Merge pull request #3051 from janhq/dev

chore: release 0.5.1 to main
2024-06-17 15:28:38 +07:00 · 2024-06-17 15:28:38 +07:00 · 25d92c3677
commit 25d92c3677
parent 0a150b373c e9e1435aff
38 changed files with 906 additions and 244 deletions
--- a/.github/ISSUE_TEMPLATE/bug_report.md
+++ b/.github/ISSUE_TEMPLATE/bug_report.md
@ -1,37 +0,0 @@
---
-name: Bug report
-about: Create a report to help us improve Jan
-title: 'bug: [DESCRIPTION]'
-labels: 'type: bug'
-assignees: ''
-
---
-
-**Describe the bug**
-A clear and concise description of what the bug is.
-
-**Steps to reproduce**
-Steps to reproduce the behavior:
-1. Go to '...'
-2. Click on '....'
-3. Scroll down to '....'
-4. See error
-
-**Expected behavior**
-A clear and concise description of what you expected to happen.
-
-**Screenshots**
-If applicable, add screenshots to help explain your issue.
-
-**Environment details**
- Operating System: [Specify your OS. e.g., MacOS Sonoma 14.2.1, Windows 11, Ubuntu 22, etc]
- Jan Version: [e.g., 0.4.xxx nightly or manual]
- Processor: [e.g., Apple M1, Intel Core i7, AMD Ryzen 5, etc]
- RAM: [e.g., 8GB, 16GB]
- Any additional relevant hardware specifics: [e.g., Graphics card, SSD/HDD]
-
-**Logs**
-If the cause of the error is not clear, kindly provide your usage logs: https://jan.ai/docs/troubleshooting#how-to-get-error-logs
-
-**Additional context**
-Add any other context or information that could be helpful in diagnosing the problem.
--- a/.github/ISSUE_TEMPLATE/bug_report.yml
+++ b/.github/ISSUE_TEMPLATE/bug_report.yml
@ -0,0 +1,82 @@
+name: "\U0001F41B Bug Report"
+description: "If something isn't working as expected \U0001F914"
+labels: [ "type: bug" ]
+title: 'bug: [DESCRIPTION]'
+
+body:
+  - type: markdown
+    attributes:
+      value: "Thanks for taking the time to fill out this bug report!"
+
+  - type: checkboxes
+    attributes:
+      label: "#"
+      description: "Please search [here](./?q=is%3Aissue) to see if an issue already exists for the bug you encountered"
+      options:
+        - label: "I have searched the existing issues"
+          required: false
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Current behavior"
+      description: "A clear and concise description of what the bug is"
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Minimum reproduction step"
+      description: |
+        Please list out steps to reproduce the behavior
+      placeholder: |
+        1. Go to '...'
+        2. Click on '....'
+        3. Scroll down to '....'
+        4. See error
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Expected behavior"
+      description: "A clear and concise description of what you expected to happen"
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Screenshots / Logs"
+      description: |
+        Kindly provide your screenshots / [usage logs](https://jan.ai/docs/troubleshooting#how-to-get-error-logs) that could be helpful in diagnosing the problem
+        **Tip:** You can attach images, recordings or log files by clicking this area to highlight it and then dragging files in
+  - type: markdown
+    attributes:
+      value: |
+        ---
+
+  - type: input
+    validations:
+      required: true
+    attributes:
+      label: "Jan version"
+      description: "**Tip:** The version is located in the lower right conner of the Jan app"
+      placeholder: "e.g. 0.5.x-xxx nightly or stable"
+
+  - type: checkboxes
+    attributes:
+      label: "In which operating systems have you tested?"
+      options:
+        - label: macOS
+        - label: Windows
+        - label: Linux
+
+  - type: textarea
+    attributes:
+      label: "Environment details"
+      description: |
+        - Operating System: [Specify your OS details: e.g., MacOS Sonoma 14.2.1, Windows 11, Ubuntu 22, etc]
+        - Processor: [e.g., Apple M1, Intel Core i7, AMD Ryzen 5, etc]
+        - RAM: [e.g., 8GB, 16GB]
+        - Any additional relevant hardware specifics: [e.g., Graphics card, SSD/HDD]
--- a/.github/ISSUE_TEMPLATE/config.yml
+++ b/.github/ISSUE_TEMPLATE/config.yml
@ -0,0 +1,7 @@
+## To encourage contributors to use issue templates, we don't allow blank issues
+blank_issues_enabled: true
+
+contact_links:
+  - name: "\u2753 Our GitHub Discussions page"
+    url: "https://github.com/orgs/janhq/discussions/categories/q-a"
+    about: "Please ask and answer questions here!"
--- a/.github/ISSUE_TEMPLATE/discussion-thread.md
+++ b/.github/ISSUE_TEMPLATE/discussion-thread.md
@ -1,16 +0,0 @@
---
-name: Discussion thread
-about: Start an open ended discussion
-title: 'Discussion: [TOPIC HERE]'
-labels: 'type: discussion'
-assignees: ''
-
---
-
-**Motivation**
-
-**Discussion**
-
-**Alternatives**
-
-**Resources**
--- a/.github/ISSUE_TEMPLATE/documentation-request.md
+++ b/.github/ISSUE_TEMPLATE/documentation-request.md
@ -1,5 +1,5 @@
 ---
-name: Documentation request
+name: "📖 Documentation request"
 about: Documentation requests
 title: 'docs: TITLE'
 labels: 'type: documentation'
--- a/.github/ISSUE_TEMPLATE/epic-request.md
+++ b/.github/ISSUE_TEMPLATE/epic-request.md
@ -1,5 +1,5 @@
 ---
-name: Epic request
+name: "💥 Epic request"
 about: Suggest an idea for this project
 title: 'epic: [DESCRIPTION]'
 labels: 'type: epic'
--- a/.github/ISSUE_TEMPLATE/feature_request.md
+++ b/.github/ISSUE_TEMPLATE/feature_request.md
@ -1,17 +0,0 @@
---
-name: Feature request
-about: Suggest an idea for this project
-title: 'feat: [DESCRIPTION]'
-labels: 'type: feature request'
-assignees: ''
-
---
-
-**Problem**
-A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
-
-**Success Criteria**
-A clear and concise description of what you want to happen.
-
-**Additional context**
-Add any other context or screenshots about the feature request here.
--- a/.github/ISSUE_TEMPLATE/feature_request.yml
+++ b/.github/ISSUE_TEMPLATE/feature_request.yml
@ -0,0 +1,44 @@
+name: "\U0001F680 Feature Request"
+description: "Suggest an idea for this project \U0001F63B!"
+title: 'feat: [DESCRIPTION]'
+labels: 'type: feature request'
+body:
+  - type: markdown
+    attributes:
+      value: "Thanks for taking the time to fill out this form!"
+
+  - type: checkboxes
+    attributes:
+      label: "#"
+      description: "Please search [here](./?q=is%3Aissue) to see if an issue already exists for the feature you are requesting"
+      options:
+        - label: "I have searched the existing issues"
+          required: false
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Is your feature request related to a problem? Please describe it"
+      description: "A clear and concise description of what the problem is"
+      placeholder: |
+        I'm always frustrated when ...
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "Describe the solution"
+      description: "Description of what you want to happen. Add any considered drawbacks"
+
+  - type: textarea
+    attributes:
+      label: "Teachability, documentation, adoption, migration strategy"
+      description: "Explain how users will be able to use this and possibly write out something for the docs. Maybe a screenshot or design?"
+
+  - type: textarea
+    validations:
+      required: true
+    attributes:
+      label: "What is the motivation / use case for changing the behavior?"
+      description: "Describe the motivation or the concrete use case"
--- a/.github/workflows/jan-openai-api-test.yml
+++ b/.github/workflows/jan-openai-api-test.yml
@ -1,101 +0,0 @@
-name: Test - OpenAI API Pytest collection
-on:
-  workflow_dispatch:
-    inputs:
-      endpoints:
-        description: 'comma-separated list (see available at endpoints_mapping.json e.g. GET /users,POST /transform)'
-        required: false
-        default: all
-        type: string
-
-  push:
-    branches:
-      - main
-      - dev
-      - release/**
-    paths:
-      - "docs/**"
-
-  pull_request:
-    branches:
-      - main
-      - dev
-      - release/**
-    paths:
-      - "docs/**"
-
-jobs:
-  openai-python-tests:
-    runs-on: [self-hosted, Linux, ubuntu-desktop]
-    if: (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository) || github.event_name == 'push' || github.event_name == 'workflow_dispatch'
-    steps:
-      - name: Getting the repo
-        uses: actions/checkout@v3
-        with:
-          fetch-depth: 0
-
-      - name: Installing node
-        uses: actions/setup-node@v3
-        with:
-          node-version: 20
-
-      - name: "Cleanup cache"
-        continue-on-error: true
-        run: |
-          rm -rf ~/jan
-          make clean
-
-      - name: Install dependencies
-        run: |
-          npm install -g @stoplight/prism-cli
-
-      - name: Create python virtual environment and run test
-        run: |
-          python3 -m venv /tmp/jan
-          source /tmp/jan/bin/activate
-          # Clone openai-api-python repo
-          OPENAI_API_PYTHON_TAG=$(cat docs/openapi/version.txt)
-          git clone https://github.com/openai/openai-python.git
-          cd openai-python
-          git checkout $OPENAI_API_PYTHON_TAG
-
-          python3 -m venv /tmp/jan
-          source /tmp/jan/bin/activate
-          pip install -r requirements-dev.lock
-          pip install pytest-reportportal pytest-html
-          
-          # Create pytest.ini file with content
-          cat ../docs/tests/pytest.ini >> pytest.ini
-          echo "rp_api_key = ${{ secrets.REPORT_PORTAL_API_KEY }}" >> pytest.ini
-          echo "rp_endpoint = ${{ secrets.REPORT_PORTAL_URL_PYTEST }}" >> pytest.ini
-          cat pytest.ini
-          
-          # Append to conftest.py
-          cat ../docs/tests/conftest.py >> tests/conftest.py
-          cat ../docs/tests/endpoints_mapping.json >> tests/endpoints_mapping.json
-          
-          # start mock server and run test then stop mock server
-          prism mock ../docs/openapi/jan.yaml > prism.log & prism_pid=$! &&
-          pytest --endpoint "$ENDPOINTS" --reportportal --html=report.html && kill $prism_pid
-          deactivate
-        env:
-          ENDPOINTS: ${{ github.event.inputs.endpoints }}
-
-      - name: Upload Artifact
-        uses: actions/upload-artifact@v2
-        with:
-          name: report
-          path: |
-            openai-python/report.html
-            openai-python/assets
-            openai-python/prism.log
-
-      - name: Clean up
-        if: always()
-        run: |
-          rm -rf /tmp/jan
-          rm -rf openai-python
-          rm -rf report.html
-          rm -rf report.zip
-  
-
--- a/.github/workflows/jan-server-build-nightly.yml
+++ b/.github/workflows/jan-server-build-nightly.yml
@ -9,7 +9,7 @@ on:
      - 'README.md'
      - 'docs/**'
  schedule:
-    - cron: '0 20 * * 1,2,3' # At 8 PM UTC on Monday, Tuesday, and Wednesday which is 3 AM UTC+7 Tuesday, Wednesday, and Thursday
+    - cron: '0 21 * * 1,2,3' # At 8 PM UTC on Monday, Tuesday, and Wednesday which is 4 AM UTC+7 Tuesday, Wednesday, and Thursday
  workflow_dispatch:

