CI fixes and Model Export logger silencing (#283)

AnishaUdayakumar · adrianboguszewski · Copilot · web-flow · commit c23fef1a4809 · 2025-06-06T11:57:17.000+02:00
* Fix CI for multimodal generator * Allowed server to accpet different models * Replaced prints with logger * Moved model selection to the UI * Moved image size selection back to the UI * Update demos/paint_your_dreams_demo/main.py Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Remove model selection from setup scripts * Replaced command line conversion with python code (#276) * Replaced command line with python code * Update demos/paint_your_dreams_demo/main.py Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update main.py --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * docs: include medium article to AI agent RAG kit (#277) Signed-off-by: Antonio Martinez <jose.antonio.martinez.torres@intel.com> * Enabled more models in paint your dreams (#279) * Enabled more models in paint your dreams * Enabled more tokenizers * Fixed sdxl inference * Used better pipeline for conversion * Added nncf * Removed a few models * Updated CI to skip HF login if token not provided (#280) * Updated CI to skip HF login if token not provided * Update sanity-check-demos.yml * Bump tar-fs from 3.0.8 to 3.0.9 in /demos/hide_your_mess_behind_demo (#282) Bumps [tar-fs](https://github.com/mafintosh/tar-fs) from 3.0.8 to 3.0.9. - [Commits](mafintosh/tar-fs@v3.0.8...v3.0.9) --- updated-dependencies: - dependency-name: tar-fs dependency-version: 3.0.9 dependency-type: indirect ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * CI fixes and Model Export logger silencing * Refactor test.py with log-based FastAPI readiness check and adding missing file as copilot suggested * Move Demo is ready to test.py after API tests, revert action.yml with updated Windows handling * Fix CRLF to LF in action.yml for clean CI diff * Update ai_ref_kits/multimodal_ai_visual_generator/ci/test.py --------- Signed-off-by: Antonio Martinez <jose.antonio.martinez.torres@intel.com> Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Adrian Boguszewski <adrian.boguszewski@intel.com> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Antonio Martinez <jose.antonio.martinez.torres@intel.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Adrian Boguszewski <adekboguszewski@gmail.com>
diff --git a/.github/reusable-steps/gradio-action/action.yml b/.github/reusable-steps/gradio-action/action.yml
@@ -24,7 +24,7 @@ runs:
         else
           python ${{ inputs.script }} 2>&1 | tee gradio_log.txt &
         fi
-        
+
         # Assign process ID
         app_pid=$(ps aux | grep -i '[p]ython ${{ inputs.script }}' | awk '{print $2}')
 
@@ -44,33 +44,62 @@ runs:
 
         # Exit with the readiness check status
         exit $status
+
     - name: Run Gradio App (Windows)
       if: ${{ runner.os == 'Windows' }}
       shell: powershell
       run: |
         cd ${{ inputs.project }}
+        Write-Output "==> Running script: ${{ inputs.script }}"
 
-        # Start the Gradio app in the background and redirect both output and error to the same file
-        Start-Process -NoNewWindow -FilePath "python" -ArgumentList "${{ inputs.script }}" -RedirectStandardOutput gradio_stdout.txt -RedirectStandardError gradio_stderr.txt
+        $timeout = ${{ inputs.timeout }}
+        $start_time = Get-Date
+        $success = $false
 
-        # Wait a moment to ensure the process starts
-        Start-Sleep -Seconds 5
+        if ("${{ inputs.script }}" -like "*test.py") {
+          Write-Output "==> test.py detected. Running in foreground..."
 
-        # Assign process ID
-        $app_pid = (Get-Process | Where-Object { $_.Name -eq "python.*" }).Id
+          $output = python "${{ inputs.script }}"
+          $output | Out-File -FilePath gradio_log.txt -Encoding utf8
+          Get-Content -Path gradio_log.txt
 
-        $timeout = ${{ inputs.timeout }}
-        # Wait for the specific log message
-        $start_time = Get-Date
-        Get-Content -Wait gradio_stderr.txt, gradio_stdout.txt | ForEach-Object {
-          if ($_ -match "Demo is ready!") { break }
-          if (((Get-Date) - $start_time).TotalSeconds -ge $timeout) {
-            exit 1
+          if ($LASTEXITCODE -eq 0) {
+            $success = $true
+          } else {
+            Write-Error "Script exited with code $LASTEXITCODE"
+          }
+
+        } else {
+          Write-Output "==> Long-running app detected. Launching in background..."
+          $proc = Start-Process -NoNewWindow -FilePath "python" -ArgumentList "${{ inputs.script }}" 
+                   -RedirectStandardOutput gradio_stdout.txt -RedirectStandardError gradio_stderr.txt -PassThru
+          $app_pid = $proc.Id
+          Write-Output "==> App PID: $app_pid"
+
+          while ($true) {
+            if (Test-Path gradio_stdout.txt) {
+              $content = Get-Content gradio_stdout.txt -Raw
+              if ($content -match "Demo is ready!") {
+                $success = $true
+                break
+              }
+            }
+            if (((Get-Date) - $start_time).TotalSeconds -ge $timeout) {
+              Write-Output "==> Timeout waiting for readiness."
+              break
+            }
+            Start-Sleep -Seconds 2
           }
+
+          Write-Output "==> Stopping background process..."
+          Stop-Process -Id $app_pid -Force -ErrorAction SilentlyContinue
         }
 
