fix(transformers/blip2): update blip to v4.50.3 (#1130)

Author: Cui-yshoho

mindone/transformers/models/auto/configuration_auto.py

@@ -91,6 +91,7 @@
         ("mpt", "MptConfig"),
         ("starcoder2", "Starcoder2Config"),
         ("mt5", "MT5Config"),
+        ("opt", "OPTConfig"),
         ("megatron-bert", "MegatronBertConfig"),
         ("mixtral", "MixtralConfig"),
         ("owlvit", "OwlViTConfig"),
@@ -197,6 +198,7 @@
         ("mobilebert", "MobileBERT"),
         ("mpt", "MPT"),
         ("mt5", "MT5"),
+        ("opt", "OPT"),
         ("megatron-bert", "Megatron-BERT"),
         ("mistral", "Mistral"),
         ("mixtral", "Mixtral"),

mindone/transformers/models/auto/modeling_auto.py

@@ -65,6 +65,7 @@
         ("glm", "GlmModel"),
         ("glpn", "GLPNModel"),
         ("gpt2", "GPT2Model"),
+        ("opt", "OPTModel"),
         ("granite", "GraniteModel"),
         ("granitemoe", "GraniteMoeModel"),
         ("granitemoeshared", "GraniteMoeSharedModel"),
@@ -204,6 +205,7 @@
         ("granitemoe", "GraniteMoeForCausalLM"),
         ("granitemoeshared", "GraniteMoeSharedForCausalLM"),
         ("llama", "LlamaForCausalLM"),
+        ("opt", "OPTForCausalLM"),
         ("megatron-bert", "MegatronBertForCausalLM"),
         ("mistral", "MistralForCausalLM"),
         ("mpt", "MptForCausalLM"),
@@ -454,6 +456,7 @@
         ("starcoder2", "Starcoder2ForSequenceClassification"),
         ("canine", "CanineForSequenceClassification"),
         ("llama", "LlamaForSequenceClassification"),
+        ("opt", "OPTForSequenceClassification"),
         ("persimmon", "PersimmonForSequenceClassification"),
         ("mobilebert", "MobileBertForSequenceClassification"),
         ("convbert", "ConvBertForSequenceClassification"),
@@ -491,6 +494,7 @@
         ("mobilebert", "MobileBertForQuestionAnswering"),
         ("megatron-bert", "MegatronBertForQuestionAnswering"),
         ("mistral", "MistralForQuestionAnswering"),
+        ("opt", "OPTForQuestionAnswering"),
         ("qwen2", "Qwen2ForQuestionAnswering"),
         ("rembert", "RemBertForQuestionAnswering"),
         ("t5", "T5ForQuestionAnswering"),
@@ -503,7 +507,11 @@
     ]
 )
 
-MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict()
+MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
+    [
+        ("blip-2", "Blip2ForConditionalGeneration"),
+    ]
+)
 
 MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES = OrderedDict(
     [
@@ -611,6 +619,7 @@
         # Model for Zero Shot Image Classification mapping
         ("blip", "BlipModel"),
         ("siglip", "SiglipModel"),
+        ("blip-2", "Blip2ForImageTextRetrieval"),
     ]
 )
 

mindone/transformers/models/blip_2/__init__.py

@@ -14,10 +14,4 @@
 # 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_blip_2 import (
-    Blip2ForConditionalGeneration,
-    Blip2Model,
-    Blip2PreTrainedModel,
-    Blip2QFormerModel,
-    Blip2VisionModel,
-)
+from .modeling_blip_2 import *