fixed mistakes caused by select all paste

innovationcore · Dec 6, 2024 · 0ed1114 · 0ed1114
1 parent fc8324a
commit 0ed1114
Show file tree

Hide file tree

Showing 8 changed files with 1,312 additions and 433 deletions.
diff --git a/src/transformers/__init__.py b/src/transformers/__init__.py
diff --git a/src/transformers/commands/add_new_model_like.py b/src/transformers/commands/add_new_model_like.py
@@ -766,6 +766,7 @@ def retrieve_info_for_model(model_type, frameworks: Optional[List[str]] = None):
         image_processor_class = image_processor_classes[0]  # we take the slow image processor class.
     else:
         image_processor_class = image_processor_classes
+
     feature_extractor_class = auto_module.feature_extraction_auto.FEATURE_EXTRACTOR_MAPPING_NAMES.get(model_type, None)
     processor_class = auto_module.processing_auto.PROCESSOR_MAPPING_NAMES.get(model_type, None)
 
@@ -1714,4 +1715,4 @@ def get_user_input():
         )
         frameworks = list(set(frameworks.split(" ")))
 
-    return (old_model_type, model_patterns, add_copied_from, frameworks, old_checkpoint)
+    return (old_model_type, model_patterns, add_copied_from, frameworks, old_checkpoint)
diff --git a/src/transformers/models/__init__.py b/src/transformers/models/__init__.py
@@ -16,6 +16,7 @@
     albert,
     align,
     altclip,
+    aria,
     audio_spectrogram_transformer,
     auto,
     autoformer,
@@ -98,6 +99,7 @@
     gemma,
     gemma2,
     git,
+    glm,
     glpn,
     gpt2,
     gpt_bigcode,
@@ -106,6 +108,8 @@
     gpt_neox_japanese,
     gpt_sw3,
     gptj,
+    granite,
+    granitemoe,
     grounding_dino,
     groupvit,
     herbert,
@@ -114,6 +118,8 @@
     ibert,
     idefics,
     idefics2,
+    idefics3,
+    ijepa,
     imagegpt,
     informer,
     instructblip,
@@ -132,6 +138,7 @@
     llava,
     llava_next,
     llava_next_video,
+    llava_onevision,
     longformer,
     longt5,
     luke,
@@ -148,27 +155,34 @@
     megatron_bert,
     megatron_gpt2,
     mgp_str,
+    mimi,
     mistral,
     mixtral,
+    mllama,
     mluke,
     mobilebert,
     mobilenet_v1,
     mobilenet_v2,
     mobilevit,
     mobilevitv2,
+    moshi,
     mpnet,
     mpt,
     mra,
     mt5,
     musicgen,
     musicgen_melody,
     mvp,
+    myt5,
     nemotron,
     nllb,
     nllb_moe,
     nougat,
     nystromformer,
     olmo,
+    olmo2,
+    olmoe,
+    omdet_turbo,
     oneformer,
     openai,
     opt,
@@ -183,8 +197,10 @@
     persimmon,
     phi,
     phi3,
+    phimoe,
     phobert,
     pix2struct,
+    pixtral,
     plbart,
     poolformer,
     pop2piano,
@@ -194,6 +210,7 @@
     qwen2,
     qwen2_audio,
     qwen2_moe,
+    qwen2_vl,
     rag,
     recurrent_gemma,
     reformer,
@@ -271,5 +288,6 @@
     xmod,
     yolos,
     yoso,
+    zamba,
     zoedepth,
-)
+)
diff --git a/src/transformers/models/auto/configuration_auto.py b/src/transformers/models/auto/configuration_auto.py
@@ -35,6 +35,8 @@
         ("albert", "AlbertConfig"),
         ("align", "AlignConfig"),
         ("altclip", "AltCLIPConfig"),
+        ("aria", "AriaConfig"),
+        ("aria_text", "AriaTextConfig"),
         ("audio-spectrogram-transformer", "ASTConfig"),
         ("autoformer", "AutoformerConfig"),
         ("bark", "BarkConfig"),
@@ -60,6 +62,7 @@
         ("chinese_clip_vision_model", "ChineseCLIPVisionConfig"),
         ("clap", "ClapConfig"),
         ("clip", "CLIPConfig"),
+        ("clip_text_model", "CLIPTextConfig"),
         ("clip_vision_model", "CLIPVisionConfig"),
         ("clipseg", "CLIPSegConfig"),
         ("clvp", "ClvpConfig"),
@@ -114,6 +117,7 @@
         ("gemma", "GemmaConfig"),
         ("gemma2", "Gemma2Config"),
         ("git", "GitConfig"),
+        ("glm", "GlmConfig"),
         ("glpn", "GLPNConfig"),
         ("gpt-sw3", "GPT2Config"),
         ("gpt2", "GPT2Config"),
