tabby/CHANGELOG.md

# Changelog
All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html),
and is generated by [Changie](https://github.com/miniscruff/changie).

## v0.20.0 (2024-11-08)

### Features

* Search results can now be edited directly.
* Allow switching backend chat models in Answer Engine.
* Added a connection test button in the `System` tab to test the connection to the backend LLM server.

### Fixes and Improvements

* Optimized CR-LF inference in code completion. ([#3279](https://github.com/TabbyML/tabby/issues/3279))
* Bumped `llama.cpp` version to `b3995`.

## v0.19.0 (2024-10-30)

### Features

* For Answer Engine, when the file content is reasonably short (e.g., less than 200 lines of code), include the entire file content directly instead of only the chunk ([#3096](https://github.com/TabbyML/tabby/issues/3096)).
* Allowed adding additional languages through the `config.toml` file.
* Allowed customizing the `system_prompt` for Answer Engine.

### Fixes and Improvements

* Redesigned homepage to make team activities (e.g., threads discussed in Answer Engine) discoverable.
* Supported downloading models with multiple partitions (e.g., Qwen-2.5 series).

## v0.18.0 (2024-10-08)

### Notice

* The Chat Side Panel implementation has been redesigned in version 0.18, necessitating an extension version bump for compatibility with 0.18.0.
  - VSCode: >= 1.12.0
  - IntelliJ: >= 1.8.0

### Features

* User Groups Access Control: Server Administrators can now assign user groups to specific context providers to precisely control which contexts can be accessed by which user groups.

## v0.17.0 (2024-09-10)

### Notice

* We've reworked the `Web` (a beta feature) context provider into the `Developer Docs` context provider. Previously added context in the `Web` tab has been cleared and needs to be manually migrated to `Developer Docs`.

### Features

* Extensive rework has been done in the answer engine search box.
  - Developer Docs / Web search is now triggered by `@`.
  - Repository Context is now selected using `#`.

* Supports OCaml

## v0.16.1 (2024-08-27)

### Notice
* Starting from this version, we are utilizing websockets for features that require streaming (e.g., Answer Engine and Chat Side Panel). If you are deploying tabby behind a reverse proxy, you may need to configure the proxy to support websockets.

### Features

* Discussion threads in the Answer Engine are now persisted, allowing users to share threads with others.

### Fixed and Improvements

* Fixed an issue where the llama-server subprocess was not being reused when reusing a model for Chat / Completion together (e.g., Codestral-22B) with the local model backend.
* Updated llama.cpp to version b3571 to support the jina series embedding models.

## v0.15.0 (2024-08-08)

### Features

* The search bar in the Code Browser has been reworked and integrated with file navigation functionality.
* GraphQL syntax highlighting support in Code Browser.

### Fixed and Improvements

* For linked GitHub repositories, issues and PRs are now only returned when the repository is selected.
* Fixed GitLab issues/MRs indexing - no longer panics if the description field is null.
* When connecting to localhost model servers, proxy settings are now skipped.
* Allow set code completion's `max_input_length` and `max_output_tokens` in config.toml

## v0.14.0 (2024-07-23)

### Features
* Code search functionality is now available in the `Code Browser` tab. Users can search for code using regex patterns and filter by language, repository, and branch.
* Initial experimental support for natural language to codebase conversation in `Answer Engine`.

### Fixed and Improvements

* Incremental issues / PRs indexing by checking `updated_at`.
* Canonicalize `git_url` before performing a relevant code search. Previously, for git_urls with credentials, the canonicalized git_url was used in the index, but the query still used the raw git_url.
* bump llama.cpp to b3370 - which fixes Qwen2 model series inference

## v0.13.1 (2024-07-10)

### Fixed and Improvements

* Bump llama.cpp version to b3334, supporting Deepseek V2 series models.
* Turn on fast attention for Qwen2-1.5B model to fix the quantization error.
* Properly set number of GPU layers (to zero) when device is CPU.
## v0.13.0 (2024-06-28)

### Features

* Introduced a new Home page featuring the Answer Engine, which activates when the chat model is loaded.
* Enhanced the Answer Engine's context by indexing issues and pull requests.
* Supports web page crawling to further enrich the Answer Engine's context.
* Enabled navigation through various git trees in the git browser.

### Fixed and Improvements

* Turn on sha256 checksum verification for model downloading.
* Added an environment variable `TABBY_HUGGINGFACE_HOST_OVERRIDE` to override `huggingface.co` with compatible mirrors (e.g., `hf-mirror.com`) for model downloading.
* Bumped `llama.cpp` version to [b3166](https://github.com/ggerganov/llama.cpp/releases/tag/3166).
* Improved logging for the `llama.cpp` backend.
* Added support for triggering background jobs in the admin UI.
* Enhanced logging for backend jobs in the admin UI.