-        # Stop the Gradio app process
-        Stop-Process -Id $app_pid -Force -ErrorAction SilentlyContinue
+        Write-Output "==> Gradio Log Output:"
+        if (Test-Path gradio_log.txt) { Get-Content gradio_log.txt }
+        if (Test-Path gradio_stdout.txt) { Get-Content gradio_stdout.txt }
+        if (Test-Path gradio_stderr.txt) { Get-Content gradio_stderr.txt }
 
-        # Exit with the readiness check status
-        exit 0
+        if (-not $success) {
+          exit 1
+        }
diff --git a/ai_ref_kits/multimodal_ai_visual_generator/ci/test.py b/ai_ref_kits/multimodal_ai_visual_generator/ci/test.py
@@ -19,12 +19,14 @@
 from convert_and_optimize_text2image import convert_image_model
 
 # ----- Configuration -----
+
 MODEL_DIR = Path("models")
 LLM_MODEL_TYPE = "tiny-llama-1b-chat"
 IMAGE_MODEL_TYPE = "lcm"
 PRECISION = "int4"
+LOG_FILE = Path("gradio_log.txt")
 
-# ----- Step 1: Export Models if Needed -----
+# ----- Step 1: Export Models if Needed (will handle download internally) -----
 logger.info("Checking and exporting LLM + Text2Image models if necessary...")
 convert_chat_model(LLM_MODEL_TYPE, PRECISION, MODEL_DIR)
 convert_image_model(IMAGE_MODEL_TYPE, PRECISION, MODEL_DIR)
@@ -38,22 +40,29 @@
     "MODEL_PRECISION": PRECISION
 })
 
-process = subprocess.Popen(
-    [sys.executable, "-m", "uvicorn", "main:app", "--host", "127.0.0.1", "--port", "8000"],
-    env=env
-)
+with LOG_FILE.open("w") as lf:
+    process = subprocess.Popen(
+        [sys.executable, "-m", "uvicorn", "main:app", "--host", "127.0.0.1", "--port", "8000"],
+        env=env,
+        stdout=lf,
+        stderr=subprocess.STDOUT
+    )
 
 try:
-    # Wait up to ~130 seconds (130 retries x 1s sleep) for FastAPI server to come up
-    for _ in range(130):
-        try:
-            r = requests.get("http://localhost:8000/docs", timeout=2)
-            if r.status_code == 200:
+    # ----- Wait for Readiness from Logs -----
+    logger.info("Waiting for FastAPI log to report readiness...")
+    start_time = time.time()
+    timeout = 130  # seconds
+
+    while time.time() - start_time < timeout:
+        if LOG_FILE.exists():
+            content = LOG_FILE.read_text()
+            if "Uvicorn running on" in content or "Application startup complete." in content:
+                logger.info("FastAPI server is up.")
                 break
-        except requests.ConnectionError:
-            time.sleep(1)
+        time.sleep(1)
     else:
-        raise RuntimeError("FastAPI server did not start within 130 seconds.")
+        raise RuntimeError("FastAPI server did not start within timeout period.")
 
     # ----- Step 3: Test Story Prompt Generation -----
     logger.info("Testing /generate_story_prompts endpoint...")
@@ -63,8 +72,7 @@
     )
     assert response1.status_code == 200, f"Story generation failed: {response1.text}"
     scenes = response1.json()["scenes"]
-    logger.info("Generated scenes: %s", scenes)
-    logger.info("Scene prompt generation test passed.")
+    logger.info("Scene prompt generation test passed. Example: %s", scenes)
 
     # ----- Step 4: Test Image Generation -----
     logger.info("Testing /generate_images endpoint...")
@@ -74,10 +82,10 @@
     )
     assert response2.status_code == 200, f"Image generation failed: {response2.text}"
     image = response2.json()["image"]
-    logger.info("Image string (truncated): %s", image[:100])
-    logger.info("Image generation test passed.")
+    logger.info("Image generation test passed. Base64 (truncated): %s", image[:100])
+    logger.info("Demo is ready!")
 
 finally:
     logger.info("Shutting down FastAPI server...")
     process.terminate()
-    process.wait()
+    process.wait()
diff --git a/ai_ref_kits/multimodal_ai_visual_generator/convert_and_optimize_llm.py b/ai_ref_kits/multimodal_ai_visual_generator/convert_and_optimize_llm.py
@@ -75,15 +75,7 @@ def validate_export(output_dir: Path) -> list[str]:
     Validates the presence of all required files in the exported model directory.
     Returns a list of any missing critical files.
     """
-    logger.info("Verifying exported files:")
-    missing = []
-    for file in CRITICAL_FILES:
-        if not (output_dir / file).exists():
-            logger.error(f"Missing: {file}")
-            missing.append(file)
-        else:
-            logger.info(f"Found: {file}")
-    return missing
+    return [file for file in CRITICAL_FILES if not (output_dir / file).exists()]
 
 def quantize_model(model, precision: str, output_dir: Path):
     """
@@ -137,7 +129,7 @@ def convert_chat_model(model_type: str, precision: str, model_dir: Path) -> Path
     }
     (output_dir / "model_index.json").write_text(json.dumps(model_index, indent=2))
 
