Windows and cuda support #163
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
**Description:**
This PR adds various features, fixes, and improvement that were
implemented in preparation of the 2.6.0-beta release.
**Changes Made:**
##### Create
* allow to drag and drop images from create tab into workflow settings
* adjust default image generation counts for default modes
* add quick select for default mode image quality and resolution
* switch image output folder naming scheme to YYYYmmdd
* adjust various parts of the default mode image generation to adapt to
new dependency versions
* ComfyUI
* add link to open comfyui in browser
* categorize comfyui workflows in workflow drop-down menu
* bump version to latest
* add Wan2.1 workflows
* add drag and drop for generated images
* add all-in-one flux kontext workflow
* add Acer Vision Art workflow for devices where Acer Vision Art is
installed
* fix issue where comfyui workflows were not refreshed after being
updated on disk
* change comfyui workflow update behavior to load workflows from a
release specific branch
##### Enhance
* automatically set aspect ratio in Enhance based on loaded image, when
image is loaded or mode is switched
##### Answer
* remove ipex-llm backend
* upgrade openvino
* add additional openvino npu models
* add additional llama-cpp gguf models
* add context size input for llama-cpp backend
* add loading overlay for backend reloads
* improve markdown parsing and code block copy functionality
* make code block header sticky to allow easier copying of long code
blocks
* support pulling ollama models directly from AI Playground
(experimental feature -> settings.json `enablePreviewFeatures: true` )
##### Other
* adjust model download temporary folder naming to avoid issues with
running into windows max path length
* add demo mode and allow to select starting tab via command line
argument (@HardikLalcheta)
* make selected theme persistent across restarts
* add version settings for all backends
* make settings tab not automatically hide when clicking other parts of
the UI
* make settings tab displace the main area instead of overlaying
* remove libuv from python-env
* add missing translations
* allow to override language via settings.json, e.g. `overrideLanguage:
"en-US"` (current options
https://github.com/intel/AI-Playground/tree/main/WebUI/src/assets/i18n)
* Upgrade dependencies
* Upgrade to PyTorch 2.8
* upgrade various python dependencies
* upgrade npm dependencies
**Testing Done:**
Tested locally on B580, A770, A750
**Checklist:**
- [x] I have tested the changes locally.
- [x] I have self-reviewed the code changes.
**Description:** This PR upgrades various dependencies. It also replaces BasicSR with a different package to avoid a vulnerability in BasicSR. **Changes Made:** * upgrade various dependencies * upgrade OpenVINO version **Testing Done:** Tested locally on B580 **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes.
**Description:** This PR implements various changes in preparation of release 2.6.1-beta. **Changes Made:** * fix `CopyFace` and `FaceSwap-HD` workflows * add Nomic embeddings model * upgrade Llama.cpp version * improve conversation manager UX (@KyleHagy) * upgrade dependencies * upgrade OpenVINO to 2025.3 * add context size selector for OpenVINO on NPU devices * improve zoom UX * zooming in is now also possible via `Ctrl`+`+` * zooming in and out is now also possible via `Ctrl` and Mousewheel * improve backend installation and startup error display * add pip freeze output of the failing backend to error details * capture error messages during backend startup and add to error details **Testing Done:** Tested locally on B580 **Screenshots:** <img width="500" height="500" alt="image" src="https://github.com/user-attachments/assets/9aef2f7b-8cce-4dc7-83fc-5d1a8dbdf57e" /> **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes. --------- Co-authored-by: kyle <[email protected]> Co-authored-by: KyleHagy <[email protected]> Co-authored-by: Daniel Hirsch <[email protected]>
**Description:** This PR improves the resolution of dll required by the inference backends. It also updates npm dependencies and fixes the ollama PoC integration. **Changes Made:** * add `Library/bin` of the respective inference backend to the front of the `PATH` env variable to improve dll resolution * update npm dependencies * fix ollama PoC **Testing Done:** Tested locally on B580 **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes.
…#346) **Description:** Fix an issue, where the `bin` folder, containing required DLLs was not correctly added to the `PATH` environment variable during device detection. **Changes Made:** * correctly set `PATH` during device detection. **Testing Done:** Tested locally on B580. **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes.
**Description:** This PR improves UX by remembering the last used device for each backend after restarting AI Playgronud **Changes Made:** * use last used device per backend if still available when AI Playground is restarted. **Testing Done:** Tested locally on B580. **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes.