## v0.12.0 (2024-05-31)

### Features

* Support Gitlab SSO
* Support connect with Self-Hosted Github / Gitlab
* Repository Context is now utilizied in "Code Browser" as well

### Fixed and Improvements

* llama-server from llama.cpp is now distributed as an individual binary, allowing for more flexible configuration
* HTTP API is out of experimental - you can connect tabby to models through HTTP API. Right now following APIs are supported:
  - llama.cpp
  - ollama
  - mistral / codestral
  - openai

## v0.11.1 (2024-05-14)

### Fixed and Improvements

* Fixed display of files where the path contains special characters. ([#2081](https://github.com/TabbyML/tabby/issues/2081))
* Fixed non-admin users not being able to see the repository in Code Browser. ([#2110](https://github.com/TabbyML/tabby/discussions/2110))

## v0.11.0 (05/10/2024)

### Notice

* The `--webserver` flag is now enabled by default in `tabby serve`. To turn off the webserver and only use OSS features, use the `--no-webserver` flag.
* The `/v1beta/chat/completions` endpoint has been moved to `/v1/chat/completions`, while the old endpoint is still available for backward compatibility.

### Features
* Upgraded [llama.cpp](https://github.com/ggerganov/llama.cpp) to version [b2715](https://github.com/ggerganov/llama.cpp/releases/tag/b2715).
* Added support for integrating repositories from GitHub and GitLab using personal access tokens.
* Introduced a new **Activities** page to view user activities.
* Implemented incremental indexing for faster repository context updates.
* Added storage usage statistics in the **System** page.
* Included an `Ask Tabby` feature in the source code browser to provide in-context help from AI.

### Fixes and Improvements
* Changed the default model filename from `q8_0.v2.gguf` to `model.gguf` in MODEL_SPEC.md.
* Excluded activities from deactivated users in reports.

## v0.10.0 (04/22/2024)

### Features
* Introduced the `--chat-device` flag to specify the device used to run the chat model.
* Added a "Reports" tab in the web interface, which provides team-wise statistics for Tabby IDE and Extensions usage (e.g., completions, acceptances).
* Enabled the use of segmented models with the `tabby download` command.
* Implemented the "Go to file" functionality in the Code Browser.

### Fixes and Improvements
* Fix worker unregisteration misfunctioning caused by unmatched address.
* Accurate repository context filtering using fuzzy matching on `git_url` field.
* Support the use of client-side context, including function/class declarations from LSP, and relevant snippets from local changed files.

## v0.9.1 (03/19/2024)

### Fixes and Improvements
* Fix worker registration check against enterprise licenses.
* Fix default value of `disable_client_side_telemetry` when `--webserver` is not used.

## v0.9.0 (03/06/2024)

### Features

* Support for SMTP configuration in the user management system.
* Support for SSO and team management as features in the Enterprise tier.
* Fully managed repository indexing using `--webserver`, with job history logging available in the web interface.

## v0.8.3 (02/06/2024)

### Fixes and Improvements

* Ensure `~/.tabby/repositories` exists for tabby scheduler jobs: https://github.com/TabbyML/tabby/pull/1375
* Add cpu only binary `tabby-cpu` to docker distribution.

## v0.8.0 (02/02/2024)

### Notice

* Due to format changes, re-executing `tabby scheduler --now` is required to ensure that `Code Browser` functions properly.

### Features

* Introducing a preview release of the `Source Code Browser`, featuring visualization of code snippets utilized for code completion in RAG.
* Added a Windows CPU binary distribution.
* Added a Linux ROCm (AMD GPU) binary distribution.

### Fixes and Improvements

* Fixed an issue with cached permanent redirection in certain browsers (e.g., Chrome) when the `--webserver` flag is disabled.
* Introduced the `TABBY_MODEL_CACHE_ROOT` environment variable to individually override the model cache directory.
* The `/v1beta/chat/completions` API endpoint is now compatible with OpenAI's chat completion API.
* Models from our official registry can now be referred to without the TabbyML prefix. Therefore, for the model TabbyML/CodeLlama-7B, you can simply refer to it as CodeLlama-7B everywhere.

## v0.7.0 (12/15/2023)

### Features

* Tabby now includes built-in user management and secure access, ensuring that it is only accessible to your team.
* The `--webserver` flag is a new addition to `tabby serve` that enables secure access to the tabby server. When this flag is on, IDE extensions will need to provide an authorization token to access the instance.
  - Some functionalities that are bound to the webserver (e.g. playground) will also require the `--webserver` flag.


