Merge branch preview into preview-fork

Note: list below was generated using Claude Opus from the more verbose original generated squashed commit message, so some items are missing and some may be hallucinated. A quick eyeball check indicated it's very close to what a human would create manually 🐛 fix merge mistake Merge branch 'main' into preview 🐛 fix language undefined bug 🚧 update redirect Revert "📝 remove redundant docs redirect" 📝 remove redundant docs redirect Merge branch 'main' of https://github.com/continuedev/continue 📝 restore /docs base url 🍱 update continue logo Update README image and description (continuedev#1169) Merge branch 'main' of https://github.com/continuedev/continue 📝 update docs base path 🔥 remove disable indexing 💄 add llama3 to UI 📝 remove note about parallel ollama models fix type for gemma of groq (continuedev#1142) 📝 update embeddings docs 📝 update custom commands docs 📝 fix shortcut in docs 📝 gemini 1.5 pro in changelog 📝 update changelog Preview (continuedev#1114) - ✨ shared indexing - 🎨 indexing - 🧑‍💻 npm i --no-save in prepackage.js - 🚚 rename addLogs to addPromptCompletionPair - 🩹 add filter for midline imports/top-level keywords and encoding header - 🩹 add .t. to stop words - 🔥 Improved Ctrl/Cmd+I (continuedev#1023) - ⚡️ improved diff streaming algo - 🎨 better messaging/formatting for further cmd/ctrl+I instructions - ⚡️ more reliably filter out unwanted explanations - 🚸 better follow up edits - 💄 accept/reject diffs block-by-block - ✨ cmd/ctrl+z to reject diff - 🚚 rename variables - 💄 allow switching files when inline diff still visible - 🚸 don't show quick pick if not ctx providers exist - 🚧 (sort of) allow switching editors while streaming diff - 💄 show model being used for cmd/ctrl+I - 💄 don't add undo stops when generating diff - 🐛 fix shortcuts for accept/reject diff blocks - ✨ improved GPT edit prompt, taking prefix/suffix into account - ✨ improved prompting for empty selection ctrl/cmd+I - ⚡️ immediately refresh codelens - 🐛 use first model if default undefined - ⚡️ refresh codelens after diff cleared - 💄 update keyboard shortcuts - ⚡️ Improved edit prompts for OS models (continuedev#1029) - 💄 refresh codelens more frequently - ⚡️ improved codellama edit prompt - ⚡️ better codellama prompt - ⚡️ use same improved prompt for most OS models - 🎨 refactor chat templates - 🎨 refactor llama2 prompt to allow ending assistant message - ⚡️ separate os models prompt when no prefix/suffix - 🎨 refactor to allow putting words in the model's mouth - ⚡️ prune code around cmd/ctrl+I - 🚚 rename to cmd/ctrl+I - 🎨 make raw a base completion option - 🩹 small improvements - 🩹 use different prompt when completions not supported - Keep the same statusBar item when updating it to prevent flickering of the status bar. (continuedev#1022) - 🎨 add getRepoName to IDE, use for indexing - 🎨 implement server client interface - 📌 pin to vectordb=0.4.12 - 🧑‍💻 mark xhr-sync-worker.js as external in esbuild - 🎨 break out ignore defaults into core - 🎨 update getRepoName - 🐛 fix import error - 🩹 fix chat.jsonl logging - ⚡️ improved OpenAI autocomplete support - 🐛 fix bug causing part of completions to be skipped - 🔥 remove URLContextProvider - ✨ Add Groq as an official provider - 🩹 make sure autocomplete works with claude - 💄 update positioning of code block toolbar to not cover code - ✨ Run in terminal button - ✨ insert at cursor button - ✨ Regenerate and copy buttons - ✨ Button to force re-indexing - 🐛 make sure tooltip IDs are unique - ✨ Button to continue truncated response - 🚧 WIP on inline edit browser embedding - 🚧 inline TipTapEditor - 🚧 WIP on inline TipTapEditor - 🔥 remove unused test component - 🚧 native inline edit - 💄 nicer looking input box - ✨ Diff Streaming in JetBrains - 💄 line highlighting - 💄 arial font - ✨ Retry with further instructions - 🚧 drop shadow - ✨ accept/reject diffs - ✨ accept/reject diffs - 🐛 fix off-by-one errors - 🚧 swap out button on enter - 💄 styling and auto-resize - 💄 box shadow - 🚧 fix keyboard shortcuts to accept/reject diff - 💄 improve small interactions - 💄 loading icon, cancellation logic - 🐛 handle next.value being undefined - ✨ latex support - Bug Fix: Add ternary operator to prevent nonexistant value error (continuedev#1052) - Update completionProvider.ts Add \r\n\r\n stop to tab completion - 📌 update package-locks - 🐛 fix bug in edit prompt - 🔊 print when SSL verification disabled - 📌 pin esbuild version to match our hosted binary - 🔥 remove unused package folder - 👷 add note about pinning esbuild - 🚚 rename pkg to binary - ⚡️ update an important stop word for starcoder2, improve dev data - 🐛 fix autocomplete bug - Update completionProvider.ts as @rootedbox suggested - ⏪ revert back to esbuild ^0.17.19 to solve no backend found error with onnxruntime - 🩹 set default autocomplete temp to 0.01 to be strictly positive - make the useCopyBuffer option effective (continuedev#1062) - Con-1037: Toggle full screen bug (continuedev#1065) - Resolve conflict, accept branch being merged in (continuedev#1076) - continuedev#1073: update outdated documentation (continuedev#1074) - 🩹 small tweaks to stop words - Add abstraction for fetch to easily allow using request options (continuedev#1059) - Add a new slash command to review code. (continuedev#1071) - 🩹 add new starcoder artifact as stopword - 💄 slight improvements to inline edit UI - 🔖 update default models, bump gradle version - 📝 recommend starcoder2 - 🐛 fix jetbrains encoding issue - 🩹 don't index site-packages - 🩹 error handling in JetBrains - 🐛 fix copy to clipboard in jetbrains - fix: cursor focus issue causing unwanted return to text area (continuedev#1086) - 📝 mention autocomplete in jetbrains - 📝 Tab-autocomplete README - 🔥 remove note about custom ctx providers only being on VS Code - 📝 docs about http context provider - 👥 pull request template - Update from Claude 2 to Claude 3 (continuedev#1078) - 📝 add FAQ about single-line completions - 📝 update autocomplete docs - fix cursor focus issue causing unwanted return to text area - 🔧 option to disable autocomplete from config.json - ✨ option to disable streaming with anthropic - ✅ Test to verify that files are packaged - Add FIM template for CodeGemma (continuedev#1097) Also pass stop tokens to llama.cpp. - ✨ customizable rerankers (continuedev#1088)
pzaback · Apr 23, 2024 · 5c5b168 · 5c5b168
1 parent 93fbc10
commit 5c5b168
Show file tree

