Release SDS-1-0-2

-Seed Instruction Upload via JSON (List Format) —Users can now upload multiple seed instructions as a list within a JSON file, replacing the earlier manual entry process. This significantly reduces time and effort, especially when working with large or complex datasets.

-Dynamic Rendering of Seed Instructions
—Upon uploading a JSON file, the seed instructions are now automatically rendered in the UI, allowing users to review the pre-filled data in real time.

-Flexible Seed Management Users still retain full control over the seed data — they can edit, remove, or add new instructions directly from the interface even after uploading a file.

-Enhanced Prompt Generation with AI Assistance —The Generate Prompt feature has been upgraded with:
A new AI Assistant icon to generate intelligent prompts based on the uploaded examples and instructions.

—A Restore icon that enables users to revert to their previously generated or edited prompt, offering more flexibility and iteration during prompt crafting.

-Schema-aware Prompt Synthesis
—The AI-assisted prompt generation leverages user-provided examples and instructions to produce a context-aware prompt. This ensures that the generated prompt is well-aligned with the schema of the sample data, improving the relevance and consistency of the generated output.

—Stability Improvements and Bug Fixes Addressed and resolved intermittent errors experienced during example uploads. These fixes enhance overall reliability and provide a smoother, uninterrupted user experience.

Author: Khauneesh-AI

.github/workflows/python-app.yml

@@ -14,25 +14,30 @@ jobs:
         python-version: ["3.10"]
 
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
 
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.python-version }}
 
-    - name: Install dependencies
+    # ---------- install uv -------------------------------------------------
+    - name: Install uv (fast dependency resolver)
       run: |
-        python -m pip install --upgrade pip
-        pip install pytest pytest-asyncio pytest-cov requests-mock boto3 moto
-        pip install -r requirements.txt
+        python -m pip install --upgrade pip uv
 
+    # ---------- reproduce locked env --------------------------------------
+    - name: Sync project dependencies from uv.lock
+      run: |
+        uv sync
+    
+    # ---------- run test suite -----------------------------------
     - name: Run tests and generate coverage report
       run: |
-        pytest tests/ --cov=app --cov-report=xml
+        uv run pytest tests/ --cov=app --cov-report=xml
 
     - name: Upload coverage report as an artifact
       uses: actions/upload-artifact@v4
       with:
         name: coverage-report
-        path: coverage.xml
+        path: coverage.xml

.gitignore

@@ -74,4 +74,13 @@ app/frontend/
 app/launch_streamlit.py
 
 # Virtual environment
-.venv/
+.venv/freeform_data_claude_*
+row_data_claude_*
+freeform_data_claude_*
+housing_example.json
+seeds_test.json
+test.csv
+sample_200x100.csv
+Raw_Web_Visit_sample.csv
+Raw_Web_Visit_Sample.csv
+Raw_Web_Visit_Sample.csv

.project-metadata.yaml

@@ -11,30 +11,37 @@ date: "2024-10-30"
 
 environment_variables:
   AWS_ACCESS_KEY_ID:
-    default: "your access key"
+    default: null
     description: >-
       AWS Key ID. Check the Amazon Bedrock documentation for information about role access
   AWS_SECRET_ACCESS_KEY:
-    default: "your secret access key"
+    default: null
     description: >-
       AWS Secret Key
   AWS_DEFAULT_REGION:
     default: "us-east-1"
     description: >-
       AWS region
   hf_token:
-    default: "hf_token"
+    default: null
     description: >-
       your huggingface token
   hf_username:
-    default: "hf_username"
+    default: null
     description: >-
       your huggingface username
   CDP_TOKEN:
     default: null
     description: >-
      API key for Cloudera AI Inference
-
+  OPENAI_API_KEY:
+    default: null
+    description: >-
+      OpenAI API Key. Check the OpenAI documentation for information about role access
+  GEMINI_API_KEY:
+    default: null
+    description: >-
+      Gemini API Key. Check the Google Gemini documentation for information about role access
 # runtimes
 runtimes:
   - editor: JupyterLab

app/client/.eslintrc.json

@@ -0,0 +1,5 @@
+{
+    "rules": {
+      "@typescript-eslint/no-explicit-any": "warn"
+    }
+  }

app/client/src/Container.tsx

@@ -55,6 +55,15 @@ const StyledText = styled.div`
   font-weight: normal;
 `;
 
+const StyledSection = styled.div`
+  margin-left: 24px;
+  color: #adb2b6;
+  flex-grow: 1;
+  font-family: Roboto Mono, monospace;
+  font-size: 12px;
+  overflow: hidden;
+`;
+
 const pages: MenuItem[] = [
   {
     key: Pages.HOME,
@@ -96,6 +105,11 @@ const pages: MenuItem[] = [
               <span style={{ marginLeft: '4px' }}>Join the discussion on GitHub</span>
             </a>
           </div>
+          <Flex justify='center' align='center' vertical style={{ marginTop: '12px' }}>
+            <StyledSection>
+              {`SDS-1.0.2`}
+            </StyledSection>
+          </Flex>
           </Flex>
           <br/>
         </div>

app/client/src/pages/DataGenerator/CustomPromptButton.tsx

@@ -5,6 +5,7 @@ import styled from "styled-components";
 import { fetchCustomPrompt } from "./hooks";
 import Loading from "../Evaluator/Loading";
 import AiAssistantIcon from "./AiAssistantIcon";
+import { isEmpty } from "lodash";
 
 interface Props {
     model_id: string;
@@ -56,11 +57,18 @@ const StyledIcon = styled.div`
 const CustomPromptButton: React.FC<Props> = ({ model_id, inference_type, caii_endpoint, use_case, example_path, setPrompt }) => {
   const [form] = Form.useForm();
   const [showModal, setShowModal] = useState(false);
+  const [disabled, setDisabled] = useState(false);
+  const custom_prompt_instructions = Form.useWatch('custom_prompt_instructions', { form, preserve: true });
+  console.log('custom_prompt_instructions', custom_prompt_instructions);
 
   const mutation = useMutation({
     mutationFn: fetchCustomPrompt
   });
 
+  useEffect(() => {
+    setDisabled(isEmpty(custom_prompt_instructions));
+  }, [custom_prompt_instructions]);
+
   useEffect(() => {
       if (mutation.isError) {
           notification.error({
@@ -112,12 +120,13 @@ const CustomPromptButton: React.FC<Props> = ({ model_id, inference_type, caii_en
               okText={`Generate`}
               title={`Generate Custom Prompt`}
               onClose={() => setShowModal(false)}
+              onCancel={() => setShowModal(false)}
               footer={
                 <Row>
                   <Col sm={12} />    
                   <Col sm={12}>
                     <StyledFlex key="footer-right">
-                      <Button type="primary" style={{ marginLeft: '12px' }} disabled={mutation.isPending} onClick={() => onFinish()}>{'Generate Custom Prompt'}</Button>
+                      <Button type="primary" style={{ marginLeft: '12px' }} disabled={mutation.isPending || disabled} onClick={() => onFinish()}>{'Generate Custom Prompt'}</Button>
                       <Button disabled={mutation.isPending} style={{ marginLeft: '12px' }} onClick={() => setShowModal(false)}>{'Cancel'}</Button>
                     </StyledFlex>
                   </Col>

app/client/src/pages/DataGenerator/DataGenerator.tsx

@@ -1,7 +1,8 @@
 import isEmpty from 'lodash/isEmpty';
 import isString from 'lodash/isString';
-import { useRef, useState } from 'react';
-import { useLocation } from 'react-router-dom';
+import { useEffect, useRef, useState } from 'react';
+import { useLocation, useParams } from 'react-router-dom';
+
 import { Button, Flex, Form, Layout, Steps } from 'antd';
 import type { FormInstance } from 'antd';
 import ArrowBackIcon from '@mui/icons-material/ArrowBack';
@@ -17,7 +18,8 @@ import Finish from './Finish';
 
 import { DataGenWizardSteps, WizardStepConfig, WorkflowType } from './types';
 import { WizardCtx } from './utils';
-import { useGetDatasetDetails } from '../DatasetDetails/hooks';
+import { fetchDatasetDetails, useGetDatasetDetails } from '../DatasetDetails/hooks';
+import { useMutation } from '@tanstack/react-query';
 
 const { Content } = Layout;
 // const { Title } = Typography;
@@ -97,12 +99,30 @@ const DataGenerator = () => {
     const [current, setCurrent] = useState(0);
     const [maxStep, setMaxStep] = useState(0);
     const [isStepValid, setIsStepValid] = useState<boolean>(false);
+    
     // Data passed from listing table to prepopulate form
     const location = useLocation();
-    console.log('location?.state?.data:', location?.state?.data);
+    const { generate_file_name } = useParams();
     const initialData = location?.state?.data;
+    const mutation = useMutation({
+        mutationFn: fetchDatasetDetails
+    });
+
+
+    useEffect(() => {
+        if (generate_file_name && !mutation.data) {
+            mutation.mutate(generate_file_name);
+        }
+        if (mutation.data && mutation?.data?.dataset) {
+            form.setFieldsValue({
+                ...initialData,
+                ...(mutation?.data?.dataset as any)
+            });
+        }
+
+    }, [generate_file_name]);
+    
 
-    const datasetDetailsReq = location?.state?.data &&  useGetDatasetDetails(location?.state?.data?.generate_file_name)
     if (initialData?.technique) {
         initialData.workflow_type = initialData?.technique === 'sft' ? 
         WorkflowType.SUPERVISED_FINE_TUNING :

app/client/src/pages/DataGenerator/Finish.tsx

@@ -132,8 +132,9 @@ const Finish = () => {
 
     const isDemo = isDemoMode(num_questions, topics, form);
 
-    useEffect(() => { 
+    useEffect(() => {
         const formValues = form.getFieldsValue(true);
+        
         const doc_paths = formValues.doc_paths;
         if (Array.isArray(doc_paths) && !isEmpty(doc_paths)) {
             if (formValues.workflow_type === WorkflowType.SUPERVISED_FINE_TUNING) {
@@ -178,10 +179,12 @@ const Finish = () => {
         }
 
         if (formValues.workflow_type === WorkflowType.FREE_FORM_DATA_GENERATION) {
+            formValues.example_custom = formValues.examples;
             delete formValues.examples;
         }
 
         const args = {...formValues, is_demo: isDemo, model_params: formValues.model_parameters }
+        console.log('Triggering data generation with args:', args);
         triggerPost(args)
     }, []);