 jobs:
--- a/.github/workflows/nightly-integrate-cortex-cpp.yml
+++ b/.github/workflows/nightly-integrate-cortex-cpp.yml
@ -0,0 +1,127 @@
+name: Nightly Update cortex cpp
+
+on:
+  schedule:
+    - cron: '30 19 * * 1-5' # At 01:30 on every day-of-week from Monday through Friday UTC +7
+  workflow_dispatch:
+
+jobs:
+  update-submodule:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+      actions: write
+
+    outputs:
+      pr_number: ${{ steps.check-update.outputs.pr_number }}
+      pr_created: ${{ steps.check-update.outputs.pr_created }}
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v3
+      with:
+        submodules: recursive
+        ref: dev
+        fetch-depth: 0
+        token: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+
+    - name: Configure Git
+      run: |
+        git config --global user.name 'github-actions[bot]'
+        git config --global user.email 'github-actions[bot]@users.noreply.github.com'
+
+    - name: Update submodule to latest release
+      id: check-update
+      env:
+        GITHUB_TOKEN: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+      run: |
+        curl -s https://api.github.com/repos/janhq/cortex/releases > /tmp/github_api_releases.json
+        latest_prerelease_name=$(cat /tmp/github_api_releases.json | jq -r '.[] | select(.prerelease) | .name' | head -n 1)
+
+        get_asset_count() {
+          local version_name=$1
+          cat /tmp/github_api_releases.json | jq -r --arg version_name "$version_name" '.[] | select(.name == $version_name) | .assets | length'
+        }
+
+        cortex_cpp_version_file_path="extensions/inference-nitro-extension/bin/version.txt"
+        current_version_name=$(cat "$cortex_cpp_version_file_path" | head -n 1)
+
+        current_version_asset_count=$(get_asset_count "$current_version_name")
+        latest_prerelease_asset_count=$(get_asset_count "$latest_prerelease_name")
+
+        if [ "$current_version_name" = "$latest_prerelease_name" ]; then
+          echo "cortex cpp remote repo doesn't have update today, skip update cortex-cpp for today nightly build"
+          echo "::set-output name=pr_created::false"
+          exit 0
+        fi
+
+        if [ "$current_version_asset_count" != "$latest_prerelease_asset_count" ]; then
+          echo "Latest prerelease version has different number of assets, somethink went wrong, skip update cortex-cpp for today nightly build"
+          echo "::set-output name=pr_created::false"
+          exit 1
+        fi
+
+        echo $latest_prerelease_name > $cortex_cpp_version_file_path
+        echo "Updated version from $current_version_name to $latest_prerelease_name."
+        echo "::set-output name=pr_created::true"
+
+        git add -f $cortex_cpp_version_file_path
+        git commit -m "Update cortex cpp nightly to version $latest_prerelease_name"
+        branch_name="update-nightly-$(date +'%Y-%m-%d-%H-%M')"
+        git checkout -b $branch_name
+        git push origin $branch_name
+
+        pr_title="Update cortex cpp nightly to version $latest_prerelease_name"
+        pr_body="This PR updates the Update cortex cpp nightly to version $latest_prerelease_name"
+
+        gh pr create --title "$pr_title" --body "$pr_body" --head $branch_name --base dev --reviewer Van-QA
+
+        pr_number=$(gh pr list --head $branch_name --json number --jq '.[0].number')
+        echo "::set-output name=pr_number::$pr_number"
+
+  check-and-merge-pr:
+    needs: update-submodule
+    if: needs.update-submodule.outputs.pr_created == 'true'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v3
+      with:
+        submodules: recursive
+        fetch-depth: 0
+        token: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+
+    - name: Wait for CI to pass
+      env:
+          GITHUB_TOKEN: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+      run: |
+        pr_number=${{ needs.update-submodule.outputs.pr_number }}
+        while true; do
+          ci_completed=$(gh pr checks $pr_number --json completedAt --jq '.[].completedAt')
+          if echo "$ci_completed" | grep -q "0001-01-01T00:00:00Z"; then
+            echo "CI is still running, waiting..."
+            sleep 60
+          else
+            echo "CI has completed, checking states..."
+            ci_states=$(gh pr checks $pr_number --json state --jq '.[].state')
+            if echo "$ci_states" | grep -vqE "SUCCESS|SKIPPED"; then
+              echo "CI failed, exiting..."
+              exit 1
+            else
+              echo "CI passed, merging PR..."
+              break
+            fi
+          fi
+        done
+
+    - name: Merge the PR
+      env:
+          GITHUB_TOKEN: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+      run: |
+        pr_number=${{ needs.update-submodule.outputs.pr_number }}
+        gh pr merge $pr_number --merge --admin
--- a/2
+++ b/2
@ -23,7 +23,7 @@ install-and-build: build-joi
 ifeq ($(OS),Windows_NT)
 	yarn config set network-timeout 300000
 endif
-	yarn global add turbo
+	yarn global add turbo@1.13.2
 	yarn build:core
 	yarn build:server
 	yarn install
--- a/electron/handlers/native.ts
+++ b/electron/handlers/native.ts
@ -1,4 +1,4 @@
-import { app, ipcMain, dialog, shell, nativeTheme, screen } from 'electron'
+import { app, ipcMain, dialog, shell, nativeTheme } from 'electron'
 import { join } from 'path'
 import { windowManager } from '../managers/window'
 import {
@ -41,12 +41,9 @@ export function handleAppIPCs() {
    windowManager.mainWindow?.minimize()
  })

