chore: improvement hybrid deployment (#766)

src/concrete/ml/deployment/fhe_client_server.py

@@ -239,7 +239,7 @@ def _export_model_to_json(self, is_training: bool = False) -> Path:
 
         return json_path
 
-    def save(self, mode: DeploymentMode = DeploymentMode.INFERENCE, via_mlir: bool = False):
+    def save(self, mode: DeploymentMode = DeploymentMode.INFERENCE, via_mlir: bool = True):
         """Export all needed artifacts for the client and server.
 
         Arguments:

src/concrete/ml/torch/hybrid_model.py

@@ -539,7 +539,7 @@ def _save_fhe_circuit(self, path: Path, via_mlir=False):
             )
             model_dev.save(via_mlir=via_mlir)
 
-    def save_and_clear_private_info(self, path: Path, via_mlir=False):
+    def save_and_clear_private_info(self, path: Path, via_mlir=True):
         """Save the PyTorch model to the provided path and also saves the corresponding FHE circuit.
 
         Args:

use_case_examples/hybrid_model/compile_hybrid_llm.py

@@ -34,7 +34,7 @@ def compile_model(
     models_dir.mkdir(exist_ok=True)
     model_dir = models_dir / model_name
     print(f"Saving to {model_dir}")
-    via_mlir = bool(int(os.environ.get("VIA_MLIR", 0)))
+    via_mlir = bool(int(os.environ.get("VIA_MLIR", 1)))
     hybrid_model.save_and_clear_private_info(model_dir, via_mlir=via_mlir)
 
 
@@ -101,7 +101,9 @@ def module_names_parser(string: str) -> List[str]:
     max_context_size = 20
     num_samples = 50
 
-    dataset = load_dataset("wikipedia", "20220301.en")
+    dataset = load_dataset(
+        "wikimedia/wikipedia", "20231101.en", revision="b04c8d1ceb2f5cd4588862100d08de323dccfbaa"
+    )
     print(model)
     models_dir = Path(__file__).parent / os.environ.get("MODELS_DIR_NAME", "compiled_models")
     models_dir.mkdir(exist_ok=True)

use_case_examples/hybrid_model/requirements.txt

@@ -1,6 +1,7 @@
 concrete-ml
 accelerate
 datasets
+transformers
 apache_beam==2.49.0 
 mwparserfromhell==0.6.4
 loguru==0.7.0