Brunobkr/llama.cpp_AlgMor24_github
ΩFFFΣLLIa • llama.cpp • AlgMor24 ██████╗ ███████╗███████╗███████╗██╗ ██╗ ██╗ █████╗ ██╔═══██╗██╔════╝██╔════╝██╔════╝██║ ██║ ██║██╔══██╗ ██║ ██║█████╗ █████╗ █████╗ ██║ ██║ ██║███████║ ██║ ██║██╔══╝ ██╔══╝ ██╔══╝ ██║ ██║ ██║██╔══██║ ╚██████╔╝██║ ██║ ███████╗███████╗███████╗██║██║ ██║ ╚═════╝ ╚═╝ ╚═╝ ╚══════╝╚══════╝╚══════╝╚═╝╚═╝ ╚═╝ High-Performance LLM / VLM Inference & Autonomous Agentic Ecosystem… See the full description on the dataset page: https://huggingface.co/datasets/Brunobkr/llama.cpp_AlgMor24_github.
03.1k
1import {supportedCodepoint} from "./unicodeScripts";2 3import type {Mode} from "./types";4import type {CharacterMetrics, CharacterMetricsTuple, FontMetrics} from "./types/fonts";5// This map contains a mapping from font name and character code to character6// metrics, including height, depth, italic correction, and skew (kern from the7// character to the corresponding \skewchar)8// This map is generated via `make metrics`. It should not be changed manually.9import metricMap from "./fontMetricsData";10 11/**12 * This file contains metrics regarding fonts and individual symbols. The sigma13 * and xi variables, as well as the metricMap map contain data extracted from14 * TeX, TeX font metrics, and the TTF files. These data are then exposed via the15 * `metrics` variable and the getCharacterMetrics function.16 */17 18// In TeX, there are actually three sets of dimensions, one for each of19// textstyle (size index 5 and higher: >=9pt), scriptstyle (size index 3 and 4:20// 7-8pt), and scriptscriptstyle (size index 1 and 2: 5-6pt). These are21// provided in the arrays below, in that order.22//23// The font metrics are stored in fonts cmsy10, cmsy7, and cmsy5 respectively.24// This was determined by running the following script:25//26// latex -interaction=nonstopmode \27// '\documentclass{article}\usepackage{amsmath}\begin{document}' \28// '$a$ \expandafter\show\the\textfont2' \29// '\expandafter\show\the\scriptfont2' \30// '\expandafter\show\the\scriptscriptfont2' \31// '\stop'32//33// The metrics themselves were retrieved using the following commands:34//35// tftopl cmsy1036// tftopl cmsy737// tftopl cmsy538//39// The output of each of these commands is quite lengthy. The only part we40// care about is the FONTDIMEN section. Each value is measured in EMs.41const sigmasAndXis: Record<string, [number, number, number]> = {42 slant: [0.250, 0.250, 0.250], // sigma143 space: [0.000, 0.000, 0.000], // sigma244 stretch: [0.000, 0.000, 0.000], // sigma345 shrink: [0.000, 0.000, 0.000], // sigma446 xHeight: [0.431, 0.431, 0.431], // sigma547 quad: [1.000, 1.171, 1.472], // sigma648 extraSpace: [0.000, 0.000, 0.000], // sigma749 num1: [0.677, 0.732, 0.925], // sigma850 num2: [0.394, 0.384, 0.387], // sigma951 num3: [0.444, 0.471, 0.504], // sigma1052 denom1: [0.686, 0.752, 1.025], // sigma1153 denom2: [0.345, 0.344, 0.532], // sigma1254 sup1: [0.413, 0.503, 0.504], // sigma1355 sup2: [0.363, 0.431, 0.404], // sigma1456 sup3: [0.289, 0.286, 0.294], // sigma1557 sub1: [0.150, 0.143, 0.200], // sigma1658 sub2: [0.247, 0.286, 0.400], // sigma1759 supDrop: [0.386, 0.353, 0.494], // sigma1860 subDrop: [0.050, 0.071, 0.100], // sigma1961 delim1: [2.390, 1.700, 1.980], // sigma2062 delim2: [1.010, 1.157, 1.420], // sigma2163 axisHeight: [0.250, 0.250, 0.250], // sigma2264 65 // These font metrics are extracted from TeX by using tftopl on cmex10.tfm;66 // they correspond to the font parameters of the extension fonts (family 3).67 // See the TeXbook, page 441. In AMSTeX, the extension fonts scale; to68 // match cmex7, we'd use cmex7.tfm values for script and scriptscript69 // values.70 defaultRuleThickness: [0.04, 0.049, 0.049], // xi8; cmex7: 0.04971 bigOpSpacing1: [0.111, 0.111, 0.111], // xi972 bigOpSpacing2: [0.166, 0.166, 0.166], // xi1073 bigOpSpacing3: [0.2, 0.2, 0.2], // xi1174 bigOpSpacing4: [0.6, 0.611, 0.611], // xi12; cmex7: 0.61175 bigOpSpacing5: [0.1, 0.143, 0.143], // xi13; cmex7: 0.14376 77 // The \sqrt rule width is taken from the height of the surd character.78 // Since we use the same font at all sizes, this thickness doesn't scale.79 sqrtRuleThickness: [0.04, 0.04, 0.04],80 81 // This value determines how large a pt is, for metrics which are defined82 // in terms of pts.83 // This value is also used in katex.scss; if you change it make sure the84 // values match.85 ptPerEm: [10.0, 10.0, 10.0],86 87 // The space between adjacent `|` columns in