-  ipcMain.handle(NativeRoute.setMaximizeApp, async () => {
+  ipcMain.handle(NativeRoute.setMaximizeApp, async (_event) => {
    if (windowManager.mainWindow?.isMaximized()) {
-      // const bounds = await getBounds()
-      // windowManager.mainWindow?.setSize(bounds.width, bounds.height)
-      // windowManager.mainWindow?.setPosition(Number(bounds.x), Number(bounds.y))
-      windowManager.mainWindow.restore()
+      windowManager.mainWindow.unmaximize()
    } else {
      windowManager.mainWindow?.maximize()
    }
--- a/electron/managers/mainWindowConfig.ts
+++ b/electron/managers/mainWindowConfig.ts
@ -4,11 +4,12 @@ export const mainWindowConfig: Electron.BrowserWindowConstructorOptions = {
  skipTaskbar: false,
  minWidth: DEFAULT_MIN_WIDTH,
  show: true,
+  transparent: true,
+  frame: false,
  titleBarStyle: 'hidden',
  vibrancy: 'fullscreen-ui',
  visualEffectState: 'active',
  backgroundMaterial: 'acrylic',
-  maximizable: false,
  autoHideMenuBar: true,
  trafficLightPosition: {
    x: 16,
--- a/extensions/inference-nitro-extension/bin/version.txt
+++ b/extensions/inference-nitro-extension/bin/version.txt
@ -1 +1 @@
-0.4.11
+0.4.13
--- a/extensions/inference-nitro-extension/package.json
+++ b/extensions/inference-nitro-extension/package.json
@ -1,7 +1,7 @@
 {
  "name": "@janhq/inference-cortex-extension",
  "productName": "Cortex Inference Engine",
-  "version": "1.0.11",
+  "version": "1.0.12",
  "description": "This extension embeds cortex.cpp, a lightweight inference engine written in C++. See https://nitro.jan.ai.\nAdditional dependencies could be installed to run without Cuda Toolkit installation.",
  "main": "dist/index.js",
  "node": "dist/node/index.cjs.js",
--- a/extensions/inference-nitro-extension/resources/models/qwen2-7b/model.json
+++ b/extensions/inference-nitro-extension/resources/models/qwen2-7b/model.json
@ -0,0 +1,36 @@
+{
+    "sources": [
+      {
+        "filename": "Qwen2-7B-Instruct-Q4_K_M.gguf",
+        "url": "https://huggingface.co/bartowski/Qwen2-7B-Instruct-GGUF/resolve/main/Qwen2-7B-Instruct-Q4_K_M.gguf"
+      }
+    ],
+    "id": "qwen2-7b",
+    "object": "model",
+    "name": "Qwen 2 Instruct 7B Q4",
+    "version": "1.0",
+    "description": "Qwen is optimized at Chinese, ideal for everyday tasks.",
+    "format": "gguf",
+    "settings": {
+      "ctx_len": 32768,
+      "prompt_template": "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant",
+      "llama_model_path": "Qwen2-7B-Instruct-Q4_K_M.gguf",
+      "ngl": 28
+    },
+    "parameters": {
+      "temperature": 0.7,
+      "top_p": 0.95,
+      "stream": true,
+      "max_tokens": 32768,
+      "stop": [],
+      "frequency_penalty": 0,
+      "presence_penalty": 0
+    },
+    "metadata": {
+      "author": "Alibaba",
+      "tags": ["7B", "Finetuned"],
+      "size": 4680000000
+    },
+    "engine": "nitro"
+  }
+  
--- a/extensions/inference-nitro-extension/rollup.config.ts
+++ b/extensions/inference-nitro-extension/rollup.config.ts
@ -39,6 +39,8 @@ const aya8bJson = require('./resources/models/aya-23-8b/model.json')
 const aya35bJson = require('./resources/models/aya-23-35b/model.json')
 const phimediumJson = require('./resources/models/phi3-medium/model.json')
 const codestralJson = require('./resources/models/codestral-22b/model.json')
+const qwen2Json = require('./resources/models/qwen2-7b/model.json')
+