-    logger.info("Verifying critical files:")
+    logger.info("Checking for missing critical files silently...")
     missing_files = validate_export(output_dir)
     if missing_files:
         logger.warning("Export completed with missing files.")
diff --git a/ai_ref_kits/multimodal_ai_visual_generator/convert_and_optimize_text2image.py b/ai_ref_kits/multimodal_ai_visual_generator/convert_and_optimize_text2image.py
@@ -55,22 +55,11 @@ def run_optimum_export(model_id: str, output_dir: Path, precision: str):
     subprocess.run(cmd, shell=(platform.system() == "Windows"), check=True)
 
 def validate_export(output_dir: Path, critical_files: list[str]) -> list[str]:
-    logger.info("Verifying exported files:")
-    missing = []
-
-    for file in critical_files:
-        if not (output_dir / file).exists():
-            logger.error(f"Missing: {file}")
-            missing.append(file)
-        else:
-            logger.info(f"Found: {file}")
-
-    if missing:
-        logger.warning("Export completed with missing files.")
-    else:
-        logger.info("All critical files verified successfully.")
-
-    return missing
+    """
+    Silently checks for the presence of required exported files.
+    Returns a list of missing files, without logging each one.
+    """
+    return [file for file in critical_files if not (output_dir / file).exists()]
 
 def convert_image_model(model_type: str, precision: str, model_dir: Path) -> Path:
     """
@@ -88,12 +77,15 @@ def convert_image_model(model_type: str, precision: str, model_dir: Path) -> Pat
             logger.info("Skipping re-export.\n")
             return output_dir
         else:
-            logger.warning(f"Export folder exists but missing files: {missing}")
             logger.info("Re-exporting model...\n")
 
     # Run export and validate output
     run_optimum_export(model_id, output_dir, precision)
     missing_files = validate_export(output_dir, CRITICAL_FILES)
+    if missing_files:
+        logger.warning(f"Export completed with missing files: {missing_files}")
+    else:
+        logger.info("All critical files verified successfully.")
 
     logger.info(f"Model exported to: {output_dir}\n")
     return output_dir
diff --git a/ai_ref_kits/multimodal_ai_visual_generator/main.py b/ai_ref_kits/multimodal_ai_visual_generator/main.py
@@ -11,6 +11,7 @@
 import sys
 import yaml
 import openvino_genai as ov_genai
+import openvino as ov
 import logging
 import random
 
@@ -61,29 +62,34 @@
 # ---------- Load Config ----------
 with open(CONFIG_PATH, "r") as f:
     config = yaml.safe_load(f)
+    
+# ---------- Determine Device (GPU if available, else fallback) ----------
+core = ov.Core()
+preferred_device = "GPU" if "GPU" in core.available_devices else "CPU"
+print(f"Using OpenVINO device: {preferred_device}")
 
-# ---------- Lazy load models if available ----------
+# ---------- Load models ----------
 image_pipe = None
 llm_pipe = None
 
 if image_model_dir.exists():
     try:
-        image_pipe = ov_genai.Text2ImagePipeline(image_model_dir, device="GPU")
-        logger.info("Image model loaded.")
+        image_pipe = ov_genai.Text2ImagePipeline(image_model_dir, device=preferred_device)
+        logger.info("Image model loaded successfully.")
     except Exception as e:
-        logger.error(f"Failed to load Image model: {e}")
+        logger.error(f"Failed to load image model: {e}")
 else:
     logger.warning(f"Image model not found at {image_model_dir}")
 
 if llm_model_dir.exists():
     try:
-        llm_pipe = ov_genai.LLMPipeline(str(llm_model_dir), device="GPU")
-        logger.info("LLM model loaded.")
+        llm_pipe = ov_genai.LLMPipeline(str(llm_model_dir), device=preferred_device)
+        logger.info("LLM model loaded successfully.")
     except Exception as e:
         logger.error(f"Failed to load LLM model: {e}")
 else:
     logger.warning(f"LLM model not found at {llm_model_dir}")
-    
+
 llm_config = ov_genai.GenerationConfig()
 llm_config.max_new_tokens = 256
 llm_config.apply_chat_template = False
@@ -209,7 +215,6 @@ def callback(step, num_steps, latent):
 
 # ---------- Server Start Print ----------
 if image_pipe or llm_pipe:
-    logger.info("Demo is ready!")
     logger.info("FastAPI backend is running.")
     logger.info("In a separate terminal, start the Streamlit app using: streamlit run streamlit_app.py")
 else:
