GLM-V update with new processor (#42122)

* init

* update

* add

* Update video_processing_glm46v.py

* update doc

* Update modular_glm46v.py

* 2

* Update processing_glm46v.py

* 21

* Update check_repo.py

* Update check_repo.py

* Update test_processor_glm46v.py

* Update modeling_auto.py

* update

* Update glm46v.md

* Update configuration_auto.py

* 2

* update with glm46v import

* uppercase

* upload

* upload

* upload with modular

* 1

* -

* update

* 1

* 2

* 1

* 2

* 2

* 1

* update config

* 1

* update as automoel

* 1

* try remove

* delete

* delete

* test

* update

* 1

* Update modular_glm46v.py

* Update test_modeling_glm46v.py

* update 1513

* 1

* use PreTrainedConfig

* Update modular_glm46v.py

* Update configuration_glm46v.py

* model_type = "glm46v"

* remove glm46v_text

* Update image_processing_auto.py

* 1

* update readme

* GLM-4.6V

* update

* update

* Update __init__.py

* update

* update doc

* Update check_docstrings.py

* update doc

* fix copies for tied weight keys!

* more fixup

---------

Co-authored-by: Raushan Turganbay <raushan@huggingface.co>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Arthur <arthur.zucker@gmail.com>

Author: 3 people

docs/source/en/_toctree.yml

@@ -1066,6 +1066,8 @@
         title: Gemma3n
       - local: model_doc/git
         title: GIT
+      - local: model_doc/glm46v
+        title: Glm46V
       - local: model_doc/glm4v
         title: glm4v
       - local: model_doc/glm4v_moe

docs/source/en/model_doc/glm46v.md

@@ -0,0 +1,34 @@
+# GLM-4.6V
+
+## Glm46VConfig
+
+[[autodoc]] Glm46VConfig
+
+## Glm46VImageProcessor
+
+[[autodoc]] Glm46VImageProcessor
+    - preprocess
+
+## Glm46VVideoProcessor
+
+[[autodoc]] Glm46VVideoProcessor
+    - preprocess
+
+## Glm46VImageProcessorFast
+
+[[autodoc]] Glm46VImageProcessorFast
+    - preprocess
+
+## Glm46VProcessor
+
+[[autodoc]] Glm46VProcessor
+
+## Glm46VModel
+
+[[autodoc]] Glm46VModel
+    - forward
+
+## Glm46VForConditionalGeneration
+
+[[autodoc]] Glm46VForConditionalGeneration
+    - forward

docs/source/en/model_doc/glm4v.md

@@ -170,6 +170,11 @@ print(output_text)
 
 [[autodoc]] Glm4vConfig
 
+
+## Glm4vVisionConfig
+
+[[autodoc]] Glm4vVisionConfig
+
 ## Glm4vTextConfig
 
 [[autodoc]] Glm4vTextConfig
@@ -193,6 +198,11 @@ print(output_text)
 
 [[autodoc]] Glm4vProcessor
 
+## Glm4vVisionModel
+
+[[autodoc]] Glm4vVisionModel
+    - forward
+
 ## Glm4vTextModel
 
 [[autodoc]] Glm4vTextModel

docs/source/en/model_doc/glm4v_moe.md

@@ -22,7 +22,7 @@ rendered properly in your Markdown viewer.
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">    </div>
 </div>
 
-# Glm4vMoe
+# Glm4vMoeMoe
 
 ## Overview
 
@@ -48,10 +48,20 @@ The model also introduces a **Thinking Mode** switch, allowing users to balance
 
 [[autodoc]] Glm4vMoeConfig
 
+
+## Glm4vMoeVisionConfig
+
+[[autodoc]] Glm4vMoeVisionConfig
+
 ## Glm4vMoeTextConfig
 
 [[autodoc]] Glm4vMoeTextConfig
 
+## Glm4vMoeVisionModel
+
+[[autodoc]] Glm4vMoeVisionModel
+    - forward
+
 ## Glm4vMoeTextModel
 
 [[autodoc]] Glm4vMoeTextModel
@@ -65,4 +75,4 @@ The model also introduces a **Thinking Mode** switch, allowing users to balance
 ## Glm4vMoeForConditionalGeneration
 