 export default [
  {
@ -84,7 +86,8 @@ export default [
          phimediumJson,
          aya8bJson,
          aya35bJson,
-          codestralJson
+          codestralJson,
+          qwen2Json
        ]),
        NODE: JSON.stringify(`${packageJson.name}/${packageJson.node}`),
        DEFAULT_SETTINGS: JSON.stringify(defaultSettingJson),
--- a/extensions/inference-nvidia-extension/README.md
+++ b/extensions/inference-nvidia-extension/README.md
@ -0,0 +1,79 @@
+# Nvidia Engine Extension
+
+Created using Jan extension example
+
+# Create a Jan Extension using Typescript
+
+Use this template to bootstrap the creation of a TypeScript Jan extension. 🚀
+
+## Create Your Own Extension
+
+To create your own extension, you can use this repository as a template! Just follow the below instructions:
+
+1. Click the Use this template button at the top of the repository
+2. Select Create a new repository
+3. Select an owner and name for your new repository
+4. Click Create repository
+5. Clone your new repository
+
+## Initial Setup
+
+After you've cloned the repository to your local machine or codespace, you'll need to perform some initial setup steps before you can develop your extension.
+
+> [!NOTE]
+>
+> You'll need to have a reasonably modern version of
+> [Node.js](https://nodejs.org) handy. If you are using a version manager like
+> [`nodenv`](https://github.com/nodenv/nodenv) or
+> [`nvm`](https://github.com/nvm-sh/nvm), you can run `nodenv install` in the
+> root of your repository to install the version specified in
+> [`package.json`](./package.json). Otherwise, 20.x or later should work!
+
+1. :hammer_and_wrench: Install the dependencies
+
+   ```bash
+   npm install
+   ```
+
+1. :building_construction: Package the TypeScript for distribution
+
+   ```bash
+   npm run bundle
+   ```
+
+1. :white_check_mark: Check your artifact
+
+   There will be a tgz file in your extension directory now
+
+## Update the Extension Metadata
+
+The [`package.json`](package.json) file defines metadata about your extension, such as
+extension name, main entry, description and version.
+
+When you copy this repository, update `package.json` with the name, description for your extension.
+
+## Update the Extension Code
+
+The [`src/`](./src/) directory is the heart of your extension! This contains the
+source code that will be run when your extension functions are invoked. You can replace the
+contents of this directory with your own code.
+
+There are a few things to keep in mind when writing your extension code:
+
+- Most Jan Extension functions are processed asynchronously.
+  In `index.ts`, you will see that the extension function will return a `Promise<any>`.
+
+  ```typescript
+  import { events, MessageEvent, MessageRequest } from '@janhq/core'
+
+  function onStart(): Promise<any> {
+    return events.on(MessageEvent.OnMessageSent, (data: MessageRequest) =>
+      this.inference(data)
+    )
+  }
+  ```
+
+  For more information about the Jan Extension Core module, see the
+  [documentation](https://github.com/janhq/jan/blob/main/core/README.md).
+
+So, what are you waiting for? Go ahead and start customizing your extension!
--- a/extensions/inference-nvidia-extension/package.json
+++ b/extensions/inference-nvidia-extension/package.json
@ -0,0 +1,43 @@
+{
+  "name": "@janhq/inference-nvidia-extension",
+  "productName": "NVIDIA NIM Inference Engine",
+  "version": "1.0.1",
+  "description": "This extension enables NVIDIA chat completion API calls",
+  "main": "dist/index.js",
+  "module": "dist/module.js",
+  "engine": "nvidia",
+  "author": "Jan <service@jan.ai>",
+  "license": "AGPL-3.0",
+  "scripts": {
+    "build": "tsc -b . && webpack --config webpack.config.js",
+    "build:publish": "rimraf *.tgz --glob && yarn build && npm pack && cpx *.tgz ../../pre-install"
+  },
+  "exports": {
+    ".": "./dist/index.js",
+    "./main": "./dist/module.js"
+  },
+  "devDependencies": {
+    "cpx": "^1.5.0",
+    "rimraf": "^3.0.2",
+    "webpack": "^5.88.2",
+    "webpack-cli": "^5.1.4",
+    "ts-loader": "^9.5.0"
+  },
+  "dependencies": {
+    "@janhq/core": "file:../../core",
+    "fetch-retry": "^5.0.6",
+    "path-browserify": "^1.0.1",
+    "ulidx": "^2.3.0"
+  },
+  "engines": {
+    "node": ">=18.0.0"
+  },
+  "files": [
+    "dist/*",
+    "package.json",
+    "README.md"
+  ],
+  "bundleDependencies": [
+    "fetch-retry"
+  ]
+}
--- a/extensions/inference-nvidia-extension/resources/models.json
+++ b/extensions/inference-nvidia-extension/resources/models.json
@ -0,0 +1,31 @@
+[
+  {
+    "sources": [
+      {
+        "url": "https://integrate.api.nvidia.com/v1/chat/completions"
+      }
+    ],
+    "id": "mistralai/mistral-7b-instruct-v0.2",
+    "object": "model",
+    "name": "Mistral 7B",
+    "version": "1.1",
+    "description": "Mistral 7B with NVIDIA",
+    "format": "api",
+    "settings": {},
+    "parameters": {
+      "max_tokens": 1024,
+      "temperature": 0.3,
+      "top_p": 1,
+      "stream": false,
+      "frequency_penalty": 0,
+      "presence_penalty": 0,
+      "stop": null,
+      "seed": null
+    },
+    "metadata": {
+      "author": "NVIDIA",
+      "tags": ["General"]
+    },
+    "engine": "nvidia"
+  }
+]
--- a/extensions/inference-nvidia-extension/resources/settings.json
+++ b/extensions/inference-nvidia-extension/resources/settings.json
@ -0,0 +1,24 @@
+[
+  {
+    "key": "chat-completions-endpoint",
+    "title": "Chat Completions Endpoint",
+    "description": "The endpoint to use for chat completions. See the [NVIDIA API documentation](https://www.nvidia.com/en-us/ai/) for more information.",
+    "controllerType": "input",
+    "controllerProps": {
+      "placeholder": "https://integrate.api.nvidia.com/v1/chat/completions",
+      "value": "https://integrate.api.nvidia.com/v1/chat/completions"
+    }
+  },
+  {
+    "key": "nvidia-api-key",
+    "title": "API Key",
+    "description": "The NVIDIA API uses API keys for authentication. Visit your [API Keys](https://org.ngc.nvidia.com/setup/personal-keys) page to retrieve the API key you'll use in your requests..",
+    "controllerType": "input",
+    "controllerProps": {
+      "placeholder": "nvapi-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
+      "value": "",
+      "type": "password",
+      "inputActions": ["unobscure", "copy"]
+    }
+  }
+]
--- a/extensions/inference-nvidia-extension/src/index.ts
+++ b/extensions/inference-nvidia-extension/src/index.ts
@ -0,0 +1,66 @@
+/**
+ * @file This file exports a class that implements the InferenceExtension interface from the @janhq/core package.
+ * The class provides methods for initializing and stopping a model, and for making inference requests.
+ * It also subscribes to events emitted by the @janhq/core package and handles new message requests.
+ * @version 1.0.0
+ * @module inference-mistral-extension/src/index
+ */
+
+import { RemoteOAIEngine } from '@janhq/core'
+
+declare const SETTINGS: Array<any>
+declare const MODELS: Array<any>
+
+enum Settings {
+  apiKey = 'nvidia-api-key',
+  chatCompletionsEndPoint = 'chat-completions-endpoint',
+}
+/**
+ * A class that implements the InferenceExtension interface from the @janhq/core package.
+ * The class provides methods for initializing and stopping a model, and for making inference requests.
+ * It also subscribes to events emitted by the @janhq/core package and handles new message requests.
+ */
+export default class JanNVIDIANIMInferenceEngine extends RemoteOAIEngine {
+  inferenceUrl: string = ''
+  provider: string = 'nvidia'
+
+  override async onLoad(): Promise<void> {
+    super.onLoad()
+
+    // Register Settings
+    this.registerSettings(SETTINGS)
+    this.registerModels(MODELS)
+
+    this.apiKey = await this.getSetting<string>(Settings.apiKey, '')
+    this.inferenceUrl = await this.getSetting<string>(
+      Settings.chatCompletionsEndPoint,
+      ''
+    )
+
+    if (this.inferenceUrl.length === 0) {
+      SETTINGS.forEach((setting) => {
+        if (setting.key === Settings.chatCompletionsEndPoint) {
+          this.inferenceUrl = setting.controllerProps.value as string
+        }
+      })
+    }
+  }
+
+  onSettingUpdate<T>(key: string, value: T): void {
+    if (key === Settings.apiKey) {
+      this.apiKey = value as string
+    } else if (key === Settings.chatCompletionsEndPoint) {
+      if (typeof value !== 'string') return
+
+      if (value.trim().length === 0) {
+        SETTINGS.forEach((setting) => {
+          if (setting.key === Settings.chatCompletionsEndPoint) {
+            this.inferenceUrl = setting.controllerProps.value as string
+          }
+        })
+      } else {
+        this.inferenceUrl = value
+      }
+    }
+  }
+}
--- a/extensions/inference-nvidia-extension/tsconfig.json
+++ b/extensions/inference-nvidia-extension/tsconfig.json
@ -0,0 +1,14 @@
+{
+  "compilerOptions": {
+    "target": "es2016",
+    "module": "ES6",
+    "moduleResolution": "node",
+    "outDir": "./dist",
+    "esModuleInterop": true,
+    "forceConsistentCasingInFileNames": true,
+    "strict": false,
+    "skipLibCheck": true,
+    "rootDir": "./src"
+  },
+  "include": ["./src"]
+}
--- a/extensions/inference-nvidia-extension/webpack.config.js
+++ b/extensions/inference-nvidia-extension/webpack.config.js
@ -0,0 +1,42 @@
+const path = require('path')
+const webpack = require('webpack')
+const packageJson = require('./package.json')
+const settingJson = require('./resources/settings.json')
+const modelsJson = require('./resources/models.json')
+
+module.exports = {
+  experiments: { outputModule: true },
+  entry: './src/index.ts', // Adjust the entry point to match your project's main file
+  mode: 'production',
+  module: {
+    rules: [
+      {
+        test: /\.tsx?$/,
+        use: 'ts-loader',
+        exclude: /node_modules/,
+      },
+    ],
+  },
+  plugins: [
+    new webpack.DefinePlugin({
+      SETTINGS: JSON.stringify(settingJson),
+      ENGINE: JSON.stringify(packageJson.engine),
+      MODELS: JSON.stringify(modelsJson),
+    }),
+  ],
+  output: {
+    filename: 'index.js', // Adjust the output file name as needed
+    path: path.resolve(__dirname, 'dist'),
+    library: { type: 'module' }, // Specify ESM output format
+  },
+  resolve: {
+    extensions: ['.ts', '.js'],
+    fallback: {
+      path: require.resolve('path-browserify'),
+    },
+  },
+  optimization: {
+    minimize: false,
+  },
+  // Add loaders and other configuration as needed for your project
+}
--- a/extensions/model-extension/package.json
+++ b/extensions/model-extension/package.json
@ -1,7 +1,7 @@
 {
  "name": "@janhq/model-extension",
  "productName": "Model Management",
-  "version": "1.0.31",
+  "version": "1.0.32",
  "description": "Model Management Extension provides model exploration and seamless downloads",
  "main": "dist/index.js",
  "node": "dist/node/index.cjs.js",
--- a/extensions/model-extension/src/index.ts
+++ b/extensions/model-extension/src/index.ts
@ -417,6 +417,30 @@ export default class JanModelExtension extends ModelExtension {
    )
  }

