Merge pull request #62 from cloudera/dev

freeform job synthesis parameters update for freeformflag

Author: Khauneesh-AI

app/main.py

@@ -466,10 +466,10 @@ async def generate_freeform_data(request: SynthesisRequest):
     else:
         # Pass additional parameter to indicate this is a freeform request
         request_dict = request.model_dump()
-        request_dict['generation_type'] = 'freeform'
+        freeform = True
         # Convert back to SynthesisRequest object
         freeform_request = SynthesisRequest(**request_dict)
-        return synthesis_job.generate_job(freeform_request, core, mem, request_id=request_id)
+        return synthesis_job.generate_job(freeform_request, core, mem, request_id=request_id, freeform = freeform)
 
 @app.post("/synthesis/evaluate", 
     include_in_schema=True,
@@ -519,10 +519,10 @@ async def evaluate_freeform(request: EvaluationRequest):
         return evaluator_service.evaluate_row_data(request, request_id=request_id)
     else:
         request_dict = request.model_dump()
-        request_dict['generation_type'] = 'freeform'
+        freeform = True
         # Convert back to SynthesisRequest object
         freeform_request = EvaluationRequest(**request_dict)
-        return synthesis_job.evaluate_job(freeform_request, request_id=request_id)
+        return synthesis_job.evaluate_job(freeform_request, request_id=request_id,freeform = freeform)
 
 
 @app.post("/model/alignment",

app/run_eval_job.py

@@ -9,25 +9,7 @@
 
     os.chdir("/home/cdsw/synthetic-data-studio")
 
-# def check_and_install_requirements():
-#     """Check and install requirements from requirements.txt"""
-#     # Get the current working directory instead of using __file__
-#     current_dir = os.getcwd()
-#     requirements_path = os.path.join(current_dir, 'requirements.txt')
-    
-#     if os.path.exists(requirements_path):
-#         try:
-#             print(f"Installing requirements from: {requirements_path}")
-#             subprocess.check_call([sys.executable, '-m', 'pip', 'install', '-r', requirements_path])
-#         except subprocess.CalledProcessError as e:
-#             print(f"Error installing requirements: {e}")
-#             sys.exit(1)
-#     else:
-#         print("No requirements.txt found, continuing with existing packages")
-
-# # Run installation check at start
-# check_and_install_requirements()
-# Get the current notebook's directory
+
 notebook_dir = os.getcwd()
 
 # Detect the Python version dynamically
@@ -69,7 +51,7 @@ async def run_freeform_eval(request, job_name, request_id):
     """Run freeform data synthesis job"""
     try:
         job = EvaluatorService()
-        result = await job.evaluate_row_data(request, job_name, is_demo=False, request_id=request_id)
+        result = job.evaluate_row_data(request, job_name, is_demo=False, request_id=request_id)
         return result
     except Exception as e:
         print(f"Error in freeform synthesis: {e}")

app/run_job.py

@@ -8,25 +8,9 @@
 if os.getenv("IS_COMPOSABLE"):
     os.chdir("/home/cdsw/synthetic-data-studio")
 
-# def check_and_install_requirements():
-#     """Check and install requirements from requirements.txt"""
-#     # Get the current working directory instead of using __file__
-#     current_dir = os.getcwd()
-#     requirements_path = os.path.join(current_dir, 'requirements.txt')
-    
-#     if os.path.exists(requirements_path):
-#         try:
-#             print(f"Installing requirements from: {requirements_path}")
-#             subprocess.check_call([sys.executable, '-m', 'pip', 'install', '-r', requirements_path])
-#         except subprocess.CalledProcessError as e:
-#             print(f"Error installing requirements: {e}")
-#             sys.exit(1)
-#     else:
-#         print("No requirements.txt found, continuing with existing packages")
-
-
-# # Run installation check at start
-# check_and_install_requirements()
+
+
+
 
 # Get the current notebook's directory
 notebook_dir = os.getcwd()

app/services/synthesis_job.py

@@ -52,7 +52,8 @@ def _create_and_run_job(
         params: Dict[str, Any],
         cpu: int,
         memory: int,
-        request_id = None
+        request_id = None,
+        freeform = None
     ) -> tuple:
         """Common job creation and execution logic"""
         script_path = self.path_manager.get_str_path(f"app/{script_name}")
@@ -62,6 +63,8 @@ def _create_and_run_job(
         job_name = f"{display_name}_{random_id}" if display_name else f"{job_prefix}_{random_id}"
         params['job_name'] = job_name
         params['request_id'] = request_id
+        if freeform:
+            params['generation_type'] = 'freeform'
         file_name = f"{job_prefix}_args_{random_id}.json"
 
         # Write parameters to file
@@ -103,7 +106,7 @@ def _get_job_creator_name(self, job_run_id: str) -> str:
         ).job_runs[0].creator.name
 
     #@track_job("generate")
-    def generate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id = None) -> Dict[str, str]:
+    def generate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id = None, freeform = False) -> Dict[str, str]:
         """Create and run a synthesis generation job"""
         json_str = request.model_dump_json()
         params = json.loads(json_str)
@@ -114,7 +117,9 @@ def generate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id =
             params,
             cpu=cpu,
             memory=memory,
-            request_id=request_id
+            request_id=request_id,
+          freeform = freeform
+            
         )
 
         # Calculate total count
@@ -159,7 +164,7 @@ def generate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id =
         return {"job_name": job_name, "job_id": job_run.job_id}
 
     #@track_job("evaluate")
-    def evaluate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id = None) -> Dict[str, str]:
+    def evaluate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id = None, freeform = None) -> Dict[str, str]:
         """Create and run an evaluation job"""
         json_str = request.model_dump_json()
         params = json.loads(json_str)
@@ -170,7 +175,8 @@ def evaluate_job(self, request: Any, cpu: int = 2, memory: int = 4, request_id =
             params,
             cpu=cpu,
             memory=memory,
-            request_id=request_id
+            request_id=request_id,
+            freeform = freeform
         )
 
         custom_prompt_str = PromptHandler.get_default_custom_eval_prompt(