Hide file tree

Showing 31 changed files with 314 additions and 146 deletions.
diff --git a/.changes/extensions/intellij/0.0.42.md b/.changes/extensions/intellij/0.0.42.md
@@ -0,0 +1,7 @@
+## 0.0.42 - 2024-04-12
+### Added
+* Inline cmd/ctrl+I in JetBrains
+### Fixed
+* Fixed character encoding error causing display issues
+* Fixed error causing input to constantly demand focus
+* Fixed automatic reloading of config.json
diff --git a/.changes/extensions/vscode/0.8.24.md b/.changes/extensions/vscode/0.8.24.md
@@ -0,0 +1,5 @@
+## 0.8.24 - 2024-04-12
+### Added
+* Support for improved retrieval models (Voyage embeddings/reranking)
+* New @code context provider
+* Personal usage analytics
diff --git a/.changes/unreleased/Added-20240412-160513.yaml b/.changes/unreleased/Added-20240412-160513.yaml
@@ -0,0 +1,4 @@
+project: extensions/vscode
+kind: Added
+body: Support for Gemini 1.5 Pro
+time: 2024-04-12T16:05:13.251485-07:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,7 +1,6 @@
 # Changelog
 
-Separate changelogs are kept for each part of the Continue repository:
+Separate changelogs are kept for each extension:
 
 - [VS Code Extension](./extensions/vscode/CHANGELOG.md)
 - [Intellij Extension](./extensions/intellij/CHANGELOG.md)
-- [Continue Server](./server/CHANGELOG.md)
diff --git a/README.md b/README.md
@@ -1,10 +1,14 @@
-![Continue logo](media/c_d.png)
+<div align="center">
+
+![Continue logo](media/readme.png)
+
+</div>
 
 <h1 align="center">Continue</h1>
 
 <div align="center">
 
-**[Continue](https://continue.dev/docs) is an open-source autopilot for [VS Code](https://marketplace.visualstudio.com/items?itemName=Continue.continue) and [JetBrains](https://plugins.jetbrains.com/plugin/22707-continue-extension)—the easiest way to code with any LLM**
+**[Continue](https://continue.dev/docs) keeps developers in flow. Our open-source [VS Code](https://marketplace.visualstudio.com/items?itemName=Continue.continue) and [JetBrains](https://plugins.jetbrains.com/plugin/22707-continue-extension) extensions enable you to easily create your own modular AI software development system that you can improve.**
 
 </div>
 

diff --git a/core/index.d.ts b/core/index.d.ts
@@ -523,16 +523,21 @@ export type ModelName =
   | "gpt-4-turbo"
   | "gpt-4-turbo-preview"
   | "gpt-4-vision-preview"
-  // Open Source
+  // Mistral
   | "mistral-7b"
   | "mistral-8x7b"
+  // Llama 2
   | "llama2-7b"
   | "llama2-13b"
   | "llama2-70b"
   | "codellama-7b"
   | "codellama-13b"
   | "codellama-34b"
   | "codellama-70b"
+  // Llama 3
+  | "llama3-8b"
+  | "llama3-70b"
+  // Other Open-source
   | "phi2"
   | "phind-codellama-34b"
   | "wizardcoder-7b"

diff --git a/core/llm/llms/Groq.ts b/core/llm/llms/Groq.ts
@@ -11,6 +11,8 @@ class Groq extends OpenAI {
     "llama2-70b": "llama2-70b-4096",
     "mistral-8x7b": "mixtral-8x7b-32768",
     gemma: "gemma-7b-it",
+    "llama3-8b": "llama3-8b-8192",
+    "llama3-70b": "llama3-70b-8192",
   };
   protected _convertModelName(model: string): string {
     return Groq.modelConversion[model] ?? model;

diff --git a/core/llm/llms/Ollama.ts b/core/llm/llms/Ollama.ts
@@ -76,6 +76,8 @@ class Ollama extends BaseLLM {
         "codellama-13b": "codellama:13b",
         "codellama-34b": "codellama:34b",
         "codellama-70b": "codellama:70b",
+        "llama3-8b": "llama3:8b",
+        "llama3-70b": "llama3:70b",
         "phi-2": "phi:2.7b",
         "phind-codellama-34b": "phind-codellama:34b-v2",
         "wizardcoder-7b": "wizardcoder:7b-python",

diff --git a/core/llm/llms/Replicate.ts b/core/llm/llms/Replicate.ts
@@ -16,6 +16,8 @@ class Replicate extends BaseLLM {
       "meta/codellama-70b-instruct:a279116fe47a0f65701a8817188601e2fe8f4b9e04a518789655ea7b995851bf",
     "llama2-7b": "meta/llama-2-7b-chat" as any,
     "llama2-13b": "meta/llama-2-13b-chat" as any,
+    "llama3-8b": "meta/meta-llama-3-8b-instruct" as any,
+    "llama3-70b": "meta/meta-llama-3-70b-instruct" as any,
     "zephyr-7b":
       "nateraw/zephyr-7b-beta:b79f33de5c6c4e34087d44eaea4a9d98ce5d3f3a09522f7328eea0685003a931",
     "mistral-7b":

