Implemented parallelism to run inference on mental and physical model

main.py

@@ -5,6 +5,7 @@
 import os
 import anthropic
 from contextlib import asynccontextmanager
+import asyncio
 from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel
@@ -146,25 +147,13 @@ async def predict(request: PredictRequest):
         # 1. Enrich text with goal
         enriched_text = f"Goal: {request.user_profile.primary_goal}. {request.text}"
 
-        # 2. Run physical model
-        physical_output = run_inference(
-            ml_models["physical_model"],
-            ml_models["tokenizer"],
-            enriched_text,
-            PHYSICAL_LABEL_COLS,
-            PHYSICAL_DECODERS
+        # 2. Run physical and mental model using asyncio
+        physical_output, mental_output = await asyncio.gather(
+            asyncio.to_thread(run_inference, ml_models["physical_model"], ml_models["tokenizer"], enriched_text, PHYSICAL_LABEL_COLS, PHYSICAL_DECODERS),
+            asyncio.to_thread(run_inference, ml_models["mental_model"],   ml_models["tokenizer"], enriched_text, MENTAL_LABEL_COLS,   MENTAL_DECODERS)
         )
 
-        # 3. Run mental model
-        mental_output = run_inference(
-            ml_models["mental_model"],
-            ml_models["tokenizer"],
-            enriched_text,
-            MENTAL_LABEL_COLS,
-            MENTAL_DECODERS
-        )
-
-        # 4. Build Claude prompt
+        # 3. Build Claude prompt
         prompt = build_claude_prompt(
             request.user_profile.primary_goal,
             request.user_profile.modifiers,
@@ -172,7 +161,7 @@ async def predict(request: PredictRequest):
             mental_output
         )
 
-        # 5. Call Claude API
+        # 4. Call Claude API
         # ANTHROPIC_API_KEY must be set as a HuggingFace Space secret
         claude_client = anthropic.Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
         message       = claude_client.messages.create(
@@ -181,7 +170,7 @@ async def predict(request: PredictRequest):
             messages=[{"role": "user", "content": prompt}]
         )
 
-        # 6. Return full response
+        # 5. Return full response
         return PredictResponse(
             physical=physical_output,
             mental=mental_output,