use is_composition for pixtral

Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>

docs/source/en/model_doc/pixtral.md

@@ -78,10 +78,6 @@ output = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up
 
 [[autodoc]] PixtralVisionConfig
 
-## PixtralTextConfig
-
-[[autodoc]] PixtralTextConfig
-
 ## PixtralVisionModel
 
 [[autodoc]] PixtralVisionModel

src/transformers/__init__.py

@@ -700,7 +700,7 @@ _import_structure = {
         "Pix2StructTextConfig",
         "Pix2StructVisionConfig",
     ],
-    "models.pixtral": ["PixtralProcessor", "PixtralVisionConfig", "PixtralTextConfig"],
+    "models.pixtral": ["PixtralProcessor", "PixtralVisionConfig"],
     "models.plbart": ["PLBartConfig"],
     "models.poolformer": ["PoolFormerConfig"],
     "models.pop2piano": ["Pop2PianoConfig"],

src/transformers/models/auto/configuration_auto.py

@@ -232,7 +232,6 @@ CONFIG_MAPPING_NAMES = OrderedDict(
         ("phimoe", "PhimoeConfig"),
         ("pix2struct", "Pix2StructConfig"),
         ("pixtral", "PixtralVisionConfig"),
-        ("pixtral_text", "PixtralTextConfig"),
         ("plbart", "PLBartConfig"),
         ("poolformer", "PoolFormerConfig"),
         ("pop2piano", "Pop2PianoConfig"),
@@ -575,7 +574,6 @@ MODEL_NAMES_MAPPING = OrderedDict(
         ("phobert", "PhoBERT"),
         ("pix2struct", "Pix2Struct"),
         ("pixtral", "Pixtral"),
-        ("pixtral_text", "PixtralMistral"),
         ("plbart", "PLBart"),
         ("poolformer", "PoolFormer"),
         ("pop2piano", "Pop2Piano"),
@@ -742,7 +740,6 @@ SPECIAL_MODEL_TYPE_TO_MODULE_NAME = OrderedDict(
         ("chinese_clip_vision_model", "chinese_clip"),
         ("rt_detr_resnet", "rt_detr"),
         ("granitevision", "llava_next"),
-        ("pixtral_text", "pixtral"),
     ]
 )
 

src/transformers/models/auto/modeling_auto.py

@@ -555,7 +555,6 @@ MODEL_FOR_CAUSAL_LM_MAPPING_NAMES = OrderedDict(
         ("phi", "PhiForCausalLM"),
         ("phi3", "Phi3ForCausalLM"),
         ("phimoe", "PhimoeForCausalLM"),
-        ("pixtral_text", "MistralForCausalLM"),
         ("plbart", "PLBartForCausalLM"),
         ("prophetnet", "ProphetNetForCausalLM"),
         ("qdqbert", "QDQBertLMHeadModel"),

src/transformers/models/llava/configuration_llava.py

@@ -78,6 +78,7 @@ class LlavaConfig(PretrainedConfig):
 
     model_type = "llava"
     sub_configs = {"text_config": AutoConfig, "vision_config": AutoConfig}
+    is_composition = True
 
     def __init__(
         self,

src/transformers/models/pixtral/configuration_pixtral.py

@@ -14,7 +14,6 @@
 """Pixtral model configuration"""
 
 from ...configuration_utils import PretrainedConfig
-from ...models.mistral.configuration_mistral import MistralConfig
 from ...utils import logging
 
 
@@ -104,116 +103,4 @@ class PixtralVisionConfig(PretrainedConfig):
         self.initializer_range = initializer_range
 