**Description:** This PR prevents pip to use global config files during package installations. **Changes Made:** * set env variable to disable global config file loading **Testing Done:** Tested locally on B580. **Checklist:** - [x] I have tested the changes locally. - [x] I have self-reviewed the code changes.
**Description:** This PR contains all changes for the upcoming Intel AI Playground 3.0.0-alpha release. **Changes Made:** # AI Playground 3.0.0-alpha Release Notes ## 🎨 Major UI/UX Overhaul ### New Unified Interface - Complete UI redesign with new unified viewport architecture - Light theme support alongside existing dark theme - Responsive side modals that switch between overlay and sidebar modes - Updated branding and styling throughout the application - Preset-based architecture for all features --- ## 🎤 Speech-to-Text - Whisper integration for voice input - Microphone selector and audio level visualization - Silence detection for automatic recording stop --- ## 🤖 Enhanced LLM Chat Features ### Vercel AI SDK Integration - Full migration to Vercel AI SDK (`@ai-sdk/vue`, `@ai-sdk/openai-compatible`) - Streaming responses with proper token tracking (not supported on OpenVINO NPU) - Built-in tool calling support ### Tool Calling / Agentic Features - ComfyUI image generation as a tool - Generate images from chat - Object detection as a tool - Detect objects in images ### Vision Language Model Support - Images can be added to the chat as a user message - Insert images via drag and drop, file input, or clipboard paste ### Chat UI Improvements - Reset chat preset button - Improved conversation management - Temperature selector - Context usage display --- ## 🖼️ Image Generation Improvements ### Inpaint/Outpaint - Enhance Features (Inpaint, Outpaint, Upscale) moved to ComfyUI backend - Improved mask editor with better performance - Improved outpaint widget with more control ### Model Support - Various model updates - Z-Image support ### Safety Features - Safety checker for all image generation presets and inpaint/outpaint presets --- ## 📦 3D Model Generation - Image-to-3D preset using Hunyuan3D - Inline 3D model viewer component --- ## 🔌 Backend Improvements ### OpenVINO Backend - Major OpenVINO backend service rewrite to OpenVINO Model Server - OVMS Whisper integration ### LlamaCPP Backend - LlamaCPP backend service rewrite to LlamaCPP Model Server and Vulkan Backend ### ComfyUI Backend - Updated to ComfyUI 0.4.0 - PyTorch 2.10 RC1 integration - ComfyUI Manager integration - Improved model path handling - ComfyUI dependency UI ### Python Environment - UV-based backend management - Removed legacy image generation code (`paint_biz.py`, `sd_adapter.py`, `realesrgan.py`, etc.) --- ## 🔧 User Experience & Small Improvements ### Build & Tooling - Silent installer support ### Version Management - Version-based user guide - Backend version display - Improved version handling UX **Testing Done:** Tested locally on B580 and 155H **Checklist:** - [x] I have tested the changes locally. --------- Co-authored-by: Timo Wuttke <[email protected]> Co-authored-by: Stefan Petschauer <[email protected]> Co-authored-by: Daniel Hirsch <[email protected]>
Signed-off-by: Leslie Lazzarino <[email protected]>
leslielazzarino
force-pushed
the
windows-and-cuda-support
branch
from
6bf774b to
c22b7fa
Compare
Also exposes the CPU as device for debugging Signed-off-by: Leslie Lazzarino <[email protected]>
leslielazzarino
force-pushed
the
windows-and-cuda-support
branch
from
c22b7fa to
c924541
Compare
Signed-off-by: Leslie Lazzarino <[email protected]>
leslielazzarino
force-pushed
the
windows-and-cuda-support
branch
from
fd11dc4 to
b3698b2
Compare
Signed-off-by: Leslie Lazzarino <[email protected]>
leslielazzarino
force-pushed
the
windows-and-cuda-support
branch
from
b3698b2 to
10b3177
Compare
Signed-off-by: Leslie Lazzarino <[email protected]>
leslielazzarino
force-pushed
the
windows-and-cuda-support
branch
from
10b3177 to
1ced592
Compare
mschuettlerTNG
force-pushed
the
dev
branch
3 times, most recently
from
2ecf46f to
cf425d3
Compare
mschuettlerTNG
force-pushed
the
dev
branch
4 times, most recently
from
f89c5ed to
b6c2451
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description:
This PR adds support for:
Changes Made:
On top of what I listed in the description:
Testing Done:
Tested Llama and ComfyUI, including video creation, on my cpu and NVIDIA RTX 4060 Laptop GPU (video creation only tested with the CPU, as the models do not fit on 8GB of VRAM). Also on an Arc GPU.
Checklist: