rixty
/
SillyTavern
mirror of https://github.com/SillyTavern/SillyTavern.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

anthropic captioning

This commit is contained in:
based 2024-03-05 07:07:38 +10:00
parent cbda133d75
commit 64d9c9dc5d
6 changed files with 83 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
public/index.html
View File

@ -1654,12 +1654,12 @@
</span>
</div>
</div>
<div class="range-block" data-source="openai,openrouter,makersuite,custom">
<div class="range-block" data-source="openai,openrouter,makersuite,claude,custom">
<label for="openai_image_inlining" class="checkbox_label flexWrap widthFreeExpand">
<input id="openai_image_inlining" type="checkbox" />
<span data-i18n="Send inline images">Send inline images</span>
<div id="image_inlining_hint" class="flexBasis100p toggle-description justifyLeft">
Sends images in prompts if the model supports it (e.g. GPT-4V or Llava 13B).
Sends images in prompts if the model supports it (e.g. GPT-4V, Claude 3 or Llava 13B).
Use the <code><i class="fa-solid fa-paperclip"></i></code> action on any message or the
<code><i class="fa-solid fa-wand-magic-sparkles"></i></code> menu to attach an image file to the chat.
</div>

6
public/scripts/extensions/caption/index.js
View File

@ -284,6 +284,7 @@ jQuery(function () {
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'openai' && (secret_state[SECRET_KEYS.OPENAI] || extension_settings.caption.allow_reverse_proxy)) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'openrouter' && secret_state[SECRET_KEYS.OPENROUTER]) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'google' && secret_state[SECRET_KEYS.MAKERSUITE]) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'anthropic' && secret_state[SECRET_KEYS.CLAUDE]) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'ollama' && textgenerationwebui_settings.server_urls[textgen_types.OLLAMA]) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'llamacpp' && textgenerationwebui_settings.server_urls[textgen_types.LLAMACPP]) ||
(extension_settings.caption.source === 'multimodal' && extension_settings.caption.multimodal_api === 'ooba' && textgenerationwebui_settings.server_urls[textgen_types.OOBA]) ||
@ -343,7 +344,7 @@ jQuery(function () {
<label for="caption_source">Source</label>
<select id="caption_source" class="text_pole">
<option value="local">Local</option>
<option value="multimodal">Multimodal (OpenAI / llama / Google)</option>
<option value="multimodal">Multimodal (OpenAI / Anthropic / llama / Google)</option>
<option value="extras">Extras</option>
<option value="horde">Horde</option>
</select>
@ -355,6 +356,7 @@ jQuery(function () {
<option value="ooba">Text Generation WebUI (oobabooga)</option>
<option value="ollama">Ollama</option>
<option value="openai">OpenAI</option>
<option value="anthropic">Anthropic</option>
<option value="openrouter">OpenRouter</option>
<option value="google">Google MakerSuite</option>
<option value="custom">Custom (OpenAI-compatible)</option>
@ -364,6 +366,8 @@ jQuery(function () {
<label for="caption_multimodal_model">Model</label>
<select id="caption_multimodal_model" class="flex1 text_pole">
<option data-type="openai" value="gpt-4-vision-preview">gpt-4-vision-preview</option>
<option data-type="anthropic" value="claude-3-opus-20240229">claude-3-opus-20240229</option>
<option data-type="anthropic" value="claude-3-sonnet-20240229">claude-3-sonnet-20240229</option>
<option data-type="google" value="gemini-pro-vision">gemini-pro-vision</option>
<option data-type="openrouter" value="openai/gpt-4-vision-preview">openai/gpt-4-vision-preview</option>
<option data-type="openrouter" value="haotian-liu/llava-13b">haotian-liu/llava-13b</option>

