Delete confusing parameter to make FT JIT easy to activate (PaddlePad…

…dle#1495)

* fix need build of faster generation

* update

* update

paddlenlp/ops/ext_utils.py

@@ -34,6 +34,8 @@
     # Clear it for other non-CUDA situations.
     CUDA_HOME = None
 
+LOADED_EXT = {}
+
 
 def _get_files(path):
     """
@@ -221,6 +223,8 @@ def load(name, build_dir=None, force=False, verbose=False, **kwargs):
         logger.warning("%s is not available because CUDA can not be found." %
                        name)
         raise NotImplementedError
+    if name in LOADED_EXT.keys():
+        return LOADED_EXT[name]
     if build_dir is None:
         # Maybe under package dir is better to avoid cmake source path conflict
         # with different source path.
@@ -247,7 +251,9 @@ def load(name, build_dir=None, force=False, verbose=False, **kwargs):
                     ext_sources, ext_filepath, 'newer'):
                 logger.debug("skipping '%s' extension (up-to-date) build" %
                              name)
-                return load_op_meta_info_and_register_op(ext_filepath)
+                ops = load_op_meta_info_and_register_op(ext_filepath)
+                LOADED_EXT[name] = ops
+                return LOADED_EXT[name]
 
     # write setup file and jit compile
     file_path = os.path.join(build_dir, "{}_setup.py".format(name))
@@ -256,7 +262,9 @@ def load(name, build_dir=None, force=False, verbose=False, **kwargs):
     if isinstance(extension, CMakeExtension):
         # Load a shared library (if exists) only to register op.
         if os.path.exists(ext_filepath):
-            load_op_meta_info_and_register_op(ext_filepath)
+            ops = load_op_meta_info_and_register_op(ext_filepath)
+            LOADED_EXT[name] = ops
+            return LOADED_EXT[name]
     else:
         # Import as callable python api
         return _import_module_from_library(name, build_base_dir, verbose)

paddlenlp/ops/faster_transformer/sample/encoder_decoding_sample.py

@@ -110,7 +110,7 @@ def do_predict(args):
 
     if args.enable_faster_encoder:
         transformer = enable_faster_encoder(
-            transformer, need_build=False, use_fp16=args.use_fp16_encoder)
+            transformer, use_fp16=args.use_fp16_encoder)
 
     src_word = generate_src_word(
         batch_size=args.infer_batch_size,

paddlenlp/ops/faster_transformer/transformer/decoder.py

@@ -21,7 +21,7 @@
 from paddle.fluid.layer_helper import LayerHelper
 from paddlenlp.transformers import WordEmbedding, PositionalEmbedding, position_encoding_init
 from paddlenlp.utils.log import logger
-from paddlenlp.ops.ext_utils import load
+from paddlenlp.ops.ext_utils import load, LOADED_EXT
 from paddlenlp.ops import transfer_param
 
 
@@ -160,8 +160,10 @@ def __init__(self,
 
         if decoder_lib is not None and os.path.isfile(decoder_lib):
             # Maybe it has been loadad by `ext_utils.load`
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoder_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoder_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoder_lib is not None:
                 logger.warning(

paddlenlp/ops/faster_transformer/transformer/decoding.py

@@ -23,7 +23,7 @@
 from paddle.fluid.layer_helper import LayerHelper
 import paddle
 
-from paddlenlp.ops.ext_utils import load
+from paddlenlp.ops.ext_utils import load, LOADED_EXT
 from paddlenlp.utils.log import logger
 
