Releases: TabbyML/tabby
Releases Β· TabbyML/tabby
v0.6.0-rc.0
π Features
π· Built-in distribution support: running completion / chat model worker on different process / machine.
![image](https://private-user-images.githubusercontent.com/388154/285098582-fdf11aec-e960-46ef-86b0-1d79362b6554.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MzQ1MzUzMDgsIm5iZiI6MTczNDUzNTAwOCwicGF0aCI6Ii8zODgxNTQvMjg1MDk4NTgyLWZkZjExYWVjLWU5NjAtNDZlZi04NmIwLTFkNzkzNjJiNjU1NC5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQxMjE4JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MTIxOFQxNTE2NDhaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT0zYzE5NmIwNDlkOTUxMmFlODFhZjhlZTliMDIzNDIxYzIzOTk1YjM4MzQ1MThlMzdlMjUxMTkyYzZiNDE1YjUyJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.Ww_SWajHkYHa51LWyk1t8Rytk2IDLPJ_qdjSJHoO7JA)
The community edition has a restriction of a maximum of one worker for code completion/chat. The enterprise edition is still in private alpha and is only available to our design partners. If you're interested, please DM Meng Zhang
on the Slack channel to apply (limited slots available).
π¬ Conversation history in chat playground.
![image](https://private-user-images.githubusercontent.com/388154/285098211-b5295acb-70b4-4837-91ab-2360ed24088e.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MzQ1MzUzMDgsIm5iZiI6MTczNDUzNTAwOCwicGF0aCI6Ii8zODgxNTQvMjg1MDk4MjExLWI1Mjk1YWNiLTcwYjQtNDgzNy05MWFiLTIzNjBlZDI0MDg4ZS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQxMjE4JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MTIxOFQxNTE2NDhaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT01MTk4OGI2NzRiMDViODNjYWFhNWQxZWE2NGRlOGVkZDUwZmNhMjJhMjA3N2Y5NjUyNDM5ZjJiM2I2NGE2N2NhJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.rQluK3gPk11gjbo25B9VhLqq0ayURv-a1RNNnzl5nE0)
π§° Fixes and Improvements
- Fix the slow repository indexing due to constraint memory arena in tantivy index writer.
- Command line argument
--model
is now optional, so users can create a chat only instance. - New command line argument
--parallelism
to control the throughput and VRAM usage: #727 - New api path
/metrics
endpoint for prometheus metrics collection.
π« New Contributors
- @liangfung made their first contribution in #702
- @erfanium made their first contribution in #742
- @costanzo made their first contribution in #748
- @darknight made their first contribution in #750
- @suside made their first contribution in #775
- @jpoisso made their first contribution in #838
- @Squadrick made their first contribution in #849
- @sonique6784 made their first contribution in #813
Full Changelog: v0.5.5...v0.6.0-rc.0
v0.5.5
β οΈ Notice
- llama.cpp backend (CPU, Metal) now requires a redownload of gguf model due to upstream format changes: #645 ggerganov/llama.cpp#3252
- Due to indexing format changes, the
~/.tabby/index
needs to be manually removed before any further runs oftabby scheduler
. TABBY_REGISTRY
is replaced withTABBY_DOWNLOAD_HOST
for the github based registry implementation.
π Features
- Improved dashboard UI.
![image](https://private-user-images.githubusercontent.com/388154/280425016-02c0c488-7fe0-4e99-bcea-99fe7d3a2ace.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MzQ1MzUzMDgsIm5iZiI6MTczNDUzNTAwOCwicGF0aCI6Ii8zODgxNTQvMjgwNDI1MDE2LTAyYzBjNDg4LTdmZTAtNGU5OS1iY2VhLTk5ZmU3ZDNhMmFjZS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQxMjE4JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MTIxOFQxNTE2NDhaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT0yNmE3ZjJkMzYzYzcxYzlmZTY5YmE1ZDVmYTAzZWZmYmQ4NjQ4YjU0NDgzODg1YTgzYzFjZDNhMDYwOTUxMWNmJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.-p1N9nT_FY4uKctX6krd-FvKOVC_ZSJEOpXSIWMNPX0)
π§° Fixes and Improvements
- Cpu backend is switched to llama.cpp: #638
- add
server.completion_timeout
to control the code completion interface timeout: #637 - Cuda backend is switched to llama.cpp: #656
- Tokenizer implementation is switched to llama.cpp, so tabby no longer need to download additional tokenizer file: #683
π« New Contributors
- @CrCs2O4 made their first contribution in #597
- @yusiwen made their first contribution in #620
- @gjedeer made their first contribution in #635
- @XpycT made their first contribution in #634
- @HKABIG made their first contribution in #662
Full Changelog: v0.4.0...v0.5.5
v0.5.4
v0.5.3
β οΈ Notice
- llama.cpp backend (CPU, Metal) now requires a redownload of gguf model due to upstream format changes: #645 ggerganov/llama.cpp#3252
- Due to indexing format changes, the
~/.tabby/index
needs to be manually removed before any further runs oftabby scheduler
. TABBY_REGISTRY
is replaced withTABBY_DOWNLOAD_HOST
for the github based registry implementation.
π Features
- Improved dashboard UI.
![image](https://private-user-images.githubusercontent.com/388154/280425016-02c0c488-7fe0-4e99-bcea-99fe7d3a2ace.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MzQ1MzUzMDgsIm5iZiI6MTczNDUzNTAwOCwicGF0aCI6Ii8zODgxNTQvMjgwNDI1MDE2LTAyYzBjNDg4LTdmZTAtNGU5OS1iY2VhLTk5ZmU3ZDNhMmFjZS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQxMjE4JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MTIxOFQxNTE2NDhaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT0yNmE3ZjJkMzYzYzcxYzlmZTY5YmE1ZDVmYTAzZWZmYmQ4NjQ4YjU0NDgzODg1YTgzYzFjZDNhMDYwOTUxMWNmJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.-p1N9nT_FY4uKctX6krd-FvKOVC_ZSJEOpXSIWMNPX0)
π§° Fixes and Improvements
- Cpu backend is switched to llama.cpp: #638
- add
server.completion_timeout
to control the code completion interface timeout: #637 - Cuda backend is switched to llama.cpp: #656
- Tokenizer implementation is switched to llama.cpp, so tabby no longer need to download additional tokenizer file: #683
π« New Contributors
- @CrCs2O4 made their first contribution in #597
- @yusiwen made their first contribution in #620
- @gjedeer made their first contribution in #635
- @XpycT made their first contribution in #634
- @HKABIG made their first contribution in #662
Full Changelog: v0.4.0...v0.5.3
v0.5.2-rc.0
v0.5.2-rc.0
v0.5.2
Release 0.5.2 http-api-bindings@0.5.2 llama-cpp-bindings@0.5.2 tabby@0.5.2 tabby-common@0.5.2 tabby-download@0.5.2 tabby-inference@0.5.2 tabby-scheduler@0.5.2 Generated by cargo-workspaces
v0.5.1-rc.2
v0.5.1-rc.2
v0.5.1-rc.1
v0.5.1-rc.1
v0.5.0-rc.4
v0.5.0-rc.4
v0.5.0
Release 0.5.0 http-api-bindings@0.5.0 llama-cpp-bindings@0.5.0 tabby@0.5.0 tabby-common@0.5.0 tabby-download@0.5.0 tabby-inference@0.5.0 tabby-scheduler@0.5.0 Generated by cargo-workspaces