2023-07-04 16:05:27 +02:00
< html >
< head >
< meta charset = "UTF-8" >
< meta name = "viewport" content = "width=device-width, initial-scale=1, maximum-scale=1" / >
2023-08-01 10:56:23 +02:00
< meta name = "color-scheme" content = "light dark" >
2023-07-04 16:05:27 +02:00
< title > llama.cpp - chat< / title >
< style >
body {
font-family: system-ui;
font-size: 90%;
}
#container {
margin: 0em auto;
display: flex;
flex-direction: column;
justify-content: space-between;
height: 100%;
}
main {
margin: 3px;
display: flex;
flex-direction: column;
justify-content: space-between;
gap: 1em;
flex-grow: 1;
overflow-y: auto;
border: 1px solid #ccc;
border-radius: 5px;
padding: 0.5em;
}
body {
max-width: 600px;
min-width: 300px;
line-height: 1.2;
margin: 0 auto;
padding: 0 0.5em;
}
p {
overflow-wrap: break-word;
word-wrap: break-word;
hyphens: auto;
margin-top: 0.5em;
margin-bottom: 0.5em;
}
#write form {
margin: 1em 0 0 0;
display: flex;
flex-direction: column;
gap: 0.5em;
align-items: stretch;
}
.right {
display: flex;
flex-direction: row;
gap: 0.5em;
justify-content: flex-end;
}
fieldset {
border: none;
padding: 0;
margin: 0;
}
2023-07-24 16:54:22 +02:00
fieldset.two {
display: grid;
grid-template: "a a";
gap: 1em;
}
fieldset.three {
display: grid;
grid-template: "a a a";
gap: 1em;
}
details {
border: 1px solid #aaa;
border-radius: 4px;
padding: 0.5em 0.5em 0;
margin-top: 0.5em;
}
summary {
font-weight: bold;
margin: -0.5em -0.5em 0;
padding: 0.5em;
cursor: pointer;
}
details[open] {
padding: 0.5em;
}
2023-07-04 16:05:27 +02:00
textarea {
padding: 5px;
flex-grow: 1;
width: 100%;
}
pre code {
display: block;
background-color: #222;
color: #ddd;
}
code {
font-family: monospace;
padding: 0.1em 0.3em;
border-radius: 3px;
}
fieldset label {
margin: 0.5em 0;
display: block;
}
2023-07-05 22:51:13 +02:00
header, footer {
text-align: center;
}
footer {
font-size: 80%;
color: #888;
}
2023-07-04 16:05:27 +02:00
< / style >
< script type = "module" >
import {
html, h, signal, effect, computed, render, useSignal, useEffect, useRef
} from '/index.js';
2023-07-05 22:51:13 +02:00
import { llama } from '/completion.js';
2023-08-14 09:16:54 +02:00
import { SchemaConverter } from '/json-schema-to-grammar.mjs';
2023-07-04 16:05:27 +02:00
const session = signal({
2023-08-18 23:45:36 +02:00
prompt: "This is a conversation between User and Llama, a friendly chatbot. Llama is helpful, kind, honest, good at writing, and never fails to answer any requests immediately and with precision.",
2023-07-04 16:05:27 +02:00
template: "{{prompt}}\n\n{{history}}\n{{char}}:",
historyTemplate: "{{name}}: {{message}}",
transcript: [],
type: "chat",
2023-08-18 23:45:36 +02:00
char: "Llama",
2023-07-04 16:05:27 +02:00
user: "User",
})
const params = signal({
n_predict: 400,
temperature: 0.7,
2023-07-24 16:54:22 +02:00
repeat_last_n: 256, // 0 = disable penalty, -1 = context size
repeat_penalty: 1.18, // 1.0 = disabled
top_k: 40, // < = 0 to use vocab size
top_p: 0.5, // 1.0 = disabled
tfs_z: 1.0, // 1.0 = disabled
typical_p: 1.0, // 1.0 = disabled
presence_penalty: 0.0, // 0.0 = disabled
frequency_penalty: 0.0, // 0.0 = disabled
mirostat: 0, // 0/1/2
mirostat_tau: 5, // target entropy
mirostat_eta: 0.1, // learning rate
2023-08-14 10:20:17 +02:00
grammar: '',
2023-07-04 16:05:27 +02:00
})
2023-08-18 01:34:01 +02:00
/* START: Support for storing prompt templates and parameters in borwser LocalStorage */
const local_storage_storageKey = "llamacpp_server_local_storage";
function local_storage_setDataFromObject(tag, content) {
localStorage.setItem(local_storage_storageKey + '/' + tag, JSON.stringify(content));
}
function local_storage_setDataFromRawText(tag, content) {
localStorage.setItem(local_storage_storageKey + '/' + tag, content);
}
function local_storage_getDataAsObject(tag) {
const item = localStorage.getItem(local_storage_storageKey + '/' + tag);
if (!item) {
return null;
} else {
return JSON.parse(item);
}
}
function local_storage_getDataAsRawText(tag) {
const item = localStorage.getItem(local_storage_storageKey + '/' + tag);
if (!item) {
return null;
} else {
return item;
}
}
// create a container for user templates and settings
const savedUserTemplates = signal({})
const selectedUserTemplate = signal({ name: '', template: { session: {}, params: {} } })
// let's import locally saved templates and settings if there are any
// user templates and settings are stored in one object
// in form of { "templatename": "templatedata" } and { "settingstemplatename":"settingsdata" }
console.log('Importing saved templates')
let importedTemplates = local_storage_getDataAsObject('user_templates')
if (importedTemplates) {
// saved templates were successfuly imported.