an array definition. From88 // `\showthe\doublerulesep` in LaTeX. Equals 2.0 / ptPerEm.89 doubleRuleSep: [0.2, 0.2, 0.2],90 91 // The width of separator lines in {array} environments. From92 // `\showthe\arrayrulewidth` in LaTeX. Equals 0.4 / ptPerEm.93 arrayRuleWidth: [0.04, 0.04, 0.04],94 95 // Two values from LaTeX source2e:96 fboxsep: [0.3, 0.3, 0.3], // 3 pt / ptPerEm97 fboxrule: [0.04, 0.04, 0.04], // 0.4 pt / ptPerEm98};99 100// These are very rough approximations. We default to Times New Roman which101// should have Latin-1 and Cyrillic characters, but may not depending on the102// operating system. The metrics do not account for extra height from the103// accents. In the case of Cyrillic characters which have both ascenders and104// descenders we prefer approximations with ascenders, primarily to prevent105// the fraction bar or root line from intersecting the glyph.106// TODO(kevinb) allow union of multiple glyph metrics for better accuracy.107const extraCharacterMap: Record<string, string> = {108 // Latin-1109 'Å': 'A',110 'Ð': 'D',111 'Þ': 'o',112 'å': 'a',113 'ð': 'd',114 'þ': 'o',115 116 // Cyrillic117 'А': 'A',118 'Б': 'B',119 'В': 'B',120 'Г': 'F',121 'Д': 'A',122 'Е': 'E',123 'Ж': 'K',124 'З': '3',125 'И': 'N',126 'Й': 'N',127 'К': 'K',128 'Л': 'N',129 'М': 'M',130 'Н': 'H',131 'О': 'O',132 'П': 'N',133 'Р': 'P',134 'С': 'C',135 'Т': 'T',136 'У': 'y',137 'Ф': 'O',138 'Х': 'X',139 'Ц': 'U',140 'Ч': 'h',141 'Ш': 'W',142 'Щ': 'W',143 'Ъ': 'B',144 'Ы': 'X',145 'Ь': 'B',146 'Э': '3',147 'Ю': 'X',148 'Я': 'R',149 'а': 'a',150 'б': 'b',151 'в': 'a',152 'г': 'r',153 'д': 'y',154 'е': 'e',155 'ж': 'm',156 'з': 'e',157 'и': 'n',158 'й': 'n',159 'к': 'n',160 'л': 'n',161 'м': 'm',162 'н': 'n',163 'о': 'o',164 'п': 'n',165 'р': 'p',166 'с': 'c',167 'т': 'o',168 'у': 'y',169 'ф': 'b',170 'х': 'x',171 'ц': 'n',172 'ч': 'n',173 'ш': 'w',174 'щ': 'w',175 'ъ': 'a',176 'ы': 'm',177 'ь': 'a',178 'э': 'e',179 'ю': 'm',180 'я': 'r',181};182 183/**184 * This function adds new font metrics to default metricMap185 * It can also override existing metrics186 */187export function setFontMetrics(188 fontName: string,189 metrics: {[key: string]: CharacterMetricsTuple}190) {191 metricMap[fontName] = metrics;192}193 194/**195 * This function is a convenience function for looking up information in the196 * metricMap table. It takes a character as a string, and a font.197 *198 * Note: the `width` property may be undefined if fontMetricsData.js wasn't199 * built using `Make extended_metrics`.200 */201export function getCharacterMetrics(202 character: string,203 font: string,204 mode: Mode,205): CharacterMetrics | null | undefined {206 if (!metricMap[font]) {207 throw new Error(`Font metrics not found for font: ${font}.`);208 }209 let ch = character.charCodeAt(0);210 let metrics = metricMap[font][ch];211 if (!metrics && character[0] in extraCharacterMap) {212 ch = extraCharacterMap[character[0]].charCodeAt(0);213 metrics = metricMap[font][ch];214 }215 216 if (!metrics && mode === 'text') {217 // We don't typically have font metrics for Asian scripts.218 // But since we support them in text mode, we need to return219 // some sort of metrics.220 // So if the character is in a script we support but we221 // don't have metrics for it, just use the metrics for222 // the Latin capital letter M. This is close enough because223 // we (currently) only care about the height of the glyph224 // not its width.225 if (supportedCodepoint(ch)) {226 metrics = metricMap[font][77]; // 77 is the charcode for 'M'227 }228 }229 230 if (metrics) {231 return {232 depth: metrics[0],233 height: metrics[1],234 italic: metrics[2],235 skew: metrics[3],236 width: metrics[4],237 };238 }239}240 241type FontSizeIndex = 0 | 1 | 2;242const fontMetricsBySizeIndex: Partial<Record<FontSizeIndex, FontMetrics>> = {};243 244/**245 * Get the font metrics for a given size.246 */247export function getGlobalMetrics(size: number): FontMetrics {248 let sizeIndex: FontSizeIndex;249 if (size >= 5) {250 sizeIndex = 0;251 } else if (size >= 3) {252 sizeIndex = 1;253 } else {254 sizeIndex = 2;255 }256 if (!fontMetricsBySizeIndex[sizeIndex]) {257 const metrics: FontMetrics = fontMetricsBySizeIndex[sizeIndex] = {258 cssEmPerMu: sigmasAndXis.quad[sizeIndex] / 18,259 };260 for (const key in sigmasAndXis) {261 if (sigmasAndXis.hasOwnProperty(key)) {262 metrics[key] = sigmasAndXis[key][sizeIndex];263 }264 }265 }266 return fontMetricsBySizeIndex[sizeIndex]!;267}268 