diff --git a/core/llm/llms/Together.ts b/core/llm/llms/Together.ts
@@ -18,6 +18,8 @@ class Together extends OpenAI {
     "codellama-13b": "togethercomputer/CodeLlama-13b-Instruct",
     "codellama-34b": "togethercomputer/CodeLlama-34b-Instruct",
     "codellama-70b": "codellama/CodeLlama-70b-Instruct-hf",
+    "llama3-8b": "meta-llama/Llama-3-8b-chat-hf",
+    "llama3-70b": "meta-llama/Llama-3-70b-chat-hf",
     "llama2-7b": "togethercomputer/llama-2-7b-chat",
     "llama2-13b": "togethercomputer/llama-2-13b-chat",
     "llama2-70b": "togethercomputer/llama-2-70b-chat",

diff --git a/core/llm/llms/index.ts b/core/llm/llms/index.ts
@@ -1,15 +1,15 @@
 import Handlebars from "handlebars";
 import { BaseLLM } from "..";
 import {
-	BaseCompletionOptions,
-	ILLM,
-	LLMOptions,
-	ModelDescription,
+  BaseCompletionOptions,
+  ILLM,
+  LLMOptions,
+  ModelDescription,
 } from "../..";
 import { DEFAULT_MAX_TOKENS } from "../constants";
 import Anthropic from "./Anthropic";
-import Cohere from "./Cohere";
 import Bedrock from "./Bedrock";
+import Cohere from "./Cohere";
 import DeepInfra from "./DeepInfra";
 import Flowise from "./Flowise";
 import FreeTrial from "./FreeTrial";
@@ -29,122 +29,122 @@ import TextGenWebUI from "./TextGenWebUI";
 import Together from "./Together";
 
 function convertToLetter(num: number): string {
-	let result = "";
-	while (num > 0) {
-		const remainder = (num - 1) % 26;
-		result = String.fromCharCode(97 + remainder) + result;
-		num = Math.floor((num - 1) / 26);
-	}
-	return result;
+  let result = "";
+  while (num > 0) {
+    const remainder = (num - 1) % 26;
+    result = String.fromCharCode(97 + remainder) + result;
+    num = Math.floor((num - 1) / 26);
+  }
+  return result;
 }
 
 const getHandlebarsVars = (
-	value: string,
+  value: string,
 ): [string, { [key: string]: string }] => {
-	const ast = Handlebars.parse(value);
+  const ast = Handlebars.parse(value);
 
-	let keysToFilepath: { [key: string]: string } = {};
-	let keyIndex = 1;
-	for (let i in ast.body) {
-		if (ast.body[i].type === "MustacheStatement") {
-			const letter = convertToLetter(keyIndex);
-			keysToFilepath[letter] = (ast.body[i] as any).path.original;
-			value = value.replace(
-				new RegExp("{{\\s*" + (ast.body[i] as any).path.original + "\\s*}}"),
-				`{{${letter}}}`,
-			);
-			keyIndex++;
-		}
-	}
-	return [value, keysToFilepath];
+  let keysToFilepath: { [key: string]: string } = {};
+  let keyIndex = 1;
+  for (let i in ast.body) {
+    if (ast.body[i].type === "MustacheStatement") {
+      const letter = convertToLetter(keyIndex);
+      keysToFilepath[letter] = (ast.body[i] as any).path.original;
+      value = value.replace(
+        new RegExp("{{\\s*" + (ast.body[i] as any).path.original + "\\s*}}"),
+        `{{${letter}}}`,
+      );
+      keyIndex++;
+    }
+  }
+  return [value, keysToFilepath];
 };
 
