o wi#I@sNddlmZmZddlZddlZddlmZddlmZddlm Z ddl m Z ddl m Z mZmZd d lmZmZd d lmZd d lmZd d lmZmZmZmZmZeeZGdddeZ GdddeZ!GdddeZ"GdddeZ#GdddeZ$GdddeZ%GdddeZ&GdddeZ'GdddeZ(gd Z)dS)!)OptionalUnionN)nn) DynamicCache)FlashAttentionKwargs)Unpack)auto_docstringcan_return_tuplelogging)Idefics3ConfigIdefics3VisionConfig)Idefics3ImageProcessor)Idefics3ImageProcessorFast)Idefics3BaseModelOutputWithPast Idefics3ForConditionalGeneration Idefics3ModelIdefics3PreTrainedModelIdefics3VisionTransformerc@eZdZdZdZdS)SmolVLMVisionConfiga This is the configuration class to store the configuration of a [`SmolVLMVisionModel`]. It is used to instantiate a SmolVLM vision encoder according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the SigLIP checkpoint [google/siglip-so400m-patch14-384](https://huggingface.co/google/siglip-so400m-patch14-384) used in SmolVLM [HuggingFaceTB/SmolVLM2-2.2B-Instruct](https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct). Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: hidden_size (`int`, *optional*, defaults to 1152): Dimensionality of the encoder layers and the pooler layer. intermediate_size (`int`, *optional*, defaults to 3072): Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder. num_hidden_layers (`int`, *optional*, defaults to 12): Number of hidden layers in the Transformer encoder. num_attention_heads (`int`, *optional*, defaults to 16): Number of attention heads for each attention layer in the Transformer encoder. num_channels (`int`, *optional*, defaults to 3): Number of channels in the input images. image_size (`int`, *optional*, defaults to 224): The size (resolution) of each image. patch_size (`int`, *optional*, defaults to 32): The size (resolution) of each patch. hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`): The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` `"quick_gelu"` are supported. layer_norm_eps (`float`, *optional*, defaults to 1e-06): The epsilon used by the layer normalization layers. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. Example: ```python >>> from transformers.models.smolvlm.modeling_smolvlm import SmolVLMVisionTransformer >>> from transformers.models.smolvlm.configuration_smolvlm import SmolVLMVisionConfig >>> # Initializing a SmolVLMVisionConfig with google/siglip-so400m-patch14-384 style configuration >>> configuration = SmolVLMVisionConfig() >>> # Initializing a SmolVLMVisionTransformer (with random weights) from the google/siglip-so400m-patch14-384 style configuration >>> model = SmolVLMVisionTransformer(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```smolvlm_visionN__name__ __module__ __qualname____doc__ model_typerrh/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/transformers/models/smolvlm/modular_smolvlm.pyr)s3rc@seZdZddZdS)SmolVLMPreTrainedModelcCst|jd|jj}t|tjtjfr,|jj j d|d|j dur*|j j dSdSt|tj rM|jj j d|d|jdurK|jj |j dSdSt|tjrb|jj d|j j dSdS)Ninitializer_range)meanstdg?)getattrconfigget_text_configr" isinstancerLinearConv2dweightdatanormal_biaszero_ Embedding padding_idx LayerNormfill_)selfmoduler%rrr _init_weightsbs    z$SmolVLMPreTrainedModel._init_weightsN)rrrr7rrrr r!as r!c@ eZdZdS)SmolVLMVisionTransformerNrrrrrrr r9rr9c@r) SmolVLMConfiga This is the configuration class to store the configuration of a [`SmolVLMModel`]. It is used to instantiate a SmolVLM model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the model of the SmolVLM [HuggingFaceTB/SmolVLM2-2.2B-Instruct](https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct) architecture. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should cache the key/value pairs of the attention mechanism. Only relevant if `config.is_decoder=True`. image_token_id (`int`, *optional*, defaults to 128257): The id of the "image" token. tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether or not to tie the word embeddings with the token embeddings. vision_config (`IdeficsVisionConfig` or `dict`, *optional*, defaults to `IdeficsVisionConfig`): Custom vision config or dict for the vision tower text_config (`PretrainedConfig` or `dict`, *optional*, defaults to `LlamaConfig`): Custom text config or dict for the text model scale_factor (`int`, *optional*, defaults to 2): The scale factor for the image encoder. pad_token_id (`int`, *optional*, defaults to 128002): The id of the padding token. Example: ```python >>> from transformers import SmolVLMModel, SmolVLMConfig >>> # Initializing configuration >>> configuration = SmolVLMConfig() >>> # Initializing a model from the configuration >>> model = SmolVLMModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```smolvlmNrrrrr r<vs%r<c@r8)SmolVLMImageProcessorNr:rrrr r>r;r>c@r8)SmolVLMImageProcessorFastNr:rrrr r?r;r?c@r8)SmolVLMBaseModelOutputWithPastNr:rrrr r@r;r@c#@s eZdZdZdejdejdejfddZddejd ejfd d Z e e d d              dde ejde ejde ejde e ejde ejde ejd e ejde ejde ede ede ede ede ejdeedeeeffddZdS) SmolVLMModelz A subclass of Idefics3Model. We do *not* remove or block the call to inputs_merger in forward. Instead, we override inputs_merger here with custom logic. input_ids inputs_embedsimage_hidden_statescCs|j\}}}||jk}|jdd}t||dkstd||}tjjj|j