### Fixes and Improvements

*  Fix https://github.com/TabbyML/tabby/issues/1036, events log should be written to dated json files.

## v0.6.0 (11/27/2023)

### Features

* Add distribution support (running completion / chat model on different process / machine).
* Add conversation history in chat playground.
* Add `/metrics` endpoint for prometheus metrics collection. 

### Fixes and Improvements

* Fix the slow repository indexing due to constraint memory arena in tantivy index writer.
* Make `--model` optional, so users can create a chat only instance.
* Add `--parallelism` to control the throughput and VRAM usage: https://github.com/TabbyML/tabby/pull/727

## v0.5.5 (11/09/2023)

### Fixes and Improvements

### Notice

* llama.cpp backend (CPU, Metal) now requires a redownload of gguf model due to upstream format changes: https://github.com/TabbyML/tabby/pull/645 https://github.com/ggerganov/llama.cpp/pull/3252
* Due to indexing format changes, the `~/.tabby/index` needs to be manually removed before any further runs of `tabby scheduler`.
* `TABBY_REGISTRY` is replaced with `TABBY_DOWNLOAD_HOST` for the github based registry implementation.

### Features

* Improved dashboard UI.

### Fixes and Improvements

* Cpu backend is switched to llama.cpp: https://github.com/TabbyML/tabby/pull/638
* add `server.completion_timeout` to control the code completion interface timeout: https://github.com/TabbyML/tabby/pull/637
* Cuda backend is switched to llama.cpp: https://github.com/TabbyML/tabby/pull/656
* Tokenizer implementation is switched to llama.cpp, so tabby no longer need to download additional tokenizer file: https://github.com/TabbyML/tabby/pull/683
* Fix deadlock issue reported in https://github.com/TabbyML/tabby/issues/718

## v0.4.0 (10/24/2023)

### Features

* Supports golang: https://github.com/TabbyML/tabby/issues/553
* Supports ruby: https://github.com/TabbyML/tabby/pull/597
* Supports using local directory for `Repository.git_url`: use `file:///path/to/repo` to specify a local directory.
* A new UI design for webserver.

### Fixes and Improvements

* Improve snippets retrieval by dedup candidates to existing content + snippets: https://github.com/TabbyML/tabby/pull/582

## v0.3.1 (10/21/2023)
### Fixes and improvements

* Fix GPU OOM issue caused the parallelism: https://github.com/TabbyML/tabby/issues/541, https://github.com/TabbyML/tabby/issues/587
* Fix git safe directory check in docker: https://github.com/TabbyML/tabby/issues/569

## v0.3.0 (10/13/2023)

### Features
#### Retrieval-Augmented Code Completion Enabled by Default

The currently supported languages are:

* Rust
* Python
* JavaScript / JSX
* TypeScript / TSX

A blog series detailing the technical aspects of Retrieval-Augmented Code Completion will be published soon. Stay tuned!

### Fixes and Improvements

* Fix [Issue #511](https://github.com/TabbyML/tabby/issues/511) by marking ggml models as optional.
* Improve stop words handling by combining RegexSet into Regex for efficiency.

## v0.2.2 (10/09/2023)
### Fixes and improvements

* Fix a critical issue that might cause request dead locking in ctranslate2 backend (when loading is heavy)

## v0.2.1 (10/03/2023)
### Features
#### Chat Model & Web Interface

We have introduced a new argument, `--chat-model`, which allows you to specify the model for the chat playground located at http://localhost:8080/playground

To utilize this feature, use the following command in the terminal:

```bash
tabby serve --device metal --model TabbyML/StarCoder-1B --chat-model TabbyML/Mistral-7B
```

#### ModelScope Model Registry

Mainland Chinese users have been facing challenges accessing Hugging Face due to various reasons. The Tabby team is actively working to address this issue by mirroring models to a hosting provider in mainland China called modelscope.cn.

```bash
## Download from the Modelscope registry
TABBY_REGISTRY=modelscope tabby download --model TabbyML/WizardCoder-1B
```

### Fixes and improvements

* Implemented more accurate UTF-8 incremental decoding in the [GitHub pull request](https://github.com/TabbyML/tabby/pull/491).
* Fixed the stop words implementation by utilizing RegexSet to isolate the stop word group.
* Improved model downloading logic; now Tabby will attempt to fetch the latest model version if there's a remote change, and the local cache key becomes stale.
* set default num_replicas_per_device for ctranslate2 backend to increase parallelism.



No releases yet, this file will be updated when generating your first release.