Releases: instill-ai/instill-core
v0.9.1-alpha
Features ✨
- fix API gateway gRPC endpoint names
- improve controller probing performance
API-gateway (v0.2.5-alpha)
Bug Fixes
Model-backend (v0.16.0-alpha)
Features
- add model initialization module (#332) (aa753a5)
- remove list and cancel operation method (#334) (3523918)
Controller (v0.1.1-alpha)
Bug Fixes
Console (v0.21.4-alpha)
Bug Fixes
v0.9.0-alpha
🎉 Highlights - Simpler model component design and a new Controller to monitor the resource state
Simpler model component
In this release, we retired the model instance concept and adopted a much simpler model design. Check out the details in this issue. To put it simply, Model becomes the core component of a pipeline now. Here is one pipeline recipe example below:
{
"source": "source-connectors/source-http",
"models": ["models/stable-diffusion"],
"destination": "destination-connectors/postgres-db"
}
When importing a model from GitHub, we specify the Git tag along with the GitHub repository. Try importing a Stable diffusion model to run on your GPU.
You may ask, how about model versioning? In the future, we plan to introduce the concept of labels to help organise models.
New Controller service to monitor the resource state
We also introduce a new service Controller into VDP. The controller periodically checks and maintains the real state of all resources (connector, model and pipeline) in an etcd database. The Console will fetch the real state from the new /watch endpoint of each service to make sure the state indicators are always up-to-date. Check the updated VDP system infrastructure.
Product Updates
Features ✨
- simplify the model component by removing
model instanceconcept - add a new Controller service to monitor the resource state
VDP(0.9.0-alpha)
Miscellaneous Chores
- release: release v0.9.0-alpha (0670c0a)
API-gateway (v0.2.4-alpha)
Miscellaneous Chores
- release v0.2.4-alpha (797410a)
Pipeline-backend (v0.11.0-alpha)
Features
Connector-backend (v0.9.0-alpha)
Features
Model-backend (v0.15.0-alpha)
Features
- controller: add model state monitoring with controller (#323) (4397826)
- remove model instance (#320) (15e1b62)
- support model caching (#317) (d15ffba)
Mgmt-backend (v0.3.3-alpha)
Miscellaneous Chores
- release v0.3.3-alpha (6c87fed)
Controller (v0.1.0-alpha)
Features
Bug Fixes
- add missing integration test files (#6) (bdb8eb7)
- hotfix dockerfile (96555e5)
- model: adopt latest model structure (#7) (d5c4264)
Console (v0.21.1-alpha)
Features
Bug Fixes
v0.8.1-alpha
Product Update
Features ✨
- Refactor all backends to have public and private API endpoints
VDP (0.8.1-alpha)
Miscellaneous Chores
- release v0.8.1-alpha (87084c9)
API-gateway (v0.2.3-alpha)
Miscellaneous Chores
- release v0.2.3-alpha (bbfa661)
Pipeline-backend (v0.10.0-alpha)
Features
Bug Fixes
Connector-backend (v0.8.0-alpha)
Features
Model-backend (v0.14.0-alpha)
Features
Bug Fixes
- config: use private port for mgmt-backend (#307) (3264e2b)
- list models and model instances pagination (#304) (1f19ed4)
Mgmt-backend (v0.3.2-alpha)
Miscellaneous Chores
- release v0.3.2-alpha (7a3d1bb)
Console (v0.20.0-alpha)
Features
- [INS-250] update the logic of tracking cookie (#372) (4a828e9)
- [INS-267] mock onboarded cookie (#378) (bbf8322)
- add playwright state (#373) (b1ff209)
- Impl pagination (#365) (6db8461)
- make test operate in order and bump playwright version (#374) (ddfcd5a)
- update how we store playwright state (#376) (9bf8de5)
Bug Fixes
v0.8.0-alpha
🎉 Highlight - Deploy VDP on Kubernetes using Helm
🤖 How to deploy VDP on Kubernetes?
Besides Docker Compose, VDP currently supports deployment via Kubernetes using Helm. It offers an elastic way of managing VDP to persist the services and scale resources horizontally and vertically.
📘 Check out the deployment documentation.