@@ -123,6 +127,8 @@
         ("gpt_neox_japanese", "GPTNeoXJapaneseConfig"),
         ("gptj", "GPTJConfig"),
         ("gptsan-japanese", "GPTSanJapaneseConfig"),
+        ("granite", "GraniteConfig"),
+        ("granitemoe", "GraniteMoeConfig"),
         ("graphormer", "GraphormerConfig"),
         ("grounding-dino", "GroundingDinoConfig"),
         ("groupvit", "GroupViTConfig"),
@@ -131,6 +137,9 @@
         ("ibert", "IBertConfig"),
         ("idefics", "IdeficsConfig"),
         ("idefics2", "Idefics2Config"),
+        ("idefics3", "Idefics3Config"),
+        ("idefics3_vision", "Idefics3VisionConfig"),
+        ("ijepa", "IJepaConfig"),
         ("imagegpt", "ImageGPTConfig"),
         ("informer", "InformerConfig"),
         ("instructblip", "InstructBlipConfig"),
@@ -149,6 +158,7 @@
         ("llava", "LlavaConfig"),
         ("llava_next", "LlavaNextConfig"),
         ("llava_next_video", "LlavaNextVideoConfig"),
+        ("llava_onevision", "LlavaOnevisionConfig"),
         ("longformer", "LongformerConfig"),
         ("longt5", "LongT5Config"),
         ("luke", "LukeConfig"),
@@ -166,13 +176,16 @@
         ("mega", "MegaConfig"),
         ("megatron-bert", "MegatronBertConfig"),
         ("mgp-str", "MgpstrConfig"),
+        ("mimi", "MimiConfig"),
         ("mistral", "MistralConfig"),
         ("mixtral", "MixtralConfig"),
+        ("mllama", "MllamaConfig"),
         ("mobilebert", "MobileBertConfig"),
         ("mobilenet_v1", "MobileNetV1Config"),
         ("mobilenet_v2", "MobileNetV2Config"),
         ("mobilevit", "MobileViTConfig"),
         ("mobilevitv2", "MobileViTV2Config"),
+        ("moshi", "MoshiConfig"),
         ("mpnet", "MPNetConfig"),
         ("mpt", "MptConfig"),
         ("mra", "MraConfig"),
@@ -187,6 +200,9 @@
         ("nougat", "VisionEncoderDecoderConfig"),
         ("nystromformer", "NystromformerConfig"),
         ("olmo", "OlmoConfig"),
+        ("olmo2", "Olmo2Config"),
+        ("olmoe", "OlmoeConfig"),
+        ("omdet-turbo", "OmDetTurboConfig"),
         ("oneformer", "OneFormerConfig"),
         ("open-llama", "OpenLlamaConfig"),
         ("openai-gpt", "OpenAIGPTConfig"),
@@ -202,7 +218,9 @@
         ("persimmon", "PersimmonConfig"),
         ("phi", "PhiConfig"),
         ("phi3", "Phi3Config"),
+        ("phimoe", "PhimoeConfig"),
         ("pix2struct", "Pix2StructConfig"),
+        ("pixtral", "PixtralVisionConfig"),
         ("plbart", "PLBartConfig"),
         ("poolformer", "PoolFormerConfig"),
         ("pop2piano", "Pop2PianoConfig"),
@@ -214,6 +232,7 @@
         ("qwen2_audio", "Qwen2AudioConfig"),
         ("qwen2_audio_encoder", "Qwen2AudioEncoderConfig"),
         ("qwen2_moe", "Qwen2MoeConfig"),
+        ("qwen2_vl", "Qwen2VLConfig"),
         ("rag", "RagConfig"),
         ("realm", "RealmConfig"),
         ("recurrent_gemma", "RecurrentGemmaConfig"),
@@ -300,6 +319,7 @@
         ("xmod", "XmodConfig"),
         ("yolos", "YolosConfig"),
         ("yoso", "YosoConfig"),
+        ("zamba", "ZambaConfig"),
         ("zoedepth", "ZoeDepthConfig"),
     ]
 )
@@ -311,6 +331,8 @@
         ("albert", "ALBERT"),
         ("align", "ALIGN"),
         ("altclip", "AltCLIP"),
+        ("aria", "Aria"),
+        ("aria_text", "AriaText"),
         ("audio-spectrogram-transformer", "Audio Spectrogram Transformer"),
         ("autoformer", "Autoformer"),
         ("bark", "Bark"),
@@ -342,6 +364,7 @@
         ("chinese_clip_vision_model", "ChineseCLIPVisionModel"),
         ("clap", "CLAP"),
         ("clip", "CLIP"),
+        ("clip_text_model", "CLIPTextModel"),
         ("clip_vision_model", "CLIPVisionModel"),
         ("clipseg", "CLIPSeg"),
         ("clvp", "CLVP"),
@@ -403,6 +426,7 @@
         ("gemma", "Gemma"),
         ("gemma2", "Gemma2"),
         ("git", "GIT"),
+        ("glm", "GLM"),
         ("glpn", "GLPN"),