 
-class PixtralTextConfig(MistralConfig):
-    r"""
-    TODO
-
-    Args:
-        vocab_size (`int`, *optional*, defaults to 32000):
-            Vocabulary size of the Mistral model. Defines the number of different tokens that can be represented by the
-            `inputs_ids` passed when calling [`MistralModel`]
-        hidden_size (`int`, *optional*, defaults to 4096):
-            Dimension of the hidden representations.
-        intermediate_size (`int`, *optional*, defaults to 14336):
-            Dimension of the MLP representations.
-        num_hidden_layers (`int`, *optional*, defaults to 32):
-            Number of hidden layers in the Transformer encoder.
-        num_attention_heads (`int`, *optional*, defaults to 32):
-            Number of attention heads for each attention layer in the Transformer encoder.
-        num_key_value_heads (`int`, *optional*, defaults to 8):
-            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
-            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
-            `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When
-            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
-            by meanpooling all the original heads within that group. For more details checkout [this
-            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
-        head_dim (`int`, *optional*, defaults to `hidden_size // num_attention_heads`):
-            The attention head dimension.
-        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
-            The non-linear activation function (function or string) in the decoder.
-        max_position_embeddings (`int`, *optional*, defaults to `4096*32`):
-            The maximum sequence length that this model might ever be used with. Mistral's sliding window attention
-            allows sequence of up to 4096*32 tokens.
-        initializer_range (`float`, *optional*, defaults to 0.02):
-            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
-        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
-            The epsilon used by the rms normalization layers.
-        use_cache (`bool`, *optional*, defaults to `True`):
-            Whether or not the model should return the last key/values attentions (not used by all models). Only
-            relevant if `config.is_decoder=True`.
-        pad_token_id (`int`, *optional*):
-            The id of the padding token.
-        bos_token_id (`int`, *optional*, defaults to 1):
-            The id of the "beginning-of-sequence" token.
-        eos_token_id (`int`, *optional*, defaults to 2):
-            The id of the "end-of-sequence" token.
-        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
-            Whether the model's input and output word embeddings should be tied.
-        rope_theta (`float`, *optional*, defaults to 10000.0):
-            The base period of the RoPE embeddings.
-        sliding_window (`int`, *optional*, defaults to 4096):
-            Sliding window attention window size. If not specified, will default to `4096`.
-        attention_dropout (`float`, *optional*, defaults to 0.0):
-            The dropout ratio for the attention probabilities.
-
-    ```python
-    >>> TODO
-    ```"""
-
-    model_type = "pixtral_text"
-
-    def __init__(
-        self,
-        vocab_size=32000,
-        hidden_size=4096,
-        intermediate_size=14336,
-        num_hidden_layers=32,
-        num_attention_heads=32,
-        num_key_value_heads=8,
-        head_dim=None,
-        hidden_act="silu",
-        max_position_embeddings=4096 * 32,
-        initializer_range=0.02,
-        rms_norm_eps=1e-6,
-        use_cache=True,
-        pad_token_id=None,
-        bos_token_id=1,
-        eos_token_id=2,
-        tie_word_embeddings=False,
-        rope_theta=10000.0,
-        sliding_window=4096,
-        attention_dropout=0.0,
-        **kwargs,
-    ):
-        self.vocab_size = vocab_size
-        self.max_position_embeddings = max_position_embeddings
-        self.hidden_size = hidden_size
-        self.intermediate_size = intermediate_size
-        self.num_hidden_layers = num_hidden_layers
-        self.num_attention_heads = num_attention_heads
-        self.sliding_window = sliding_window
-        self.head_dim = head_dim  # as opposed to MistralConfig, do not auto-populate
-
-        # for backward compatibility
-        if num_key_value_heads is None:
-            num_key_value_heads = num_attention_heads
-
-        self.num_key_value_heads = num_key_value_heads
-        self.hidden_act = hidden_act
-        self.initializer_range = initializer_range
-        self.rms_norm_eps = rms_norm_eps
-        self.use_cache = use_cache
-        self.rope_theta = rope_theta
-        self.attention_dropout = attention_dropout
-
-        PretrainedConfig.__init__(
-            self,
-            pad_token_id=pad_token_id,
-            bos_token_id=bos_token_id,
-            eos_token_id=eos_token_id,
-            tie_word_embeddings=tie_word_embeddings,
-            **kwargs,
-        )
-
-
-__all__ = ["PixtralVisionConfig", "PixtralTextConfig"]
+__all__ = ["PixtralVisionConfig"]

tests/models/llava/test_configuration_llava.py

@@ -32,8 +32,7 @@ class LlavaConfigTest(unittest.TestCase):
         }
 
         text_config = {
-            # "model_type": "mistral",
-            "model_type": "pixtral_text",
+            "model_type": "mistral",
             "hidden_size": 5120,
             "head_dim": 128,
             "num_attention_heads": 32,

utils/check_table.py

@@ -180,7 +180,6 @@ MODEL_NAMES_TO_IGNORE = [
     "CLIPVisionModel",
     "Qwen2AudioEncoder",
     "SiglipVisionModel",
-    "PixtralMistral",  # not a real model
 ]