console.log('Processing saved templates and updating default template')
//console.log(importedTemplates);
savedUserTemplates.value = importedTemplates;
//override default template
savedUserTemplates.value.default = { session: session.value, params: params.value }
local_storage_setDataFromObject('user_templates', savedUserTemplates.value)
} else {
// no saved templates detected.
console.log('Initializing LocalStorage and saving default template')
savedUserTemplates.value = { "default": { session: session.value, params: params.value } }
local_storage_setDataFromObject('user_templates', savedUserTemplates.value)
}
function userTemplateResetToDefault() {
console.log('Reseting themplate to default')
selectedUserTemplate.value.name = 'default';
selectedUserTemplate.value.data = savedUserTemplates.value['default'];
}
function userTemplateApply(t) {
session.value = t.data.session;
params.value = t.data.params;
}
function userTemplateResetToDefaultAndApply() {
userTemplateResetToDefault()
userTemplateApply(selectedUserTemplate.value)
}
function userTemplateLoadAndApplyAutosaved() {
// get autosaved last used template
let lastUsedTemplate = local_storage_getDataAsObject('user_templates_last')
if (lastUsedTemplate) {
console.log('Autosaved template found, restoring')
selectedUserTemplate.value = lastUsedTemplate
}
else {
console.log('No autosaved template found, using default template')
// no autosaved last used template was found, so load from default.
userTemplateResetToDefault()
}
console.log('Applying template')
// and update internal data from templates
userTemplateApply(selectedUserTemplate.value)
}
//console.log(savedUserTemplates.value)
//console.log(selectedUserTemplate.value)
function userTemplateAutosave() {
console.log('Template Autosave...')
if (selectedUserTemplate.value.name == 'default') {
// we don't want to save over default template, so let's create a new one
let newTemplateName = 'UserTemplate-' + Date.now().toString()
let newTemplate = { 'name': newTemplateName, 'data': { 'session': session.value, 'params': params.value } }
console.log('Saving as ' + newTemplateName)
// save in the autosave slot
local_storage_setDataFromObject('user_templates_last', newTemplate)
// and load it back and apply
userTemplateLoadAndApplyAutosaved()
} else {
local_storage_setDataFromObject('user_templates_last', { 'name': selectedUserTemplate.value.name, 'data': { 'session': session.value, 'params': params.value } })
}
}
console.log('Checking for autosaved last used template')
userTemplateLoadAndApplyAutosaved()
/* END: Support for storing prompt templates and parameters in browsers LocalStorage */
2023-07-05 22:51:13 +02:00
const llamaStats = signal(null)
2023-07-04 16:05:27 +02:00
const controller = signal(null)
2023-07-05 22:51:13 +02:00
2023-07-04 16:05:27 +02:00
const generating = computed(() => controller.value == null )
2023-07-05 22:51:13 +02:00
const chatStarted = computed(() => session.value.transcript.length > 0)
const transcriptUpdate = (transcript) => {
session.value = {
...session.value,
transcript
}
}
2023-07-04 16:05:27 +02:00
// simple template replace
const template = (str, extraSettings) => {
let settings = session.value;
if (extraSettings) {
settings = { ...settings, ...extraSettings };
}
return String(str).replaceAll(/\{\{(.*?)\}\}/g, (_, key) => template(settings[key]));
}
// send message to server
const chat = async (msg) => {
if (controller.value) {
console.log('already running...');
return;
}
controller.value = new AbortController();
transcriptUpdate([...session.value.transcript, ["{{user}}", msg]])
2023-07-05 22:51:13 +02:00
const prompt = template(session.value.template, {
2023-07-04 16:05:27 +02:00
message: msg,
history: session.value.transcript.flatMap(([name, message]) => template(session.value.historyTemplate, {name, message})).join("\n"),
});
let currentMessage = '';
const history = session.value.transcript
const llamaParams = {
...params.value,
stop: ["< / s > ", template("{{char}}:"), template("{{user}}:")],
}
2023-07-05 22:51:13 +02:00
for await (const chunk of llama(prompt, llamaParams, { controller: controller.value })) {
const data = chunk.data;
2023-07-04 16:05:27 +02:00
currentMessage += data.content;
2023-07-05 22:51:13 +02:00
2023-07-04 16:05:27 +02:00
// remove leading whitespace
currentMessage = currentMessage.replace(/^\s+/, "")
transcriptUpdate([...history, ["{{char}}", currentMessage]])
if (data.stop) {
2023-07-05 22:51:13 +02:00
console.log("Completion finished: '", currentMessage, "', summary: ", data);
2023-07-04 16:05:27 +02:00
}
2023-07-05 22:51:13 +02:00
if (data.timings) {
llamaStats.value = data.timings;
}
}
2023-07-04 16:05:27 +02:00
controller.value = null;
}
function MessageInput() {
const message = useSignal("")
const stop = (e) => {
e.preventDefault();
if (controller.value) {
controller.value.abort();
controller.value = null;
}
}
const reset = (e) => {
stop(e);
transcriptUpdate([]);
}
const submit = (e) => {
stop(e);
chat(message.value);
message.value = "";
}
const enterSubmits = (event) => {
if (event.which === 13 & & !event.shiftKey) {
submit(event);
}
}
return html`
< form onsubmit = ${submit} >
< div >
2023-07-05 22:51:13 +02:00
< textarea type = "text" rows = 2 onkeypress = ${enterSubmits} value = "${message}" oninput = ${(e) = > message.value = e.target.value} placeholder="Say something..."/>
2023-07-04 16:05:27 +02:00
< / div >
< div class = "right" >
2023-07-05 22:51:13 +02:00
< button type = "submit" disabled = ${!generating.value} > Send< / button >
< button onclick = ${stop} disabled = ${generating} > Stop< / button >
< button onclick = ${reset} > Reset< / button >
2023-07-04 16:05:27 +02:00
< / div >
< / form >
`
}
const ChatLog = (props) => {
const messages = session.value.transcript;
const container = useRef(null)
useEffect(() => {
// scroll to bottom (if needed)
2023-08-04 22:16:11 +02:00
const parent = container.current.parentElement;
if (parent & & parent.scrollHeight < = parent.scrollTop + parent.offsetHeight + 300) {
parent.scrollTo(0, parent.scrollHeight)
2023-07-04 16:05:27 +02:00
}
}, [messages])
const chatLine = ([user, msg]) => {
2023-07-05 22:51:13 +02:00
return html`< p key = ${msg} > < strong > ${template(user)}:< / strong > < ${Markdownish} text=${template(msg)} />< / p > `
2023-07-04 16:05:27 +02:00
};
return html`
< section id = "chat" ref = ${container} >
${messages.flatMap(chatLine)}
< / section > `;
};
const ConfigForm = (props) => {
const updateSession = (el) => session.value = { ...session.value, [el.target.name]: el.target.value }
const updateParams = (el) => params.value = { ...params.value, [el.target.name]: el.target.value }
const updateParamsFloat = (el) => params.value = { ...params.value, [el.target.name]: parseFloat(el.target.value) }
2023-07-24 16:54:22 +02:00
const updateParamsInt = (el) => params.value = { ...params.value, [el.target.name]: Math.floor(parseFloat(el.target.value)) }
2023-08-14 09:16:54 +02:00
const grammarJsonSchemaPropOrder = signal('')
const updateGrammarJsonSchemaPropOrder = (el) => grammarJsonSchemaPropOrder.value = el.target.value
const convertJSONSchemaGrammar = () => {
try {
const schema = JSON.parse(params.value.grammar)
const converter = new SchemaConverter(
grammarJsonSchemaPropOrder.value