+  private async getModelJsonPath(
+    folderFullPath: string
+  ): Promise<string | undefined> {
+    // try to find model.json recursively inside each folder
+    if (!(await fs.existsSync(folderFullPath))) return undefined
+    const files: string[] = await fs.readdirSync(folderFullPath)
+    if (files.length === 0) return undefined
+    if (files.includes(JanModelExtension._modelMetadataFileName)) {
+      return joinPath([
+        folderFullPath,
+        JanModelExtension._modelMetadataFileName,
+      ])
+    }
+    // continue recursive
+    for (const file of files) {
+      const path = await joinPath([folderFullPath, file])
+      const fileStats = await fs.fileStat(path)
+      if (fileStats.isDirectory) {
+        const result = await this.getModelJsonPath(path)
+        if (result) return result
+      }
+    }
+  }
+
  private async getModelsMetadata(
    selector?: (path: string, model: Model) => Promise<boolean>
  ): Promise<Model[]> {
@ -438,11 +462,11 @@ export default class JanModelExtension extends ModelExtension {
      const readJsonPromises = allDirectories.map(async (dirName) => {
        // filter out directories that don't match the selector
        // read model.json
-        const jsonPath = await joinPath([
+        const folderFullPath = await joinPath([
          JanModelExtension._homeDir,
          dirName,
-          JanModelExtension._modelMetadataFileName,
        ])
+        const jsonPath = await this.getModelJsonPath(folderFullPath)

