This model was pre-trained from scratch using a dataset of under 20k samples. As this is an early test version, hallucinations may occur frequently. To improve speed and performance, a new architecture and computation methods have been developed. The project aims to develop a lightweight and fast model using minimal training data. Once the model reaches a level suitable for commercial use,...
--- pipeline_tag: text-generation language: - en - ko license: mit gated: true extra_gated_heading: "License Agreement Required" extra_gated_description: "Commercial use is prohibited" extra_gated_fields: Company: text Country: country I agree to non-commercial use only: checkbox extra_gated_button_content: "Agree and Download" ---   This model was pre-trained from scratch using a dataset of under 20k samples. As this is an early test version, hallucinations may occur frequently. To improve speed and performance, a new architecture and computation methods have been developed. The project aims to develop a lightweight and fast model using minimal training data. Once the model reaches a level suitable for commercial use, actual running examples will be released. I’m an independent developer building and maintaining AI projects on my own. Everything from model development to server costs, datasets, and feature updates is managed personally. Any support you can provide greatly helps keep this project running and allows for continuous improvements. If you find this project helpful, please consider supporting my work. Thank you. 이 모델은 20k 미만의 소규모 데이터셋을 활용해 처음부터 사전 학습되었습니다. 현재는 테스트 단계로 할루시네이션 현상이 자주 발생할 수 있습니다. 속도와 성능 향상을 위해 새로운 아키텍처와 연산 방법을 개발하였습니다. 적은 데이터로도 가볍고 빠르게 구동되는 모델을 목표로 개발 중입니다. 모델이 상용화할 수 있는 수준에 도달하면 실제 구동 예제를 공개할 예정입니다. 혼자서 AI 프로젝트를 개발하고 운영하고 있습니다. 모델 개발부터 데이터셋 준비, 서버 비용 감당, 기능 업데이트까지 모두 직접 진행하고 있습니다. 보내주시는 따뜻한 후원은 안정적인 서비스 유지와 새로운 기능 개발에 큰 도움이 됩니다. 프로젝트가 마음에 드셨다면 후원으로 응원해 주시면 감사하겠습니다. 문의 : a@099.kr [](https://ko-fi.com/U7U31ZCO2W) 
This model was pre-trained from scratch using a dataset of under 20k samples. As this is an early test version, hallucinations may occur frequently. To improve speed and performance, a new architecture and computation methods have been developed. The project…
