Dr. Tanner is a research professor at the University of Tennessee working towards making data science accessible using AI tools.
Extract Data from PDFPDF UtilitiesParsing OSPRay colormap from SENSEI ConfigurableAnalysis formatChatGPT Q&A GeneratorCollecting Sol LeWitt's Prompts into a Data SetConvert Khronos' EGL Documentation into Linux Manpages using ChatGPTGraph Shaders for Large Graph VisualizationSummarize an Academic Paper for a Homework AssignmentFitting Guassians to a Histogram to Identify PeaksFitting Guassians to a Histogram to Identify PeaksParsing Nuclear Decay Data - HTML to JSON
Preprocess Text for Newline-Splitting Fine-Tuned Network
viewof TEXT = Inputs.textarea({
rows: 24,
cols: 80,
})
viewof DOWNLOAD_BUTTON = DOM.download(() => {
const ndjson = DATASET.map((sample) => JSON.stringify(sample)).join('\n');
return new Blob([ndjson], { type: 'application/x-ndjson' });
}, "Text Splitting Dataset.ndjson");
CONTEXT_SIZE = 16
DATASET = {
const tokenizer = GPT2Tokenizer;
const [nl] = tokenizer.encode('\n');
const tokens = tokenizer.encode(TEXT);
const ret = [];
for (let i=0, n=tokens.length; i<n+CONTEXT_SIZE+1; ++i) {
const context = tokens.slice(i, i+CONTEXT_SIZE);
const input = tokenizer.decode(context);
const output = tokens[i+CONTEXT_SIZE] === nl;
ret.push({ input, output });
}
return ret;
}
GPT3Tokenizer = {
const library = await import("https://cdn.skypack.dev/gpt3-tokenizer@1.1.5/dist-browser/gpt3-tokenizer.js");
return library.default;
}
GPT2Tokenizer = await import("https://deno.land/x/gpt_2_3_tokenizer@v0.0.2/mod.js")
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