To install the VDP chart:
helm repo add instill https://helm.instill.tech
helm repo update
helm install vdp instill/vdp --devel --namespace vdp --create-namespaceAnd visit the api-gateway (http://localhost:8080) and Console (http://localhost:3000) by port forwarding:
export APIGATEWAY_POD_NAME=$(kubectl get pods --namespace vdp -l "app.kubernetes.io/component=api-gateway,app.kubernetes.io/instance=vdp" -o jsonpath="{.items[0].metadata.name}")
export APIGATEWAY_CONTAINER_PORT=$(kubectl get pod --namespace vdp $APIGATEWAY_POD_NAME -o jsonpath="{.spec.containers[0].ports[0].containerPort}")
export CONSOLE_POD_NAME=$(kubectl get pods --namespace vdp -l "app.kubernetes.io/component=console,app.kubernetes.io/instance=vdp" -o jsonpath="{.items[0].metadata.name}")
export CONSOLE_CONTAINER_PORT=$(kubectl get pod --namespace vdp $CONSOLE_POD_NAME -o jsonpath="{.spec.containers[0].ports[0].containerPort}")
kubectl --namespace vdp port-forward $APIGATEWAY_POD_NAME 8080:${APIGATEWAY_CONTAINER_PORT} &
kubectl --namespace vdp port-forward $CONSOLE_POD_NAME 3000:${CONSOLE_CONTAINER_PORT} &Product Updates
Features ✨
- Support VDP Kubernetes deployment
VDP (0.8.0-alpha)
Miscellaneous Chores
- release v0.8.0-alpha (4e5cb3d)
Console (v0.19.0-alpha)
Features
v0.7.0-alpha
🎉 Highlight - VDP supports Text to Image AI task now
🤖 What is Text to Image?
Text to Image is a Generative AI task to generate images from text inputs. Generally, the task takes descriptive text prompts as the input, and outputs generated images in Base64 format based on the text prompts.
📘 Check out the AI task documentation.
🚀 Try out the Text to Image pipeline on our demo website
We have set up a SYNC Text to Image pipeline stable-diffusion on our demo website.
Send a request and decode the generated Base64 image
curl -X POST https://demo.instill.tech/v1alpha/pipelines/stable-diffusion/trigger -d '{
"task_inputs": [
{
"text_to_image":{
"prompt": "a monkey drinking alone in a bar by Edward Hopper",
"steps": 40,
"cfg_scale": 10.5,
"seed": 32554,
"samples": 1
}
}
]
}' | jq -r '.model_instance_outputs[0].task_outputs[0].text_to_image.images[0]' | base64 -d > vdp-stable-diffusion.jpgProduct Updates
Features ✨
- Support Text Generation AI task
VDP (0.7.0-alpha)
Miscellaneous Chores
- docs: fix stable diffusion models by @xiaofei-du in #209
- chore(main): release 0.7.0-alpha by @droplet-bot in #210
Console (v0.18.0-alpha)
Features
- Thanks @iamnamananand996 - migrate react-query to version 4 (#352) (d72196a)
Bug Fixes
- Thanks @xiaofei-du - refactor the trigger snippet codes of models(#361) (400e54b)
- Thanks @iamnamananand996 - reset error object when create model form (#357) (e664c05)
Model-backend (v0.13.1-alpha)
Bug Fixes
Contributors
Assets 2
v0.6.0-alpha
🎉 Highlight - VDP supports Text Generation AI task now
🤖 What is Text Generation?
Text generation is a Generative AI task to generate new text from text inputs. Generally, the task takes incomplete text prompts as the input, and produces new text based on the prompts. The task can fill in incomplete sentences or even generate full stories given the first words.
📘 Check out the AI task documentation.
This is just a start! VDP is capable to serve more powerful models like GPT-J and BLOOM.
🚀 Try out the Text Generation pipeline on our demo website
If you want to try it out directly, we have set up a SYNC text generation pipeline gpt2 on our demo website.
Try it out by sending a request
curl -X POST https://demo.instill.tech/v1alpha/pipelines/gpt2/trigger -d '{
"task_inputs": [
{
"text_generation": {
"prompt": "once upon a time,",
"output_len": 50,
"topk": 5,
"seed": 0
}
}
]
}'or write your story on top of the pipeline (tutorials coming soon!)
