add t5gemma

Author: alien-0119

mindone/transformers/__init__.py

@@ -1335,6 +1335,14 @@
 from .models.trocr import TrOCRForCausalLM, TrOCRPreTrainedModel
 from .models.tvp import TvpForVideoGrounding, TvpModel, TvpPreTrainedModel
 from .models.udop import UdopEncoderModel, UdopForConditionalGeneration, UdopModel, UdopPreTrainedModel
+from .models.t5gemma import (
+    T5GemmaEncoderModel,
+    T5GemmaForConditionalGeneration,
+    T5GemmaPreTrainedModel,
+    T5GemmaForSequenceClassification,
+    T5GemmaForTokenClassification,
+    T5GemmaModel,
+)
 from .models.umt5 import (
     UMT5EncoderModel,
     UMT5ForQuestionAnswering,

mindone/transformers/models/__init__.py

@@ -228,6 +228,7 @@
     trocr,
     tvp,
     udop,
+    t5gemma,
     umt5,
     unispeech,
     unispeech_sat,

mindone/transformers/models/auto/configuration_auto.py

@@ -257,6 +257,7 @@
         ("trocr", "TrOCRConfig"),
         ("tvp", "TvpConfig"),
         ("udop", "UdopConfig"),
+        ("t5gemma", "T5GemmaConfig"),
         ("umt5", "UMT5Config"),
         ("unispeech", "UniSpeechConfig"),
         ("unispeech-sat", "UniSpeechSatConfig"),
@@ -521,6 +522,7 @@
         ("swinv2", "Swin Transformer V2"),
         ("swin2sr", "Swin2SR"),
         ("t5", "T5"),
+        ("t5gemma", "T5Gemma"),
         ("t5v1.1", "T5v1.1"),
         ("table-transformer", "Table Transformer"),
         ("tapas", "TAPAS"),

mindone/transformers/models/auto/modeling_auto.py

@@ -233,6 +233,7 @@
         ("timesformer", "TimesformerModel"),
         ("tvp", "TvpModel"),
         ("udop", "UdopModel"),
+        ("t5gemma", "T5GemmaModel"),
         ("umt5", "UMT5Model"),
         ("unispeech", "UniSpeechModel"),
         ("unispeech-sat", "UniSpeechSatModel"),
@@ -328,6 +329,7 @@
         ("vipllava", "VipLlavaForConditionalGeneration"),
         ("visual_bert", "VisualBertForPreTraining"),
         ("vit_mae", "ViTMAEForPreTraining"),
+        ("t5gemma", "T5GemmaForConditionalGeneration"),
         ("wav2vec2", "Wav2Vec2ForPreTraining"),
         ("wav2vec2-conformer", "Wav2Vec2ConformerForPreTraining"),
         ("xlm", "XLMWithLMHeadModel"),
@@ -397,6 +399,7 @@
         ("squeezebert", "SqueezeBertForMaskedLM"),
         ("t5", "T5ForConditionalGeneration"),
         ("tapas", "TapasForMaskedLM"),
+        ("t5gemma", "T5GemmaForConditionalGeneration"),
         ("wav2vec2", "Wav2Vec2ForMaskedLM"),
         ("whisper", "WhisperForConditionalGeneration"),
         ("xlm", "XLMWithLMHeadModel"),
@@ -831,6 +834,7 @@
         ("seamless_m4t_v2", "SeamlessM4Tv2ForTextToText"),
         ("squeezebert", "SqueezeBertForSequenceClassification"),
         ("t5", "T5ForConditionalGeneration"),
+        ("t5gemma", "T5GemmaForConditionalGeneration"),
         ("umt5", "UMT5ForConditionalGeneration"),
         ("xlm-prophetnet", "XLMProphetNetForConditionalGeneration"),
     ]
@@ -919,6 +923,7 @@
         ("starcoder2", "Starcoder2ForSequenceClassification"),
         ("t5", "T5ForSequenceClassification"),
         ("tapas", "TapasForSequenceClassification"),
+        ("t5gemma", "T5GemmaForSequenceClassification"),
         ("umt5", "UMT5ForSequenceClassification"),
         ("xlm", "XLMForSequenceClassification"),
         ("xlm-roberta", "XLMRobertaForSequenceClassification"),
@@ -1070,6 +1075,7 @@
         ("squeezebert", "SqueezeBertForTokenClassification"),
         ("stablelm", "StableLmForTokenClassification"),
         ("t5", "T5ForTokenClassification"),
+        ("t5gemma", "T5GemmaForTokenClassification"),
         ("umt5", "UMT5ForTokenClassification"),
         ("xlm", "XLMForTokenClassification"),
         ("xlm-roberta", "XLMRobertaForTokenClassification"),
@@ -1257,6 +1263,7 @@
         ("roberta-prelayernorm", "RobertaPreLayerNormModel"),
         ("squeezebert", "SqueezeBertModel"),
         ("t5", "T5EncoderModel"),
+        ("t5gemma", "T5GemmaEncoderModel"),
         ("umt5", "UMT5EncoderModel"),
         ("xlm", "XLMModel"),
         ("xlm-roberta", "XLMRobertaModel"),

mindone/transformers/models/t5gemma/__init__.py

@@ -0,0 +1,17 @@
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# This code is adapted from https://github.com/huggingface/transformers
+# with modifications to run transformers on mindspore.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from .modeling_t5gemma import *