 export async function renderTemplatedString(
-	template: string,
-	readFile: (filepath: string) => Promise<string>,
-	inputData: any,
+  template: string,
+  readFile: (filepath: string) => Promise<string>,
+  inputData: any,
 ): Promise<string> {
-	const [newTemplate, vars] = getHandlebarsVars(template);
-	const data: any = { ...inputData };
-	for (const key in vars) {
-		const fileContents = await readFile(vars[key]);
-		data[key] = fileContents || (inputData[vars[key]] ?? vars[key]);
-	}
-	const templateFn = Handlebars.compile(newTemplate);
-	const final = templateFn(data);
-	return final;
+  const [newTemplate, vars] = getHandlebarsVars(template);
+  const data: any = { ...inputData };
+  for (const key in vars) {
+    const fileContents = await readFile(vars[key]);
+    data[key] = fileContents || (inputData[vars[key]] ?? vars[key]);
+  }
+  const templateFn = Handlebars.compile(newTemplate);
+  const final = templateFn(data);
+  return final;
 }
 
 const LLMs = [
-	Anthropic,
-	Cohere,
-	FreeTrial,
-	Gemini,
-	Llamafile,
-	Ollama,
-	Replicate,
-	TextGenWebUI,
-	Together,
-	HuggingFaceTGI,
-	HuggingFaceInferenceAPI,
-	LlamaCpp,
-	OpenAI,
-	LMStudio,
-	Mistral,
-	Bedrock,
-	DeepInfra,
-	OpenAIFreeTrial,
-	Flowise,
-	Groq,
+  Anthropic,
+  Cohere,
+  FreeTrial,
+  Gemini,
+  Llamafile,
+  Ollama,
+  Replicate,
+  TextGenWebUI,
+  Together,
+  HuggingFaceTGI,
+  HuggingFaceInferenceAPI,
+  LlamaCpp,
+  OpenAI,
+  LMStudio,
+  Mistral,
+  Bedrock,
+  DeepInfra,
+  OpenAIFreeTrial,
+  Flowise,
+  Groq,
 ];
 
 export async function llmFromDescription(
-	desc: ModelDescription,
-	readFile: (filepath: string) => Promise<string>,
-	completionOptions?: BaseCompletionOptions,
-	systemMessage?: string,
+  desc: ModelDescription,
+  readFile: (filepath: string) => Promise<string>,
+  completionOptions?: BaseCompletionOptions,
+  systemMessage?: string,
 ): Promise<BaseLLM | undefined> {
-	const cls = LLMs.find((llm) => llm.providerName === desc.provider);
+  const cls = LLMs.find((llm) => llm.providerName === desc.provider);
 
-	if (!cls) {
-		return undefined;
-	}
+  if (!cls) {
+    return undefined;
+  }
 
-	const finalCompletionOptions = {
-		...completionOptions,
-		...desc.completionOptions,
-	};
+  const finalCompletionOptions = {
+    ...completionOptions,
+    ...desc.completionOptions,
+  };
 
-	systemMessage = desc.systemMessage ?? systemMessage;
-	if (systemMessage !== undefined) {
-		systemMessage = await renderTemplatedString(systemMessage, readFile, {});
-	}
+  systemMessage = desc.systemMessage ?? systemMessage;
+  if (systemMessage !== undefined) {
+    systemMessage = await renderTemplatedString(systemMessage, readFile, {});
+  }
 
-	const options: LLMOptions = {
-		...desc,
-		completionOptions: {
-			...finalCompletionOptions,
-			model: (desc.model || cls.defaultOptions?.model) ?? "codellama-7b",
-			maxTokens:
-				finalCompletionOptions.maxTokens ??
-				cls.defaultOptions?.completionOptions?.maxTokens ??
-				DEFAULT_MAX_TOKENS,
-		},
-		systemMessage,
-	};
+  const options: LLMOptions = {
+    ...desc,
+    completionOptions: {
+      ...finalCompletionOptions,
+      model: (desc.model || cls.defaultOptions?.model) ?? "codellama-7b",
+      maxTokens:
+        finalCompletionOptions.maxTokens ??
+        cls.defaultOptions?.completionOptions?.maxTokens ??
+        DEFAULT_MAX_TOKENS,
+    },
+    systemMessage,
+  };
 