         ("gpt-sw3", "GPT-Sw3"),
         ("gpt2", "OpenAI GPT-2"),
@@ -412,6 +436,8 @@
         ("gpt_neox_japanese", "GPT NeoX Japanese"),
         ("gptj", "GPT-J"),
         ("gptsan-japanese", "GPTSAN-japanese"),
+        ("granite", "Granite"),
+        ("granitemoe", "GraniteMoeMoe"),
         ("graphormer", "Graphormer"),
         ("grounding-dino", "Grounding DINO"),
         ("groupvit", "GroupViT"),
@@ -421,6 +447,9 @@
         ("ibert", "I-BERT"),
         ("idefics", "IDEFICS"),
         ("idefics2", "Idefics2"),
+        ("idefics3", "Idefics3"),
+        ("idefics3_vision", "Idefics3VisionTransformer"),
+        ("ijepa", "I-JEPA"),
         ("imagegpt", "ImageGPT"),
         ("informer", "Informer"),
         ("instructblip", "InstructBLIP"),
@@ -442,6 +471,7 @@
         ("llava", "LLaVa"),
         ("llava_next", "LLaVA-NeXT"),
         ("llava_next_video", "LLaVa-NeXT-Video"),
+        ("llava_onevision", "LLaVA-Onevision"),
         ("longformer", "Longformer"),
         ("longt5", "LongT5"),
         ("luke", "LUKE"),
@@ -463,22 +493,26 @@
         ("megatron-bert", "Megatron-BERT"),
         ("megatron_gpt2", "Megatron-GPT2"),
         ("mgp-str", "MGP-STR"),
+        ("mimi", "Mimi"),
         ("mistral", "Mistral"),
         ("mixtral", "Mixtral"),
+        ("mllama", "Mllama"),
         ("mluke", "mLUKE"),
         ("mms", "MMS"),
         ("mobilebert", "MobileBERT"),
         ("mobilenet_v1", "MobileNetV1"),
         ("mobilenet_v2", "MobileNetV2"),
         ("mobilevit", "MobileViT"),
         ("mobilevitv2", "MobileViTV2"),
+        ("moshi", "Moshi"),
         ("mpnet", "MPNet"),
         ("mpt", "MPT"),
         ("mra", "MRA"),
         ("mt5", "MT5"),
         ("musicgen", "MusicGen"),
         ("musicgen_melody", "MusicGen Melody"),
         ("mvp", "MVP"),
+        ("myt5", "myt5"),
         ("nat", "NAT"),
         ("nemotron", "Nemotron"),
         ("nezha", "Nezha"),
@@ -487,6 +521,9 @@
         ("nougat", "Nougat"),
         ("nystromformer", "Nyströmformer"),
         ("olmo", "OLMo"),
+        ("olmo2", "OLMo2"),
+        ("olmoe", "OLMoE"),
+        ("omdet-turbo", "OmDet-Turbo"),
         ("oneformer", "OneFormer"),
         ("open-llama", "OpenLlama"),
         ("openai-gpt", "OpenAI GPT"),
@@ -502,8 +539,10 @@
         ("persimmon", "Persimmon"),
         ("phi", "Phi"),
         ("phi3", "Phi3"),
+        ("phimoe", "Phimoe"),
         ("phobert", "PhoBERT"),
         ("pix2struct", "Pix2Struct"),
+        ("pixtral", "Pixtral"),
         ("plbart", "PLBart"),
         ("poolformer", "PoolFormer"),
         ("pop2piano", "Pop2Piano"),
@@ -515,6 +554,7 @@
         ("qwen2_audio", "Qwen2Audio"),
         ("qwen2_audio_encoder", "Qwen2AudioEncoder"),
         ("qwen2_moe", "Qwen2MoE"),
+        ("qwen2_vl", "Qwen2VL"),
         ("rag", "RAG"),
         ("realm", "REALM"),
         ("recurrent_gemma", "RecurrentGemma"),
@@ -608,6 +648,7 @@
         ("xmod", "X-MOD"),
         ("yolos", "YOLOS"),
         ("yoso", "YOSO"),
+        ("zamba", "Zamba"),
         ("zoedepth", "ZoeDepth"),
     ]
 )
@@ -653,6 +694,9 @@
         ("xclip", "x_clip"),
         ("clip_vision_model", "clip"),
         ("qwen2_audio_encoder", "qwen2_audio"),
+        ("clip_text_model", "clip"),
+        ("aria_text", "aria"),
+        ("idefics3_vision", "idefics3"),
         ("siglip_vision_model", "siglip"),
         ("chinese_clip_vision_model", "chinese_clip"),
         ("rt_detr_resnet", "rt_detr"),
@@ -1037,4 +1081,4 @@ def register(model_type, config, exist_ok=False):
                 f"you passed (config has {config.model_type} and you passed {model_type}. Fix one of those so they "
                 "match!"
             )
-        CONFIG_MAPPING.register(model_type, config, exist_ok=exist_ok)
+        CONFIG_MAPPING.register(model_type, config, exist_ok=exist_ok)