add compatibility for protobuf 4.x

alfred/fm/huggingface.py

@@ -18,7 +18,8 @@
 
 logger = logging.getLogger(__name__)
 
-from transformers import LlamaPreTrainedModel, MistralPreTrainedModel
+from transformers.models.llama.modeling_llama import LlamaPreTrainedModel
+from transformers.models.mistral.modeling_mistral import MistralPreTrainedModel
 
 dtype_match = {
     "auto": "auto",

alfred/fm/remote/grpc.py

@@ -11,11 +11,11 @@
 from PIL import Image
 from tqdm.auto import tqdm
 
-from alfred.fm.query import Query, RankedQuery, CompletionQuery
-from alfred.fm.remote.protos import query_pb2
-from alfred.fm.remote.protos import query_pb2_grpc
-from alfred.fm.remote.utils import get_ip, tensor_to_bytes, bytes_to_tensor, port_finder
-from alfred.fm.response import RankedResponse, CompletionResponse
+from ..query import Query, RankedQuery, CompletionQuery
+from ..remote.protos import query_pb2
+from ..remote.protos import query_pb2_grpc
+from ..remote.utils import get_ip, tensor_to_bytes, bytes_to_tensor, port_finder
+from ..response import RankedResponse, CompletionResponse
 
 logger = logging.getLogger(__name__)
 

alfred/fm/remote/protos/query_pb2.pyi

@@ -1,46 +1,17 @@
-from typing import ClassVar as _ClassVar, Optional as _Optional
-
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import message as _message
+from typing import ClassVar as _ClassVar, Optional as _Optional
 
 DESCRIPTOR: _descriptor.FileDescriptor
 
-class EncodeRequest(_message.Message):
-    __slots__ = ["kwargs", "message", "reduction"]
-    KWARGS_FIELD_NUMBER: _ClassVar[int]
-    MESSAGE_FIELD_NUMBER: _ClassVar[int]
-    REDUCTION_FIELD_NUMBER: _ClassVar[int]
-    kwargs: str
-    message: str
-    reduction: str
-
-    def __init__(
-        self,
-        message: _Optional[str] = ...,
-        reduction: _Optional[str] = ...,
-        kwargs: _Optional[str] = ...,
-    ) -> None: ...
-
-class EncodeResponse(_message.Message):
-    __slots__ = ["embedding", "success"]
-    EMBEDDING_FIELD_NUMBER: _ClassVar[int]
-    SUCCESS_FIELD_NUMBER: _ClassVar[int]
-    embedding: bytes
-    success: bool
-
-    def __init__(
-        self, embedding: _Optional[bytes] = ..., success: bool = ...
-    ) -> None: ...
-
 class RunRequest(_message.Message):
-    __slots__ = ["candidate", "kwargs", "message"]
+    __slots__ = ("message", "candidate", "kwargs")
+    MESSAGE_FIELD_NUMBER: _ClassVar[int]
     CANDIDATE_FIELD_NUMBER: _ClassVar[int]
     KWARGS_FIELD_NUMBER: _ClassVar[int]
-    MESSAGE_FIELD_NUMBER: _ClassVar[int]
+    message: str
     candidate: str
     kwargs: str
-    message: str
-
     def __init__(
         self,
         message: _Optional[str] = ...,
@@ -49,18 +20,17 @@ class RunRequest(_message.Message):
     ) -> None: ...
 
 class RunResponse(_message.Message):
-    __slots__ = ["embedding", "logit", "message", "ranked", "success"]
-    EMBEDDING_FIELD_NUMBER: _ClassVar[int]
-    LOGIT_FIELD_NUMBER: _ClassVar[int]
+    __slots__ = ("message", "ranked", "success", "logit", "embedding")
     MESSAGE_FIELD_NUMBER: _ClassVar[int]
     RANKED_FIELD_NUMBER: _ClassVar[int]
     SUCCESS_FIELD_NUMBER: _ClassVar[int]
-    embedding: bytes
-    logit: str
+    LOGIT_FIELD_NUMBER: _ClassVar[int]
+    EMBEDDING_FIELD_NUMBER: _ClassVar[int]
     message: str
     ranked: bool
     success: bool
-
+    logit: str
+    embedding: bytes
     def __init__(
         self,
         message: _Optional[str] = ...,
@@ -69,3 +39,28 @@ class RunResponse(_message.Message):
         logit: _Optional[str] = ...,
         embedding: _Optional[bytes] = ...,
     ) -> None: ...
+
+class EncodeRequest(_message.Message):
+    __slots__ = ("message", "reduction", "kwargs")
+    MESSAGE_FIELD_NUMBER: _ClassVar[int]
+    REDUCTION_FIELD_NUMBER: _ClassVar[int]
+    KWARGS_FIELD_NUMBER: _ClassVar[int]
+    message: str
+    reduction: str
+    kwargs: str
+    def __init__(
+        self,
+        message: _Optional[str] = ...,
+        reduction: _Optional[str] = ...,
+        kwargs: _Optional[str] = ...,
+    ) -> None: ...
+
+class EncodeResponse(_message.Message):
+    __slots__ = ("embedding", "success")
+    EMBEDDING_FIELD_NUMBER: _ClassVar[int]
+    SUCCESS_FIELD_NUMBER: _ClassVar[int]
+    embedding: bytes
+    success: bool
+    def __init__(
+        self, embedding: _Optional[bytes] = ..., success: bool = ...
+    ) -> None: ...