-	return new cls(options);
+  return new cls(options);
 }
 
 export function llmFromProviderAndOptions(
-	providerName: string,
-	llmOptions: LLMOptions,
+  providerName: string,
+  llmOptions: LLMOptions,
 ): ILLM {
-	const cls = LLMs.find((llm) => llm.providerName === providerName);
+  const cls = LLMs.find((llm) => llm.providerName === providerName);
 
-	if (!cls) {
-		throw new Error(`Unknown LLM provider type "${providerName}"`);
-	}
+  if (!cls) {
+    throw new Error(`Unknown LLM provider type "${providerName}"`);
+  }
 
-	return new cls(llmOptions);
+  return new cls(llmOptions);
 }
diff --git a/docs/docs/customization/slash-commands.md b/docs/docs/customization/slash-commands.md
@@ -16,7 +16,7 @@ To use any of the built-in slash commands, open `~/.continue/config.json` and ad
 
 ### `/edit`
 
-Select code with ctrl/cmd + M (VS Code) or ctrl/cmd + J (JetBrains), and then type "/edit", followed by instructions for the edit. Continue will stream the changes into a side-by-side diff editor.
+Select code with ctrl/cmd + L (VS Code) or ctrl/cmd + J (JetBrains), and then type "/edit", followed by instructions for the edit. Continue will stream the changes into a side-by-side diff editor.
 
 ```json
 {
@@ -118,18 +118,25 @@ You can add custom slash commands by adding to the `customCommands` property in
 
 - `name`: the name of the command, which will be invoked with `/name`
 - `description`: a short description of the command, which will appear in the dropdown
-- `prompt`: a set of instructions to the LLM, which will be shown in the prompt
+- `prompt`: a templated prompt to send to the LLM
 
 Custom commands are great when you are frequently reusing a prompt. For example, if you've crafted a great prompt and frequently ask the LLM to check for mistakes in your code, you could add a command like this:
 
 ```json title="~/.continue/config.json"
 customCommands=[{
         "name": "check",
         "description": "Check for mistakes in my code",
-        "prompt": "Please read the highlighted code and check for any mistakes. You should look for the following, and be extremely vigilant:\n- Syntax errors\n- Logic errors\n- Security vulnerabilities\n- Performance issues\n- Anything else that looks wrong\n\nOnce you find an error, please explain it as clearly as possible, but without using extra words. For example, instead of saying 'I think there is a syntax error on line 5', you should say 'Syntax error on line 5'. Give your answer as one bullet point per mistake found."
+        "prompt": "{{{ input }}}\n\nPlease read the highlighted code and check for any mistakes. You should look for the following, and be extremely vigilant:\n- Syntax errors\n- Logic errors\n- Security vulnerabilities\n- Performance issues\n- Anything else that looks wrong\n\nOnce you find an error, please explain it as clearly as possible, but without using extra words. For example, instead of saying 'I think there is a syntax error on line 5', you should say 'Syntax error on line 5'. Give your answer as one bullet point per mistake found."
 }]
 ```
 
+#### Templating
+
+The `prompt` property supports templating with Handlebars syntax. You can use the following variables:
+
+- `input` (used in the example above): any additional input entered with the slash command. For example, if you type `/test only write one test`, `input` will be `only write one test`. This will also include highlighted code blocks.
+- File names: You can reference any file by providing an absolute path or a path relative to the current working directory.
+
 ### Custom Slash Commands
 
 If you want to go a step further than writing custom commands with natural language, you can write a custom function that returns the response. This requires using `config.ts` instead of `config.json`.
@@ -147,7 +154,7 @@ export function modifyConfig(config: Config): Config {
         `${diff}\n\nWrite a commit message for the above changes. Use no more than 20 tokens to give a brief description in the imperative mood (e.g. 'Add feature' not 'Added feature'):`,
         {
           maxTokens: 20,
-        }
+        },
       )) {
         yield message;
       }