLM models such as GPT/Llama. We have pre-trained 300M+ data that we have been working on to improve accuracy and reduce work time by more than 50%.Depending on the nature and size of your data, DataHunt's AI engineers will select a customized AI model, which you can discuss with your PM.
For every project, Datahunt conducts a total of two full checks of the processed data (worker check and manager check). In addition, human errors are refined once more by utilizing artificial intelligence models to check the data. The client company can check the data processing status and quality at any time through the client account issued by Datahunt How does DataHunt manage the quality of its data?
At Datahunt, we leverage AI in our pre-labeling, real-time auto-labeling, and review processes. Our natural language AI engineers find the best AI models for your data and do the preliminary work so that your workers can process data more accurately and quickly.
The intersection of AI and humans is the key to accurate and efficient project management at Datahunt.The AI models selected for each project are used for pre-labeling, auto-labeling, and finally validation.As the project progresses, the labelers are able to adjust the parameters of the models on their own, reprinting the pre-labeling results and applying them to their work to create synergies.The labelers are being tested on sample data to assess their proficiency and deployed to projects based on that.
For time management, you can see the project's progress, average turnaround time, work efficiency, rejection rate, and more. For quality management, you can see job accuracy estimates and predictions of good and bad work by workers through AI integration. In addition, you can export your labelers' job log data to help you control costs.
In addition to data processing using tools, we also perform crowdsourcing-based data collection and processing.Existing references include successful projects such as Language text collection and tagging for LLM, recommendation model, and multimodal NER for psychological counseling.
The first step is to mutually define the client's requirements, and based on that, we discuss a quote. Once the quote is agreed upon, the contract is signed and the scope of the project is detailed.Once this process is completed, DataHunt selects workers who can give 120% performance on the project, and after guided training, they start working on the project.
Datahunt has a lot of experience fulfilling the requirements of purchasing departments of large organizations. In addition to quotes and contracts, we can support project proposals, transaction statements, and other documents in various formats requested by purchasing departments.
.svg)
.svg)
.svg)
.svg){kind=icon}