5
public/scripts/extensions/shared.js
View File

@ -23,6 +23,7 @@ export async function getMultimodalCaption(base64Img, prompt) {
// OpenRouter has a payload limit of ~2MB. Google is 4MB, but we love democracy.
// Ooba requires all images to be JPEGs.
const isGoogle = extension_settings.caption.multimodal_api === 'google';
const isClaude = extension_settings.caption.multimodal_api === 'anthropic';
const isOllama = extension_settings.caption.multimodal_api === 'ollama';
const isLlamaCpp = extension_settings.caption.multimodal_api === 'llamacpp';
const isCustom = extension_settings.caption.multimodal_api === 'custom';
@ -39,7 +40,7 @@ export async function getMultimodalCaption(base64Img, prompt) {
}
const useReverseProxy =
extension_settings.caption.multimodal_api === 'openai'
(extension_settings.caption.multimodal_api === 'openai' || extension_settings.caption.multimodal_api === 'anthropic')
&& extension_settings.caption.allow_reverse_proxy
&& oai_settings.reverse_proxy
&& isValidUrl(oai_settings.reverse_proxy);
@ -87,6 +88,8 @@ export async function getMultimodalCaption(base64Img, prompt) {
switch (extension_settings.caption.multimodal_api) {
case 'google':
return '/api/google/caption-image';
case 'anthropic':
return '/api/anthropic/caption-image';
case 'llamacpp':
return '/api/backends/text-completions/llamacpp/caption-image';
case 'ollama':

3
public/scripts/openai.js
View File

@ -3836,6 +3836,7 @@ export function isImageInliningSupported() {
const gpt4v = 'gpt-4-vision';
const geminiProV = 'gemini-pro-vision';
const claude = 'claude-3';
const llava = 'llava';
if (!oai_settings.image_inlining) {
@ -3847,6 +3848,8 @@ export function isImageInliningSupported() {
return oai_settings.openai_model.includes(gpt4v);
case chat_completion_sources.MAKERSUITE:
return oai_settings.google_model.includes(geminiProV);
case chat_completion_sources.CLAUDE:
return oai_settings.claude_model.includes(claude);
case chat_completion_sources.OPENROUTER:
return !oai_settings.openrouter_force_instruct && (oai_settings.openrouter_model.includes(gpt4v) || oai_settings.openrouter_model.includes(llava));
case chat_completion_sources.CUSTOM:

3
server.js
View File

@ -505,6 +505,9 @@ app.use('/api/openai', require('./src/endpoints/openai').router);
//Google API
app.use('/api/google', require('./src/endpoints/google').router);
//Anthropic API
app.use('/api/anthropic', require('./src/endpoints/anthropic').router);
// Tokenizers
app.use('/api/tokenizers', require('./src/endpoints/tokenizers').router);

66
src/endpoints/anthropic.js Normal file
View File

@ -0,0 +1,66 @@
const { readSecret, SECRET_KEYS } = require('./secrets');
const fetch = require('node-fetch').default;
const express = require('express');
const { jsonParser } = require('../express-common');
const router = express.Router();
router.post('/caption-image', jsonParser, async (request, response) => {
try {
const mimeType = request.body.image.split(';')[0].split(':')[1];
const base64Data = request.body.image.split(',')[1];
const url = 'https://api.anthropic.com/v1/messages';
const body = {
model: request.body.model,
messages: [
{
'role': 'user', 'content': [
{
'type': 'image',
'source': {
'type': 'base64',
'media_type': mimeType,
'data': base64Data,
},
},
{ 'type': 'text', 'text': request.body.prompt },
],
},
],
max_tokens: 800,
};
console.log('Multimodal captioning request', body);
const result = await fetch(url, {
body: JSON.stringify(body),
method: 'POST',
headers: {
'Content-Type': 'application/json',
'anthropic-version': '2023-06-01',
'x-api-key': readSecret(SECRET_KEYS.CLAUDE),
},
timeout: 0,
});
if (!result.ok) {
console.log(color.red(`Claude API returned error: ${result.status} ${result.statusText}`));
return response.status(result.status).send({ error: true });
}
const generateResponseJson = await result.json();
const caption = generateResponseJson.content[0].text;
console.log('Claude response:', generateResponseJson);
if (!caption) {
return response.status(500).send('No caption found');
}
return response.json({ caption });
} catch (error) {
console.error(error);
response.status(500).send('Internal server error');
}
});
module.exports = { router };