Add fastapi endpoint that runs lm-harness

README.md

@@ -8,3 +8,8 @@ pinned: false
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+```shell
+conda activate model-eval-be
+pip install -r requirements.txt
+```

app.py

@@ -1,7 +1,25 @@
 from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from svc.router import router
 
-app = FastAPI()
+app = FastAPI(
+    title="Resume Generator API",
+    description="API for converting audio/text to structured resume with PDF generation",
+    version="1.0.0"
+)
+
+# Add CORS middleware
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # Modify this in production
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Include our router
+app.include_router(router, prefix="/api")
 
 @app.get("/")
-def greet_json():
-    return {"Hello": "World!"}
+async def health_check():
+    return {"status": "healthy"}

requirements.txt

@@ -1,2 +1,3 @@
 fastapi
-uvicorn[standard]
+uvicorn[standard]
+lm_eval

svc/router.py

@@ -0,0 +1,32 @@
+from fastapi import APIRouter, HTTPException
+import logging
+
+from lm_eval import evaluator, utils
+from svc.schemas import LMHarnessTaskRequest, LMHarnessTaskResponse
+
+router = APIRouter()
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
+logger = logging.getLogger(__name__)
+
+
+
+@router.post("/chat",  response_model=LMHarnessTaskResponse)
+def inference_model(request: LMHarnessTaskRequest):
+    try:
+        results = evaluator.simple_evaluate(
+            model=request.model,
+            model_args=request.model_args,
+            tasks=request.tasks,
+            num_fewshot=request.num_fewshot,
+            batch_size=request.batch_size,
+            device=request.device,
+            limit=request.limit,
+            write_out=request.write_out # Whether to write out an example document and model input, for checking task integrity
+        )
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"lm-harness task execution failed for model: {request.model_args}")
+
+    logger.info(results)
+    return LMHarnessTaskResponse(results=results)
+

svc/schemas.py

@@ -0,0 +1,16 @@
+from pydantic import BaseModel
+from typing import List, Optional, Union, Any
+
+class LMHarnessTaskRequest(BaseModel):
+    model: str
+    model_args: Optional[Union[str, dict]] = None
+    tasks: Optional[List[Union[str, dict, object]]] = None
+    num_fewshot: Optional[int] = None
+    batch_size: Optional[int] = None
+    device: Optional[str] = None
+    limit: Optional[Union[int, float]] = None
+    write_out: bool = False # Whether to write out an example document and model input, for checking task integrity
+
+
+class LMHarnessTaskResponse(BaseModel):
+    results: dict[Any]