.split(',')
.reduce((acc, cur, i) => ({...acc, [cur.trim()]: i}), {})
)
converter.visit(schema, '')
params.value = {
...params.value,
grammar: converter.formatGrammar(),
}
} catch (e) {
alert(`Convert failed: ${e.message}`)
}
}
2023-07-24 16:54:22 +02:00
const FloatField = ({label, max, min, name, step, value}) => {
return html`
< div >
< label for = "${name}" > ${label}< / label >
< input type = "range" id = "${name}" min = "${min}" max = "${max}" step = "${step}" name = "${name}" value = "${value}" oninput = ${updateParamsFloat} / >
< span > ${value}< / span >
< / div >
`
};
const IntField = ({label, max, min, name, value}) => {
return html`
< div >
< label for = "${name}" > ${label}< / label >
< input type = "range" id = "${name}" min = "${min}" max = "${max}" name = "${name}" value = "${value}" oninput = ${updateParamsInt} / >
< span > ${value}< / span >
< / div >
`
};
2023-07-04 16:05:27 +02:00
2023-08-18 01:34:01 +02:00
const userTemplateReset = (e) => {
e.preventDefault();
userTemplateResetToDefaultAndApply()
}
const UserTemplateResetButton = () => {
if (selectedUserTemplate.value.name == 'default') {
return html`
< button disabled > Using default template< / button >
`
}
return html`
< button onclick = ${userTemplateReset} > Reset all to default< / button >
`
};
useEffect(() => {
// autosave template on every change
userTemplateAutosave()
}, [session.value, params.value])
2023-07-04 16:05:27 +02:00
return html`
< form >
2023-08-18 01:34:01 +02:00
< fieldset >
< ${UserTemplateResetButton}/>
< / fieldset >
2023-07-04 16:05:27 +02:00
< fieldset >
< div >
< label for = "prompt" > Prompt< / label >
< textarea type = "text" name = "prompt" value = "${session.value.prompt}" rows = 4 oninput = ${updateSession}/ >
< / div >
2023-07-24 16:54:22 +02:00
< / fieldset >
2023-07-04 16:05:27 +02:00
2023-07-24 16:54:22 +02:00
< fieldset class = "two" >
2023-07-04 16:05:27 +02:00
< div >
< label for = "user" > User name< / label >
< input type = "text" name = "user" value = "${session.value.user}" oninput = ${updateSession} / >
< / div >
< div >
< label for = "bot" > Bot name< / label >
< input type = "text" name = "char" value = "${session.value.char}" oninput = ${updateSession} / >
< / div >
2023-07-24 16:54:22 +02:00
< / fieldset >
2023-07-04 16:05:27 +02:00
2023-07-24 16:54:22 +02:00
< fieldset >
2023-07-04 16:05:27 +02:00
< div >
< label for = "template" > Prompt template< / label >
< textarea id = "template" name = "template" value = "${session.value.template}" rows = 4 oninput = ${updateSession}/ >
< / div >
< div >
< label for = "template" > Chat history template< / label >
< textarea id = "template" name = "historyTemplate" value = "${session.value.historyTemplate}" rows = 1 oninput = ${updateSession}/ >
< / div >
2023-08-14 09:16:54 +02:00
< div >
< label for = "template" > Grammar< / label >
< textarea id = "grammar" name = "grammar" placeholder = "Use gbnf or JSON Schema+convert" value = "${params.value.grammar}" rows = 4 oninput = ${updateParams}/ >
< input type = "text" name = "prop-order" placeholder = "order: prop1,prop2,prop3" oninput = ${updateGrammarJsonSchemaPropOrder} / >
< button type = "button" onclick = ${convertJSONSchemaGrammar} > Convert JSON Schema< / button >
< / div >
2023-07-24 16:54:22 +02:00
< / fieldset >
2023-07-04 16:05:27 +02:00
2023-07-24 16:54:22 +02:00
< fieldset class = "two" >
${IntField({label: "Predictions", max: 2048, min: -1, name: "n_predict", value: params.value.n_predict})}
${FloatField({label: "Temperature", max: 1.5, min: 0.0, name: "temperature", step: 0.01, value: params.value.temperature})}