        if (await fs.existsSync(jsonPath)) {
          // if we have the model.json file, read it
--- a/web/containers/Layout/TopPanel/index.tsx
+++ b/web/containers/Layout/TopPanel/index.tsx
@ -9,6 +9,7 @@ import {
  PanelRightCloseIcon,
  MinusIcon,
  MenuIcon,
+  SquareIcon,
  PaletteIcon,
  XIcon,
 } from 'lucide-react'
@ -51,7 +52,7 @@ const TopPanel = () => {
            <Button
              theme="icon"
              onClick={() => {
-                window?.electronAPI.showOpenMenu(100, 100)
+                window?.electronAPI?.showOpenMenu(100, 100)
              }}
            >
              <MenuIcon size={16} />
@ -96,17 +97,23 @@ const TopPanel = () => {
            <PaletteIcon size={16} className="cursor-pointer" />
          </Button>

-          {isWindows && (
+          {!isMac && (
            <div className="flex items-center gap-x-2">
              <Button
                theme="icon"
-                onClick={() => window?.electronAPI.setMinimizeApp()}
+                onClick={() => window?.electronAPI?.setMinimizeApp()}
              >
                <MinusIcon size={16} />
              </Button>
              <Button
                theme="icon"
-                onClick={() => window?.electronAPI.setCloseApp()}
+                onClick={() => window?.electronAPI?.setMaximizeApp()}
+              >
+                <SquareIcon size={14} />
+              </Button>
+              <Button
+                theme="icon"
+                onClick={() => window?.electronAPI?.setCloseApp()}
              >
                <XIcon size={16} />
              </Button>
--- a/web/containers/Loader/ProgressCircle.tsx
+++ b/web/containers/Loader/ProgressCircle.tsx
@ -0,0 +1,50 @@
+import React from 'react'
+
+interface ProgressCircleProps {
+  percentage: number
+  size?: number
+  strokeWidth?: number
+}
+
+const ProgressCircle: React.FC<ProgressCircleProps> = ({
+  percentage,
+  size = 100,
+  strokeWidth = 14,
+}) => {
+  const radius = (size - strokeWidth) / 2
+  const circumference = 2 * Math.PI * radius
+  const offset = circumference - (percentage / 100) * circumference
+
+  return (
+    <svg
+      className="ml-0.5 h-4 w-4 rotate-[-90deg] transform text-[hsla(var(--primary-bg))]"
+      height={size}
+      width={size}
+      xmlns="http://www.w3.org/2000/svg"
+      viewBox={`0 0 ${size} ${size}`}
+    >
+      <circle
+        className="opacity-25"
+        cx={size / 2}
+        cy={size / 2}
+        r={radius}
+        stroke="currentColor"
+        strokeWidth={strokeWidth}
+        fill="none"
+      ></circle>
+      <circle
+        className="transition-stroke-dashoffset duration-300"
+        cx={size / 2}
+        cy={size / 2}
+        r={radius}
+        stroke="currentColor"
+        strokeWidth={strokeWidth}
+        fill="none"
+        strokeDasharray={circumference}
+        strokeDashoffset={offset}
+      ></circle>
+    </svg>
+  )
+}
+
+export default ProgressCircle
--- a/web/containers/ModelDropdown/index.tsx
+++ b/web/containers/ModelDropdown/index.tsx
@ -8,16 +8,19 @@ import { useAtom, useAtomValue, useSetAtom } from 'jotai'
 import { ChevronDownIcon, DownloadCloudIcon, XIcon } from 'lucide-react'
 import { twMerge } from 'tailwind-merge'

+import ProgressCircle from '@/containers/Loader/ProgressCircle'
+
 import ModelLabel from '@/containers/ModelLabel'

 import SetupRemoteModel from '@/containers/SetupRemoteModel'

 import useDownloadModel from '@/hooks/useDownloadModel'
+import { modelDownloadStateAtom } from '@/hooks/useDownloadState'
 import useRecommendedModel from '@/hooks/useRecommendedModel'

 import useUpdateModelParameters from '@/hooks/useUpdateModelParameters'

-import { toGibibytes } from '@/utils/converter'
+import { formatDownloadPercentage, toGibibytes } from '@/utils/converter'

 import { extensionManager } from '@/extension'

@ -64,6 +67,7 @@ const ModelDropdown = ({
  const [dropdownOptions, setDropdownOptions] = useState<HTMLDivElement | null>(
    null
  )
+  const downloadStates = useAtomValue(modelDownloadStateAtom)
  const setThreadModelParams = useSetAtom(setThreadModelParamsAtom)
  const { updateModelParameter } = useUpdateModelParameters()

@ -277,8 +281,8 @@ const ModelDropdown = ({
                className="h-6 gap-1 px-2"
                options={[
                  { name: 'All', value: 'all' },
-                  { name: 'Local', value: 'local' },
-                  { name: 'Remote', value: 'remote' },
+                  { name: 'On-device', value: 'local' },
+                  { name: 'Cloud', value: 'remote' },
                ]}
                onValueChange={(value) => setSearchFilter(value)}
                onOpenChange={(open) => setFilterOptionsOpen(open)}
@ -351,12 +355,29 @@ const ModelDropdown = ({
                                  <span className="font-medium">
                                    {toGibibytes(model.metadata.size)}
                                  </span>
-                                  {!isDownloading && (
+                                  {!isDownloading ? (
                                    <DownloadCloudIcon
                                      size={18}
                                      className="cursor-pointer text-[hsla(var(--app-link))]"
                                      onClick={() => downloadModel(model)}
                                    />
+                                  ) : (
+                                    Object.values(downloadStates)
+                                      .filter((x) => x.modelId === model.id)
+                                      .map((item) => (
+                                        <ProgressCircle
+                                          key={item.modelId}
+                                          percentage={
+                                            formatDownloadPercentage(
+                                              item?.percent,
+                                              {
+                                                hidePercentage: true,
+                                              }
+                                            ) as number
+                                          }
+                                          size={100}
+                                        />
+                                      ))
                                  )}
                                </div>
                              </li>
@ -397,12 +418,29 @@ const ModelDropdown = ({
                                    <span className="font-medium">
                                      {toGibibytes(model.metadata.size)}
                                    </span>
-                                    {!isDownloading && (
+                                    {!isDownloading ? (
                                      <DownloadCloudIcon
                                        size={18}
                                        className="cursor-pointer text-[hsla(var(--app-link))]"
                                        onClick={() => downloadModel(model)}
                                      />
+                                    ) : (
+                                      Object.values(downloadStates)
+                                        .filter((x) => x.modelId === model.id)
+                                        .map((item) => (
+                                          <ProgressCircle
+                                            key={item.modelId}
+                                            percentage={
+                                              formatDownloadPercentage(
+                                                item?.percent,
+                                                {
+                                                  hidePercentage: true,
+                                                }
+                                              ) as number
+                                            }
+                                            size={100}
+                                          />
+                                        ))
                                    )}
                                  </div>
                                </li>
--- a/web/containers/ModelLabel/NotEnoughMemoryLabel.tsx
+++ b/web/containers/ModelLabel/NotEnoughMemoryLabel.tsx
@ -1,32 +1,49 @@
-import { memo } from 'react'
+import { Fragment, memo } from 'react'