 
@@ -610,8 +610,10 @@ def __init__(self,
         #     raise ValueError("The path to decoding lib is not exist.")
         if decoding_lib is not None and os.path.isfile(decoding_lib):
             # Maybe it has been loadad by `ext_utils.load`
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoding_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoding_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoding_lib is not None:
                 logger.warning(
@@ -870,8 +872,10 @@ def parse_function(func_name):
 class InferGptDecoding(nn.Layer):
     def __init__(self, model, decoding_lib=None, use_fp16_decoding=False):
         if decoding_lib is not None and os.path.isfile(decoding_lib):
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoding_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoding_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoding_lib is not None:
                 logger.warning(
@@ -1078,8 +1082,10 @@ def __init__(self,
                  hidden_act="gelu"):
         if decoding_lib is not None and os.path.isfile(decoding_lib):
             # Maybe it has been loadad by `ext_utils.load`
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoding_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoding_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoding_lib is not None:
                 logger.warning(
@@ -1442,8 +1448,10 @@ class InferBartDecoding(nn.Layer):
     def __init__(self, model, decoding_lib=None, use_fp16_decoding=False):
         if decoding_lib is not None and os.path.isfile(decoding_lib):
             # Maybe it has been loadad by `ext_utils.load`
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoding_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoding_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoding_lib is not None:
                 logger.warning(
@@ -1683,8 +1691,10 @@ def __init__(self,
                  hidden_act="gelu"):
         if decoding_lib is not None and os.path.isfile(decoding_lib):
             # Maybe it has been loadad by `ext_utils.load`
-            paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
-                decoding_lib)
+            if "FasterTransformer" not in LOADED_EXT.keys():
+                ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                    decoding_lib)
+                LOADED_EXT["FasterTransformer"] = ops
         else:
             if decoding_lib is not None:
                 logger.warning(

paddlenlp/ops/faster_transformer/transformer/encoder.py

@@ -238,10 +238,7 @@ def encoder_forward(self, src, src_mask=None, cache=None):
     return output
 
 
-def enable_faster_encoder(self,
-                          need_build=True,
-                          use_fp16=False,
-                          encoder_lib=None):
+def enable_faster_encoder(self, use_fp16=False, encoder_lib=None):
     """
     Compiles fusion encoder operator intergrated FasterTransformer using the
     method of JIT(Just-In-Time) and replaces the `forward` function of
@@ -281,19 +278,21 @@ def init_func(layer):
                 convert_to_fp16(layer)
 
     if not self.training:
-        if need_build:
-            try:
-                # Pass decoding lib to prevent re-building encoder.
-                # Todo: check weather decoding lib have contained encoder or not.
-                if encoder_lib is not None:
-                    load_op_meta_info_and_register_op(encoder_lib)
-                else:
-                    load("FasterTransformer", verbose=True)
-            except Exception:
-                logger.warning(
-                    "Exception occurs when using FasterEncoder. " \
-                    "The original forward will be involved. ")
-                return self
+        try:
+            # Pass decoding lib to prevent re-building encoder.
+            # Todo: check weather decoding lib have contained encoder or not.
+            if encoder_lib is not None:
+                if "FasterTransformer" not in LOADED_EXT.keys():
+                    ops = paddle.utils.cpp_extension.load_op_meta_info_and_register_op(
+                        decoding_lib)
+                    LOADED_EXT["FasterTransformer"] = ops
+            else:
+                load("FasterTransformer", verbose=True)
+        except Exception:
+            logger.warning(
+                "Exception occurs when using FasterEncoder. " \
+                "The original forward will be involved. ")
+            return self
         for layer in self.children():
             layer.apply(init_func)
     return self

paddlenlp/ops/faster_transformer/transformer/faster_transformer.py

@@ -1161,7 +1161,7 @@ def forward(self,
                 **model_kwargs):
 
         if encoder_output is None:
-            self.encoder = enable_faster_encoder(self.encoder, need_build=False)
+            self.encoder = enable_faster_encoder(self.encoder)
             assert input_ids is not None, "You have to specify either input_ids or encoder_output."
             encoder_output = self.prepare_encoder_decoder_kwargs_for_generation(
                 input_ids, model_kwargs)["encoder_output"]
@@ -1265,7 +1265,7 @@ def forward(self,
 
         #(gongenlei) Not enable_faster_encoder temporarily
         if encoder_output is None:
-            self.encoder = enable_faster_encoder(self.encoder, need_build=False)
+            self.encoder = enable_faster_encoder(self.encoder)
             assert input_ids is not None, "You have to specify either input_ids or encoder_output."
             encoder_output = self.prepare_encoder_decoder_kwargs_for_generation(
                 input_ids, model_kwargs)["encoder_output"]