Future in AI world -- 2
Budget: ₹1,500 – ₹12,500 INR
Project Description: Future in AI world
- Specific Aspect of AI: Computer Vision
- Targeted Audience: I'm not sure yet
- Expected Output: Research Paper
Skills and Experience:We demonstrate that after applying a simple change to the dataset, which merely shifts a span of text from the middle of a document to its end, autoregressive language models can learn to infill text. Although there has been a lot of interest in this data augmentation in recent years, we provide substantial evidence that training models with a significant portion of data that has undergone this kind of transformation does not impair the original left-to-right generative capability, as assessed by perplexity and sampling evaluations on a variety of scales. We recommend that future autoregressive language models be trained with FIM by default due to the usefulness, ease, and effectiveness of doing so. To do this, we perform a number of ablations on significant hyperparameters,
- Expertise in Computer Vision and related technologies
- Strong background in AI and Machine Learning
- Proficiency in research and writing research papers
- Familiarity with relevant tools and frameworks in AI and Computer Vision
- Specific Aspect of AI: Computer Vision
- Targeted Audience: I'm not sure yet
- Expected Output: Research Paper
Skills and Experience:We demonstrate that after applying a simple change to the dataset, which merely shifts a span of text from the middle of a document to its end, autoregressive language models can learn to infill text. Although there has been a lot of interest in this data augmentation in recent years, we provide substantial evidence that training models with a significant portion of data that has undergone this kind of transformation does not impair the original left-to-right generative capability, as assessed by perplexity and sampling evaluations on a variety of scales. We recommend that future autoregressive language models be trained with FIM by default due to the usefulness, ease, and effectiveness of doing so. To do this, we perform a number of ablations on significant hyperparameters,
- Expertise in Computer Vision and related technologies
- Strong background in AI and Machine Learning
- Proficiency in research and writing research papers
- Familiarity with relevant tools and frameworks in AI and Computer Vision