alfred/fm/remote/protos/query_pb2_grpc.py

@@ -2,13 +2,7 @@
 """Client and server classes corresponding to protobuf-defined services."""
 import grpc
 
-try:
-    import alfred.fm.remote.protos.query_pb2 as query__pb2
-except ImportError:
-    try:
-        import query_pb2 as query__pb2
-    except ModuleNotFoundError:
-        from . import query_pb2 as query__pb2
+from . import query_pb2 as query__pb2
 
 
 class QueryServiceStub(object):

docs/alfred/fm/huggingface.md

@@ -12,7 +12,7 @@ Huggingface
 
 ## HuggingFaceModel
 
-[Show source in huggingface.py:43](../../../alfred/fm/huggingface.py#L43)
+[Show source in huggingface.py:44](../../../alfred/fm/huggingface.py#L44)
 
 The HuggingFaceModel class is a wrapper for HuggingFace models,
 including both Seq2Seq (Encoder-Decoder, e.g. T5, T0) and Causal

docs/alfred/fm/remote/protos/query_pb2_grpc.md

@@ -21,7 +21,7 @@ Query Pb2 Grpc
 
 ## QueryService
 
-[Show source in query_pb2_grpc.py:71](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L71)
+[Show source in query_pb2_grpc.py:65](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L65)
 
 Missing associated documentation comment in .proto file.
 
@@ -34,7 +34,7 @@ class QueryService(object):
 
 ### QueryService.Encode
 
-[Show source in query_pb2_grpc.py:74](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L74)
+[Show source in query_pb2_grpc.py:68](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L68)
 
 #### Signature
 
@@ -57,7 +57,7 @@ def Encode(
 
 ### QueryService.Run
 
-[Show source in query_pb2_grpc.py:103](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L103)
+[Show source in query_pb2_grpc.py:97](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L97)
 
 #### Signature
 
@@ -82,7 +82,7 @@ def Run(
 
 ## QueryServiceServicer
 
-[Show source in query_pb2_grpc.py:35](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L35)
+[Show source in query_pb2_grpc.py:29](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L29)
 
 Missing associated documentation comment in .proto file.
 
@@ -95,7 +95,7 @@ class QueryServiceServicer(object):
 
 ### QueryServiceServicer().Encode
 
-[Show source in query_pb2_grpc.py:38](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L38)
+[Show source in query_pb2_grpc.py:32](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L32)
 
 Missing associated documentation comment in .proto file.
 
@@ -108,7 +108,7 @@ def Encode(self, request_iterator, context):
 
 ### QueryServiceServicer().Run
 
-[Show source in query_pb2_grpc.py:44](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L44)
+[Show source in query_pb2_grpc.py:38](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L38)
 
 Missing associated documentation comment in .proto file.
 
@@ -123,7 +123,7 @@ def Run(self, request_iterator, context):
 
 ## QueryServiceStub
 
-[Show source in query_pb2_grpc.py:14](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L14)
+[Show source in query_pb2_grpc.py:8](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L8)
 
 Missing associated documentation comment in .proto file.
 
@@ -139,7 +139,7 @@ class QueryServiceStub(object):
 
 ## add_QueryServiceServicer_to_server
 
-[Show source in query_pb2_grpc.py:51](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L51)
+[Show source in query_pb2_grpc.py:45](../../../../../alfred/fm/remote/protos/query_pb2_grpc.py#L45)
 
 #### Signature
 

requirements.txt

@@ -1,8 +1,8 @@
-datasets~=2.4.0
 paramiko>=2.7.2
 pyarrow>=3.0.0
 torch>=1.8.0
 
+datasets
 transformers
 accelerate
 numpy>=1.21.0
@@ -12,8 +12,8 @@ arrow>=0.13.1
 setuptools>=58.0.4
 scipy>=1.7.1
 
-grpcio==1.48.1
-protobuf==3.20.0
+grpcio==1.62.0
+protobuf==4.25.2
 sentencepiece
 deepspeed>=0.8.2 
 

setup.py

@@ -8,5 +8,5 @@
     author_email='peilin_yu@brown.edu',
     description='Toolkit for Prompted Weak Supervisions',
     packages=find_packages(),
-    install_requires=['numpy', 'scipy', 'torch', 'tqdm', 'torchvision', "paramiko>=2.7.2", "pyarrow>=3.0.0", "grpcio==1.48.1", "protobuf==3.20.0"],
+    install_requires=['numpy', 'scipy', 'torch', 'tqdm', 'torchvision', "accelerate", "paramiko>=2.7.2", "pyarrow>=3.0.0", "grpcio", "protobuf<5"],
 )

test/unit_test/test_grpc.py

@@ -1,4 +1,5 @@
 import threading
+import time
 import unittest
 
 import grpc
@@ -36,6 +37,7 @@ def server_starter(server, client, port):
 
         self.channel = grpc.insecure_channel(f"localhost:{self.port}")
         self.stub = query_pb2_grpc.QueryServiceStub(self.channel)
+        time.sleep(2)
 
     def test_run_single_query(self):
         # Test running a single query using the client