ddddd} | dd} |j dd} | d|} | d|} | d| }t |}|||| |ddf||<t | d||}|S)NdimrzCAt least one sample has tokens not divisible by patch_size.)rEr)value) shapeimage_token_idsumtorchall ValueErrorr functionalpadcumsum unsqueeze zeros_likewhere)r5rBrCrD_ patch_size image_masknum_image_tokensblocks_per_sampleoffsets block_offsetrow_cum chunk_idx local_idx block_idx image_embeds merged_embedsrrr inputs_mergers        zSmolVLMModel.inputs_mergerN pixel_valuespixel_attention_maskcs*j\}}}}}j||gjddRjdd}dkjdd|k} t| s3d| d<| |durOtjfd d d Dtjj d }n|j||g|jddR}|| }|j j j } |j d| | d } | j d| | d } | jdddk} |j| d} | j} || } | S)a Encodes images into continuous embeddings that can be forwarded to the language model. Args: pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`): The tensors corresponding to the input images. pixel_attention_mask (`torch.LongTensor`, *optional*): The attention mask indicating padded regions in the image. r NrEr#)rIrFTrcsg|]}j|qSr)rJ).0irdrr sz3SmolVLMModel.get_image_features..)rr r)sizedtypedevice) dimensionrlstep)rIrf)rdpatch_attention_mask)rJviewnumelrLany contiguousrMonesboolrnr' vision_configrWunfold vision_modellast_hidden_state connector)r5rdre batch_size num_images num_channelsheightwidthnb_values_per_imagereal_images_indsrWpatches_subgridrqrDrrjr get_image_featuress.      zSmolVLMModel.get_image_featuresa Inputs fed to the model can have an arbitrary number of images. To account for this, pixel_values fed to the model have image padding -> (batch_size, max_num_images, 3, max_heights, max_widths) where max_num_images is the maximum number of images among the batch_size samples in the batch. Padding images are not needed beyond padding the pixel_values at the entrance of the model. For efficiency, we only pass through the vision_model's forward the real images by discarding the padding images i.e. pixel_values of size (image_batch_size, 3, height, width) where image_batch_size would be 7 when num_images_per_sample=[1, 3, 1, 2] and max_num_images would be 3. ) custom_introattention_mask position_idspast_key_values use_cacheoutput_attentionsoutput_hidden_states return_dictcache_positionkwargsreturnc Ks| dur| n|jj} | dur| n|jj} | dur| n|jj} | dur$| n|jj} |jr8|jjr8| r8t dd} |durB|j \}}n|durM|j \}}}nt dd}| r`|dur\t }| }|durp|durp|dkrpt d|dur|j||j}|dur|durt d|dur||||j}n |dur|j|j|jd}|dur|dur|j|||d}|jd ||||| | | d | d |}t|j|j|j|j|d S) NzZ`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...Fz5You have to specify either input_ids or inputs_embedsrzWWhen first calling the model, if input_embeds are passed, input_ids should not be None.zMYou cannot specify both pixel_values and image_hidden_states at the same time)rmrn)rBrCrDT) rCrrrrrrrr)r{r hidden_states attentionsrDr)r'rrruse_return_dicttraining text_modelgradient_checkpointinglogger warning_oncerJrOrget_seq_lengthget_input_embeddingstornrrmrcr@r{rrr)r5rBrrrrCrdrerDrrrrrrr} seq_lengthrVpast_seen_tokensoutputsrrr forwardsp   zSmolVLMModel.forward)N) NNNNNNNNNNNNN)rrrrrM LongTensorTensorrc FloatTensorrr r rlist BoolTensorrwrrrtupler@rrrrr rAst .        rAcs(eZdZfddZfddZZS)SmolVLMForConditionalGenerationcs<t|t||_tj|jj|jjdd|_ | dS)NF)r/) super__init__rAmodelrr* text_config hidden_size vocab_sizelm_head post_init)r5r' __class__rr r_s   z(SmolVLMForConditionalGeneration.__init__c stjdi|dS)a Example: ```python >>> import requests >>> import torch >>> from PIL import Image >>> from io import BytesIO >>> from transformers import AutoProcessor, AutoModelForImageTextToText >>> from transformers.image_utils import load_image >>> # Note that passing the image urls (instead of the actual pil images) to the processor is also possible >>> image1 = load_image("https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg") >>> image2 = load_image("https://cdn.britannica.com/59/94459-050-DBA42467/Skyline-Chicago.jpg") >>> image3 = load_image("https://cdn.britannica.com/68/170868-050-8DDE8263/Golden-Gate-Bridge-San-Francisco.jpg") >>> processor = AutoProcessor.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct") >>> model = AutoModelForImageTextToText.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct", torch_dtype=torch.bfloat16, device_map="auto") >>> # Create inputs >>> messages = [ ... { ... "role": "user", ... "content": [ ... {"type": "video", "path": path/to/video}, ... {"type": "text", "text": "What is happening in this video?"}, ... ] ... } ... ] >>> inputs = processor.apply_chat_template([messages], add_generation_prompt=True) >>> # Generate >>> generated_ids = model.generate(**inputs, max_new_tokens=256) >>> generated_texts = processor.batch_decode(generated_ids, skip_special_tokens=True) >>> print(generated_texts) ```Nr)rr)r5 super_kwargsrrr res(z'SmolVLMForConditionalGeneration.forward)rrrrr __classcell__rrrr r^s r)rr<r>r?rr!rAr9)*typingrrrMtorch.utils.checkpointr cache_utilsrmodeling_flash_attention_utilsrprocessing_utilsrutilsr r r idefics3.configuration_idefics3r r"idefics3.image_processing_idefics3r'idefics3.image_processing_idefics3_fastridefics3.modeling_idefics3rrrrr get_loggerrrrr!r9r<r>r?r@rAr__all__rrrr s0       8*3 2