Write.with.GPT2.on.VDP.mp4
Product Updates
Features ✨
- Support Text Generation AI task
- Implement new pipeline snippet based on AI task types
VDP (0.6.0-alpha)
Features
Bug Fixes
Console (0.17.0-alpha)
Features
- add health endpoint (#341) (488ef16)
- expose node env as build args (#345) (9a82499)
- further delay the model details test (bb0e955)
- further stabilize e2e test by removing the clean up function (#356) (eb0e58c)
- implement new pipeline snippet (#346) (01c8c41)
- turn off test related to long running operation (#354) (f28be7c)
- update env name (#358) (8f51ed8)
- update model task label (#347) (bb317f1)
- update tests to skip long-running operation related tests (#355) (e1ed5a8)
Bug Fixes
API-gateway (0.2.2-alpha)
Miscellaneous Chores
- release v0.2.2-alpha (d7f3dd0)
Pipeline-backend (0.9.8-alpha)
Miscellaneous Chores
- release v0.9.8-alpha (9051972)
Connector-backend (0.7.6-alpha)
Bug Fixes
- fix airbyte destination conenctor (a285180)
Model-backend (0.13.0-alpha)
Features
- add support for text generation tasks (instill-ai/model-backend#252) (767ec45)
Bug Fixes
- keep format for empty inference output (instill-ai/model-backend#258) (e2a2e48)
Mgmt-backend (0.3.1-alpha)
Bug Fixes
- fix admin and public servers mixed issue (8f8adef)
v0.5.0-alpha
Product Update
BREAKING change in Mgmt-backend 🅱️
Separate into public API (expose to client apps, such as console) and admin API (should NOT expose to the public internet traffic). Please check the latest protobufs for details.
VDP (0.5.0-alpha)
Miscellaneous Chores
- release 0.5.0-alpha (4cf3593)
API-gateway (0.2.1-alpha)
Bug Fixes
- fix endpoints and remove jwt token configuration (#13) (7b35a60)
- update configuration for headers allowed to reach the backend (#16) (a4f66b2)
Pipeline-backend (0.9.6-alpha)
Bug Fixes
- fix usage client nil issue when mgmt-backend not ready (d7c47fd)
- replace fatal logs with error logs (#102) (a410b29)
Connector-backend (0.7.5-alpha)
Bug Fixes
- fix usage client nil issue when mgmt-backend not ready (ab634a8)
- use error logs for usage (#61) (2c03f57)
Model-backend (0.12.0-alpha)
Features
Bug Fixes
Mgmt-backend (0.3.0-alpha)
Features
Bug Fixes
Console (0.16.0-alpha)
console-update
- We are at Nextjs 13 right now, a brand new world.
- Adapt newly released mgmt-backend
Features
- adapt new mgmt backend and protobuf (#337) (8f6f698)
- add pr-semantic workflow (#331) (e6084c0)
- extend the delay when test model details page (#336) (6b68a71)
- run build when test in the workflow (#339) (2b87fa1)
- upgrade nextjs13 (#334) (a488408)
Bug Fixes
v0.4.1-alpha
Product Update
Fix bugs 🐛
- add search attributes for Temporal
- fix the bug of triggering multiple images via multipart request
VDP (0.4.1-alpha)
Miscellaneous Chores
- release v0.4.1-alpha (1281217)
Pipeline-backend (v0.9.5-alpha)
Bug Fixes
- fix multipart file already closed issue (c5b6050)
Model-backend (v0.11.1-alpha)
Bug Fixes
- fix list long-run operation error (instill-ai/model-backend#220) (472696d)
v0.4.0-alpha
🚀 Highlight - New API Gateway as the single point of entry
The new API Gateway (http://localhost:8080) acts as the single point of entry for all backends. It aggregates all backends into one single entrance, so VDP users or clients can now focus on only one single interface. Find out how it works in the VDP docs.
Product Updates
Features ✨
- Support API Gateway via http://localhost:8080, check the VDP roadmap
- Support creating model asynchronously via long-run operation
Fix bugs 🐛
- Fix AI task displayed icons
- Support deploying model without Triton ensemble model
Documentation 📔
- Want to understand how VDP works behind the curtain? Check out the VDP system architecture.
VDP (0.4.0-alpha)
Features
API-gateway (0.2.0-alpha)
Features
- enable h2c (a1e3598)
Pipeline-backend (0.9.4-alpha )
Miscellaneous Chores
- release v0.9.4-alpha (e636cef)
Connector-backend (0.7.3-alpha )
Miscellaneous Chores
- release v0.7.3-alpha (f8e5028)
Model-backend (0.11.0-alpha)
Miscellaneous Chores
- release 0.11.0-alpha (d592acb)
Mgmt-backend (0.2.8-alpha)
Miscellaneous Chores
- release v0.2.8-alpha (3e7904b)
Console (0.15.0-alpha)
- Support asynchronously create model
- Support API-GATEWAY
- Implement the integration-test container and run our test in it
- Implement a new way to handle the run-time environment vairables
- Fix various bugs related to integration-test
- Fix bug related to model label displayed icons[
Features
- add InstanceSegmentationIcon into label (#327) (fabb36a)
- add SemanticSegmentationIcon icon (#329) (76cdb20)
- fix operation query issue (#328) (54d7aef)
- Impl running playwright inside docker (#318) (4da12aa)
- let maintainer switch playwright container user (#320) (f9c15e5)
- prolong test timeout (#323) (59f93e1)
- reduce the test time from 60000 ms to 15000 ms (#319) (3d4b024)
- stablize async model deploying test (#314) (94e867f)
- support async model creation (#324) (dedcd82)
- suspend the test related to cookie (#322) (bd54f71)
Bug Fixes
This PR was generated with Release Please. See documentation.