 import { Badge, Tooltip } from '@janhq/joi'
-import { InfoIcon } from 'lucide-react'
-import { twMerge } from 'tailwind-merge'
+import { AlertTriangleIcon, InfoIcon } from 'lucide-react'

 type Props = {
  compact?: boolean
  unit: string
 }

+const tooltipContent = `Your device doesn't have enough RAM to run this model. Consider upgrading your RAM or using a device with more memory capacity.`
+
 const NotEnoughMemoryLabel = ({ unit, compact }: Props) => (
-  <Badge
-    theme="destructive"
-    variant="soft"
-    className={twMerge(compact && 'h-5 w-5 p-1')}
-  >
-    {!compact && <span className="line-clamp-1">Not enough {unit}</span>}
-    <Tooltip
-      trigger={
-        compact ? (
-          <div className="h-2 w-2 cursor-pointer rounded-full bg-[hsla(var(--destructive-bg))]" />
-        ) : (
-          <InfoIcon size={14} className="ml-2 flex-shrink-0 cursor-pointer" />
-        )
-      }
-      content="This tag signals insufficient RAM for optimal model performance. It's dynamic and may change with your system's RAM availability."
-    />
-  </Badge>
+  <>
+    {compact ? (
+      <div className="flex h-5 w-5 items-center">
+        <Tooltip
+          trigger={
+            <AlertTriangleIcon
+              size={14}
+              className="cursor-pointer text-[hsla(var(--destructive-bg))]"
+            />
+          }
+          content={
+            <Fragment>
+              <b>Not enough RAM:</b> <span>{tooltipContent}</span>
+            </Fragment>
+          }
+        />
+      </div>
+    ) : (
+      <Badge theme="destructive" variant="soft">
+        <span className="line-clamp-1">Not enough {unit}</span>
+        <Tooltip
+          trigger={
+            <InfoIcon size={14} className="ml-2 flex-shrink-0 cursor-pointer" />
+          }
+          content={
+            <Fragment>
+              <b>Not enough RAM:</b> <span>{tooltipContent}</span>
+            </Fragment>
+          }
+        />
+      </Badge>
+    )}
+  </>
 )

 export default memo(NotEnoughMemoryLabel)
--- a/web/containers/ModelLabel/SlowOnYourDeviceLabel.tsx
+++ b/web/containers/ModelLabel/SlowOnYourDeviceLabel.tsx
@ -1,32 +1,49 @@
-import { memo } from 'react'
+import { Fragment, memo } from 'react'

 import { Badge, Tooltip } from '@janhq/joi'

-import { InfoIcon } from 'lucide-react'
-import { twMerge } from 'tailwind-merge'
+import { AlertTriangleIcon, InfoIcon } from 'lucide-react'

 type Props = {
  compact?: boolean
 }

+const tooltipContent = `Your device may be running low on available RAM, which can affect the speed of this model. Try closing any unnecessary applications to free up system memory.`
+
 const SlowOnYourDeviceLabel = ({ compact }: Props) => (
-  <Badge
-    theme="warning"
-    variant="soft"
-    className={twMerge(compact && 'h-5 w-5 p-1')}
-  >
-    {!compact && <span className="line-clamp-1">Slow on your device</span>}
-    <Tooltip
-      trigger={
-        compact ? (
-          <div className="h-2 w-2 cursor-pointer rounded-full bg-[hsla(var(--warning-bg))] p-0" />
-        ) : (
-          <InfoIcon size={14} className="ml-2 flex-shrink-0 cursor-pointer" />
-        )
-      }
-      content="This tag indicates that your current RAM performance may affect model speed. It can change based on other active apps. To improve, consider closing unnecessary applications to free up RAM."
-    />
-  </Badge>
+  <>
+    {compact ? (
+      <div className="flex h-5 w-5 items-center">
+        <Tooltip
+          trigger={
+            <AlertTriangleIcon
+              size={14}
+              className="cursor-pointer text-[hsla(var(--warning-bg))]"
+            />
+          }
+          content={
+            <Fragment>
+              <b>Slow on your device:</b> <span>{tooltipContent}</span>
+            </Fragment>
+          }
+        />
+      </div>
+    ) : (
+      <Badge theme="warning" variant="soft">
+        <span className="line-clamp-1">Slow on your device</span>
+        <Tooltip
+          trigger={
+            <InfoIcon size={14} className="ml-2 flex-shrink-0 cursor-pointer" />
+          }
+          content={
+            <Fragment>
+              <b>Slow on your device:</b> <span>{tooltipContent}</span>
+            </Fragment>
+          }
+        />
+      </Badge>
+    )}
+  </>
 )

 export default memo(SlowOnYourDeviceLabel)
--- a/web/containers/Providers/KeyListener.tsx
+++ b/web/containers/Providers/KeyListener.tsx
@ -8,7 +8,11 @@ import { MainViewState } from '@/constants/screens'

 import { useCreateNewThread } from '@/hooks/useCreateNewThread'

-import { mainViewStateAtom, showLeftPanelAtom } from '@/helpers/atoms/App.atom'
+import {
+  mainViewStateAtom,
+  showLeftPanelAtom,
+  showRightPanelAtom,
+} from '@/helpers/atoms/App.atom'
 import { assistantsAtom } from '@/helpers/atoms/Assistant.atom'

 type Props = {
@ -17,6 +21,7 @@ type Props = {

 export default function KeyListener({ children }: Props) {
  const setShowLeftPanel = useSetAtom(showLeftPanelAtom)
+  const setShowRightPanel = useSetAtom(showRightPanelAtom)
  const setMainViewState = useSetAtom(mainViewStateAtom)
  const { requestCreateNewThread } = useCreateNewThread()
  const assistants = useAtomValue(assistantsAtom)
@ -25,6 +30,11 @@ export default function KeyListener({ children }: Props) {
    const onKeyDown = (e: KeyboardEvent) => {
      const prefixKey = isMac ? e.metaKey : e.ctrlKey