 [[autodoc]] Glm4vMoeForConditionalGeneration
-    - forward
+    - forward

src/transformers/models/__init__.py

@@ -142,6 +142,9 @@
     from .git import *
     from .glm import *
     from .glm4 import *
+    from .glm4v import *
+    from .glm4v_moe import *
+    from .glm46v import *
     from .glpn import *
     from .got_ocr2 import *
     from .gpt2 import *

src/transformers/models/auto/configuration_auto.py

@@ -172,11 +172,14 @@
         ("git", "GitConfig"),
         ("glm", "GlmConfig"),
         ("glm4", "Glm4Config"),
+        ("glm46v", "Glm46VConfig"),
         ("glm4_moe", "Glm4MoeConfig"),
         ("glm4v", "Glm4vConfig"),
         ("glm4v_moe", "Glm4vMoeConfig"),
         ("glm4v_moe_text", "Glm4vMoeTextConfig"),
+        ("glm4v_moe_vision", "Glm4vMoeVisionConfig"),
         ("glm4v_text", "Glm4vTextConfig"),
+        ("glm4v_vision", "Glm4vVisionConfig"),
         ("glpn", "GLPNConfig"),
         ("got_ocr2", "GotOcr2Config"),
         ("gpt-sw3", "GPT2Config"),
@@ -620,11 +623,14 @@
         ("git", "GIT"),
         ("glm", "GLM"),
         ("glm4", "GLM4"),
+        ("glm46v", "Glm46V"),
         ("glm4_moe", "Glm4MoE"),
         ("glm4v", "GLM4V"),
         ("glm4v_moe", "GLM4VMOE"),
         ("glm4v_moe_text", "GLM4VMOE"),
+        ("glm4v_moe_vision", "Glm4vMoeVisionModel"),
         ("glm4v_text", "GLM4V"),
+        ("glm4v_vision", "Glm4vVisionModel"),
         ("glpn", "GLPN"),
         ("got_ocr2", "GOT-OCR2"),
         ("gpt-sw3", "GPT-Sw3"),
@@ -983,6 +989,8 @@
         ("gemma3n_audio", "gemma3n"),
         ("gemma3n_text", "gemma3n"),
         ("gemma3n_vision", "gemma3n"),
+        ("glm4v_vision", "glm4v"),
+        ("glm4v_moe_vision", "glm4v_moe"),
         ("glm4v_text", "glm4v"),
         ("glm4v_moe_text", "glm4v_moe"),
         ("idefics3_vision", "idefics3"),

src/transformers/models/auto/image_processing_auto.py

@@ -109,6 +109,7 @@
             ("gemma3", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")),
             ("gemma3n", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
             ("git", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
+            ("glm46v", ("Glm46VImageProcessor", "Glm46VImageProcessorFast")),
             ("glm4v", ("Glm4vImageProcessor", "Glm4vImageProcessorFast")),
             ("glpn", ("GLPNImageProcessor", "GLPNImageProcessorFast")),
             ("got_ocr2", ("GotOcr2ImageProcessor", "GotOcr2ImageProcessorFast")),

src/transformers/models/auto/modeling_auto.py

@@ -175,11 +175,14 @@ class _BaseModelWithGenerate(PreTrainedModel, GenerationMixin):
         ("git", "GitModel"),
         ("glm", "GlmModel"),
         ("glm4", "Glm4Model"),
+        ("glm46v", "Glm46VModel"),
         ("glm4_moe", "Glm4MoeModel"),
         ("glm4v", "Glm4vModel"),
         ("glm4v_moe", "Glm4vMoeModel"),
         ("glm4v_moe_text", "Glm4vMoeTextModel"),
+        ("glm4v_moe_vision", "Glm4vMoeVisionModel"),
         ("glm4v_text", "Glm4vTextModel"),
+        ("glm4v_vision", "Glm4vVisionModel"),
         ("glpn", "GLPNModel"),
         ("got_ocr2", "GotOcr2Model"),
         ("gpt-sw3", "GPT2Model"),
@@ -1032,6 +1035,7 @@ class _BaseModelWithGenerate(PreTrainedModel, GenerationMixin):
         ("gemma3", "Gemma3ForConditionalGeneration"),
         ("gemma3n", "Gemma3nForConditionalGeneration"),
         ("git", "GitForCausalLM"),
+        ("glm46v", "Glm46VForConditionalGeneration"),
         ("glm4v", "Glm4vForConditionalGeneration"),
         ("glm4v_moe", "Glm4vMoeForConditionalGeneration"),
         ("got_ocr2", "GotOcr2ForConditionalGeneration"),

src/transformers/models/auto/processing_auto.py

@@ -75,6 +75,7 @@
         ("gemma3", "Gemma3Processor"),
         ("gemma3n", "Gemma3nProcessor"),
         ("git", "GitProcessor"),
+        ("glm46v", "Glm46VProcessor"),
         ("glm4v", "Glm4vProcessor"),
         ("glm4v_moe", "Glm4vProcessor"),
         ("got_ocr2", "GotOcr2Processor"),

src/transformers/models/auto/tokenization_auto.py

@@ -308,6 +308,7 @@
         ("git", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
         ("glm", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),
         ("glm4", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),
+        ("glm46v", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),
         ("glm4_moe", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),
         ("glm4v", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),
         ("glm4v_moe", (None, "PreTrainedTokenizerFast" if is_tokenizers_available() else None)),