v0.3.0-alpha
xiaofei-du
Xiaofei Du
7b80636
xiaofei-du
Xiaofei Du
Product Updates
Announcement 📣
- VDP (originally, Visual Data Preparation) is officially renamed to
Versatile Data Pipeline.
We have realised that as a general ETL infrastructure, VDP is in fact capable of processing all kinds of unstructured data. We should not limit its usage to only visual data but to more general versatile data. In addition, the term, Data Preparation, has been misleading for users often thinking it has to do with only data labelling, cleaning, or wrangling. In our vision, while VDP should involve data preparation in its MLOps practice, it should not be conceptually confined to only data preparation. VDP does more than that and is focused on the overall effectiveness of the unstructured data ETL with a data-centric paradigm. The end form of a VDP infrastructure is nothing but a data pipeline. The term Data Pipeline is more precise to capture the core concept of VDP hence renamed as Versatile Data Pipeline.
Features ✨
- support new task Instance segmentation. Check out the Streamlit example
VDP (0.3.0-alpha)
Features
- support Instance segmentation task (0476f59)
- add console e2e test into vdp (#148) (a779a11)
- add instance segmentation example (#167)(c341e0c)
Bug Fixes
- fix wrong triton environment when deploying HuggingFace models (#150) (b2fda36)
- use COCO RLE format for instance segmentation (4d10e46)
- update model output protocol (e6ea88d)
- fix docker-compose (#174) (e40d607)
- fix: fix typo in makefile (#172) (c861afd)
Pipeline-backend (0.9.3-alpha)
Bug Fixes
- fix pipeline trigger model hanging (instill-ai/pipeline-backend#80) (7ba58e5)
Connector-backend (0.7.2-alpha)
Bug Fixes
- fix connector empty description update (0bc3086)
Model-backend (0.10.0-alpha)
Features
- support instance segmentation task (instill-ai/model-backend#183) (d28cfdc)
- support async deploy and undeploy model instance (instill-ai/model-backend#192) (ed36dc7)
- support semantic segmentation (instill-ai/model-backend#203) (f22262c)
Bug Fixes
- allow updating emtpy description for a model (instill-ai/model-backend#177) (100ec84)
- HuggingFace batching bug in preprocess model (b1582e8)
- model instance state update to unspecified state (instill-ai/model-backend#206) (14c87d5)
- panic error with nil object (instill-ai/model-backend#208) (a342113)
Console
Features
- extend the time span of our user cookie (instill-ai/console#289) (76a6f99)
- finish integration test and make it stable (instill-ai/console#281) (3fd8d21)
- replace prism.js with code-hike (instill-ai/console#292) (cb61708)
- unify the gap between elements in every table (instill-ai/console#291) (e743820)
- update console request URL according to new protobuf (instill-ai/console#287) (fa7ecc3)
- add hg model id field at model_instance page (instill-ai/console#300) (31a6eab)
- cleanup connector after test (instill-ai/console#295) (f9c8e4c)
- disable html report (instill-ai/console#297) (689f50d)
- enhance the warning of the resource id field (instill-ai/console#303) (6c4aa4f)
- make playwright output dot on CI (instill-ai/console#293) (e5c2958)
- support model-backend async long run operation (instill-ai/console#309) (f795ce8)
- update e2e test (instill-ai/console#313) (88bf0cd)
update how we test model detail page (instill-ai/console#310) (04c83a1) - wipe out all data after test (instill-ai/console#296) (e4085dd)
Bug Fixes
- fix pipeline e2e not stable (instill-ai/console#285) (a26e599)
- fix set-cookie api route issue due to wrong domain name (instill-ai/console#284) (c3efcdd)
This PR was generated with Release Please. See documentation.