+      if (e.key === 'b' && prefixKey && e.shiftKey) {
+        setShowRightPanel((showRightideBar) => !showRightideBar)
+        return
+      }
+
      if (e.key === 'n' && prefixKey) {
        requestCreateNewThread(assistants[0])
        setMainViewState(MainViewState.Thread)
@ -43,7 +53,13 @@ export default function KeyListener({ children }: Props) {
    }
    document.addEventListener('keydown', onKeyDown)
    return () => document.removeEventListener('keydown', onKeyDown)
-  }, [assistants, requestCreateNewThread, setMainViewState, setShowLeftPanel])
+  }, [
+    assistants,
+    requestCreateNewThread,
+    setMainViewState,
+    setShowLeftPanel,
+    setShowRightPanel,
+  ])

  return <Fragment>{children}</Fragment>
 }
--- a/web/hooks/useLoadTheme.ts
+++ b/web/hooks/useLoadTheme.ts
@ -29,11 +29,11 @@ export const useLoadTheme = async () => {
  const setNativeTheme = useCallback(
    (nativeTheme: NativeThemeProps) => {
      if (nativeTheme === 'dark') {
-        window?.electronAPI.setNativeThemeDark()
+        window?.electronAPI?.setNativeThemeDark()
        setTheme('dark')
        localStorage.setItem('nativeTheme', 'dark')
      } else {
-        window?.electronAPI.setNativeThemeLight()
+        window?.electronAPI?.setNativeThemeLight()
        setTheme('light')
        localStorage.setItem('nativeTheme', 'light')
      }
--- a/web/hooks/useSendChatMessage.ts
+++ b/web/hooks/useSendChatMessage.ts
@ -10,6 +10,7 @@ import {
  ConversationalExtension,
  EngineManager,
  ToolManager,
+  ChatCompletionMessage,
 } from '@janhq/core'
 import { atom, useAtom, useAtomValue, useSetAtom } from 'jotai'

@ -19,6 +20,7 @@ import {
  fileUploadAtom,
 } from '@/containers/Providers/Jotai'

+import { Stack } from '@/utils/Stack'
 import { compressImage, getBase64 } from '@/utils/base64'
 import { MessageRequestBuilder } from '@/utils/messageRequestBuilder'
 import { toRuntimeParams, toSettingParams } from '@/utils/modelParam'
@ -90,6 +92,33 @@ export default function useSendChatMessage() {
    selectedModelRef.current = selectedModel
  }, [selectedModel])

+  const normalizeMessages = (
+    messages: ChatCompletionMessage[]
+  ): ChatCompletionMessage[] => {
+    const stack = new Stack<ChatCompletionMessage>()
+    for (const message of messages) {
+      if (stack.isEmpty()) {
+        stack.push(message)
+        continue
+      }
+      const topMessage = stack.peek()
+
+      if (message.role === topMessage.role) {
+        // add an empty message
+        stack.push({
+          role:
+            topMessage.role === ChatCompletionRole.User
+              ? ChatCompletionRole.Assistant
+              : ChatCompletionRole.User,
+          content: '.', // some model requires not empty message
+        })
+      }
+      stack.push(message)
+    }
+
+    return stack.reverseOutput()
+  }
+
  const resendChatMessage = async (currentMessage: ThreadMessage) => {
    if (!activeThreadRef.current) {
      console.error('No active thread')
@ -140,6 +169,8 @@ export default function useSendChatMessage() {
      ) ?? []
    )

+    request.messages = normalizeMessages(request.messages ?? [])
+
    const engine =
      requestBuilder.model?.engine ?? selectedModelRef.current?.engine ?? ''

@ -258,6 +289,7 @@ export default function useSendChatMessage() {
        (assistant) => assistant.tools ?? []
      ) ?? []
    )
+    request.messages = normalizeMessages(request.messages ?? [])

    // Request for inference
    EngineManager.instance()
--- a/web/screens/Settings/Hotkeys/index.tsx
+++ b/web/screens/Settings/Hotkeys/index.tsx
@ -11,6 +11,11 @@ const availableHotkeys = [
    modifierKeys: [isMac ? '⌘' : 'Ctrl'],
    description: 'Toggle collapsible left panel',
  },
+  {
+    combination: 'Shift B',
+    modifierKeys: [isMac ? '⌘' : 'Ctrl'],
+    description: 'Toggle collapsible right panel',
+  },
  {
    combination: ',',
    modifierKeys: [isMac ? '⌘' : 'Ctrl'],
@ -21,7 +26,7 @@ const availableHotkeys = [
    description: 'Send a message',
  },
  {
-    combination: 'Shift + Enter',
+    combination: 'Shift Enter',
    description: 'Insert new line in input box',
  },
  {
--- a/web/screens/Thread/ThreadCenterPanel/SimpleTextMessage/index.tsx
+++ b/web/screens/Thread/ThreadCenterPanel/SimpleTextMessage/index.tsx
@ -268,7 +268,7 @@ const SimpleTextMessage: React.FC<ThreadMessage> = (props) => {
              ) : (
                <div
                  className={twMerge(
-                    'message flex flex-col gap-y-2 font-medium leading-relaxed',
+                    'message flex flex-col gap-y-2 leading-relaxed',
                    isUser ? 'whitespace-pre-wrap break-words' : 'p-4'
                  )}
                >
@ -279,7 +279,7 @@ const SimpleTextMessage: React.FC<ThreadMessage> = (props) => {
          ) : (
            <div
              className={twMerge(
-                'message max-width-[100%] flex flex-col gap-y-2 overflow-auto font-medium leading-relaxed',
+                'message max-width-[100%] flex flex-col gap-y-2 overflow-auto leading-relaxed',
                isUser && 'whitespace-pre-wrap break-words'
              )}
              dangerouslySetInnerHTML={{ __html: parsedText }}
--- a/web/utils/Stack.ts
+++ b/web/utils/Stack.ts
@ -0,0 +1,31 @@
+export class Stack<T> {
+  private array: T[] = []
+
+  pop(): T | undefined {
+    if (this.isEmpty()) throw new Error()
+
+    return this.array.pop()
+  }
+
+  push(data: T): void {
+    this.array.push(data)
+  }
+
+  peek(): T {
+    if (this.isEmpty()) throw new Error()
+
+    return this.array[this.array.length - 1]
+  }
+
+  isEmpty(): boolean {
+    return this.array.length === 0
+  }
+
+  size(): number {
+    return this.array.length
+  }
+
+  reverseOutput(): T[] {
+    return [...this.array]
+  }
+}
 @ -1 +1 @@
 .4.11
 .4.13