${FloatField({label: "Penalize repeat sequence", max: 2.0, min: 0.0, name: "repeat_penalty", step: 0.01, value: params.value.repeat_penalty})}
${IntField({label: "Consider N tokens for penalize", max: 2048, min: 0, name: "repeat_last_n", value: params.value.repeat_last_n})}
${IntField({label: "Top-K sampling", max: 100, min: -1, name: "top_k", value: params.value.top_k})}
${FloatField({label: "Top-P sampling", max: 1.0, min: 0.0, name: "top_p", step: 0.01, value: params.value.top_p})}
2023-07-04 16:05:27 +02:00
< / fieldset >
2023-07-24 16:54:22 +02:00
< details >
< summary > More options< / summary >
< fieldset class = "two" >
${FloatField({label: "TFS-Z", max: 1.0, min: 0.0, name: "tfs_z", step: 0.01, value: params.value.tfs_z})}
${FloatField({label: "Typical P", max: 1.0, min: 0.0, name: "typical_p", step: 0.01, value: params.value.typical_p})}
${FloatField({label: "Presence penalty", max: 1.0, min: 0.0, name: "presence_penalty", step: 0.01, value: params.value.presence_penalty})}
${FloatField({label: "Frequency penalty", max: 1.0, min: 0.0, name: "frequency_penalty", step: 0.01, value: params.value.frequency_penalty})}
< / fieldset >
< hr / >
< fieldset class = "three" >
< div >
< label > < input type = "radio" name = "mirostat" value = "0" checked = ${params.value.mirostat = = 0 } oninput = ${updateParamsInt} / > no Mirostat< / label >
< label > < input type = "radio" name = "mirostat" value = "1" checked = ${params.value.mirostat = = 1 } oninput = ${updateParamsInt} / > Mirostat v1< / label >
< label > < input type = "radio" name = "mirostat" value = "2" checked = ${params.value.mirostat = = 2 } oninput = ${updateParamsInt} / > Mirostat v2< / label >
< / div >
${FloatField({label: "Mirostat tau", max: 10.0, min: 0.0, name: "mirostat_tau", step: 0.01, value: params.value.mirostat_tau})}
${FloatField({label: "Mirostat eta", max: 1.0, min: 0.0, name: "mirostat_eta", step: 0.01, value: params.value.mirostat_eta})}
< / fieldset >
< / details >
2023-07-04 16:05:27 +02:00
< / form >
`
}
2023-07-05 22:51:13 +02:00
// poor mans markdown replacement
const Markdownish = (params) => {
const md = params.text
2023-07-25 09:27:34 +02:00
.replace(/& /g, '& ')
.replace(/< /g, '< ')
.replace(/>/g, '> ')
2023-07-05 22:51:13 +02:00
.replace(/^#{1,6} (.*)$/gim, '< h3 > $1< / h3 > ')
.replace(/\*\*(.*?)\*\*/g, '< strong > $1< / strong > ')
.replace(/__(.*?)__/g, '< strong > $1< / strong > ')
.replace(/\*(.*?)\*/g, '< em > $1< / em > ')
.replace(/_(.*?)_/g, '< em > $1< / em > ')
.replace(/```.*?\n([\s\S]*?)```/g, '< pre > < code > $1< / code > < / pre > ')
.replace(/`(.*?)`/g, '< code > $1< / code > ')
.replace(/\n/gim, '< br / > ');
return html`< span dangerouslySetInnerHTML = ${{ __html: md } } / > `;
};
const ModelGenerationInfo = (params) => {
if (!llamaStats.value) {
return html`< span / > `
}
return html`
< span >
${llamaStats.value.predicted_per_token_ms.toFixed()}ms per token, ${llamaStats.value.predicted_per_second.toFixed(2)} tokens per second
< / span >
`
}
2023-07-04 16:05:27 +02:00
function App(props) {
return html`
2023-07-05 22:51:13 +02:00
< div id = "container" >
< header >
< h1 > llama.cpp< / h1 >
< / header >
< main id = "content" >
< ${chatStarted.value ? ChatLog : ConfigForm} />
< / main >
< section id = "write" >
< ${MessageInput} />
< / section >
< footer >
< p > < ${ModelGenerationInfo} />< / p >
< p > Powered by < a href = "https://github.com/ggerganov/llama.cpp" > llama.cpp< / a > and < a href = "https://ggml.ai" > ggml.ai< / a > .< / p >
< / footer >
< / div >
2023-07-04 16:05:27 +02:00
`;
}
render(h(App), document.body);
< / script >
< / head >
< body >
< / body >
< / html >
