Hugging Face's logo

Transformers documentation

Qwen3-VL-Moe

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v4.56.2).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

This model was released on None and added to Hugging Face Transformers on 2025-08-17.

PyTorch FlashAttention SDPA

Qwen3-VL-Moe

Qwen3-VL is a multimodal vision-language model series, encompassing both dense and MoE variants, as well as Instruct and Thinking versions. Building upon its predecessors, Qwen3-VL delivers significant improvements in visual understanding while maintaining strong pure text capabilities. Key architectural advancements include: enhanced MRope with interleaved layout for better spatial-temporal modeling, DeepStack integration to effectively leverage multi-level features from the Vision Transformer (ViT), and improved video understanding through text-based time alignment—evolving from T-RoPE to text timestamp alignment for more precise temporal grounding. These innovations collectively enable Qwen3-VL to achieve superior performance in complex multimodal tasks.

Model usage

AutoModel 
import torch
from transformers import Qwen3VLMoeForConditionalGeneration, AutoProcessor

model = Qwen3VLMoeForConditionalGeneration.from_pretrained(
    "Qwen/Qwen3-VL-Moe",
    dtype=torch.float16,
    device_map="auto",
    attn_implementation="sdpa"
)
processor = AutoProcessor.from_pretrained("Qwen/Qwen3-VL-Moe")
messages = [
    {
        "role":"user",
        "content":[
            {
                "type":"image",
                "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
            },
            {
                "type":"text",
                "text":"Describe this image."
            }
        ]
    }

]

inputs = processor.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_dict=True,
    return_tensors="pt",
)
inputs.pop("token_type_ids", None)

generated_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids_trimmed = [
            out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
       generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text)

Qwen3VLMoeConfig

class transformers.Qwen3VLMoeConfig

< >

( text_config = None vision_config = None image_token_id = 151655 video_token_id = 151656 vision_start_token_id = 151652 vision_end_token_id = 151653 tie_word_embeddings = False **kwargs )

Parameters

  • text_config (Union[PreTrainedConfig, dict], optional, defaults to Qwen3VLMoeTextConfig) — The config object or dictionary of the text backbone.
  • vision_config (Union[PreTrainedConfig, dict], optional, defaults to Qwen3VLMoeVisionConfig) — The config object or dictionary of the vision backbone.
  • image_token_id (int, optional, defaults to 151655) — The image token index to encode the image prompt.
  • video_token_id (int, optional, defaults to 151656) — The video token index to encode the image prompt.
  • vision_start_token_id (int, optional, defaults to 151652) — The start token index to encode the image prompt.
  • vision_end_token_id (int, optional, defaults to 151653) — The end token index to encode the image prompt.
  • tie_word_embeddings (bool, optional, defaults to False) — Whether to tie the word embeddings.

This is the configuration class to store the configuration of a Qwen3VLMoeModel. It is used to instantiate a Qwen3-VL-MOE model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of Qwen3-VL-30B-A3B-Instruct Qwen/Qwen3-VL-30B-A3B-Instruct.

Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.

>>> from transformers import Qwen3VLMoeForConditionalGeneration, Qwen3VLMoeConfig

>>> # Initializing a Qwen3-VL-MOE style configuration
>>> configuration = Qwen3VLMoeConfig()

>>> # Initializing a model from the Qwen3-VL-30B-A3B style configuration
>>> model = Qwen3VLMoeForConditionalGeneration(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config

Qwen3VLMoeTextConfig

class transformers.Qwen3VLMoeTextConfig

< >

( vocab_size = 151936 hidden_size = 2048 intermediate_size = 5632 num_hidden_layers = 24 num_attention_heads = 16 num_key_value_heads = 16 hidden_act = 'silu' max_position_embeddings = 128000 initializer_range = 0.02 rms_norm_eps = 1e-06 use_cache = True tie_word_embeddings = False rope_theta = 5000000.0 attention_bias = False attention_dropout = 0.0 decoder_sparse_step = 1 moe_intermediate_size = 1408 num_experts_per_tok = 4 num_experts = 60 norm_topk_prob = True mlp_only_layers = None rope_scaling = None head_dim = None **kwargs )

Parameters

  • vocab_size (int, optional, defaults to 151936) — Vocabulary size of the Qwen2MoE model. Defines the number of different tokens that can be represented by the inputs_ids passed when calling Qwen2MoeModel
  • hidden_size (int, optional, defaults to 2048) — Dimension of the hidden representations.
  • intermediate_size (int, optional, defaults to 5632) — Dimension of the MLP representations.
  • num_hidden_layers (int, optional, defaults to 24) — 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_key_value_heads (int, optional, defaults to 16) — 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. If it is not specified, will default to 32.
  • 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 128000) — The maximum sequence length that this model might ever be used with.
  • 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.
  • 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 5000000.0) — The base period of the RoPE embeddings.
  • attention_bias (bool, defaults to False, optional, defaults to False) — Whether to use a bias in the query, key, value and output projection layers during self-attention.
  • attention_dropout (float, optional, defaults to 0.0) — The dropout ratio for the attention probabilities.
  • decoder_sparse_step (int, optional, defaults to 1) — The frequency of the MoE layer.
  • moe_intermediate_size (int, optional, defaults to 1408) — Intermediate size of the routed expert.
  • num_experts_per_tok (int, optional, defaults to 4) — Number of selected experts.
  • num_experts (int, optional, defaults to 60) — Number of routed experts.
  • norm_topk_prob (bool, optional, defaults to True) — Whether to normalize the topk probabilities.
  • mlp_only_layers (List[int], optional, defaults to []) — Indicate which layers use Qwen3VLMoeMLP rather than Qwen3VLMoeSparseMoeBlock The list contains layer index, from 0 to num_layers-1 if we have num_layers layers If mlp_only_layers is empty, decoder_sparse_step is used to determine the sparsity.
  • rope_scaling (Dict, optional) — Dictionary containing the scaling configuration for the RoPE embeddings. NOTE: if you apply new rope type and you expect the model to work on longer max_position_embeddings, we recommend you to update this value accordingly. Expected contents: rope_type (str): The sub-variant of RoPE to use. Can be one of [‘default’, ‘linear’, ‘dynamic’, ‘yarn’, ‘longrope’, ‘llama3’], with ‘default’ being the original RoPE implementation. factor (float, optional): Used with all rope types except ‘default’. The scaling factor to apply to the RoPE embeddings. In most scaling types, a factor of x will enable the model to handle sequences of length x original maximum pre-trained length. original_max_position_embeddings (int, optional): Used with ‘dynamic’, ‘longrope’ and ‘llama3’. The original max position embeddings used during pretraining. attention_factor (float, optional): Used with ‘yarn’ and ‘longrope’. The scaling factor to be applied on the attention computation. If unspecified, it defaults to value recommended by the implementation, using the factor field to infer the suggested value. beta_fast (float, optional): Only used with ‘yarn’. Parameter to set the boundary for extrapolation (only) in the linear ramp function. If unspecified, it defaults to 32. beta_slow (float, optional): Only used with ‘yarn’. Parameter to set the boundary for interpolation (only) in the linear ramp function. If unspecified, it defaults to 1. short_factor (List[float], optional): Only used with ‘longrope’. The scaling factor to be applied to short contexts (< original_max_position_embeddings). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2 long_factor (List[float], optional): Only used with ‘longrope’. The scaling factor to be applied to long contexts (< original_max_position_embeddings). Must be a list of numbers with the same length as the hidden size divided by the number of attention heads divided by 2 low_freq_factor (float, optional): Only used with ‘llama3’. Scaling factor applied to low frequency components of the RoPE high_freq_factor (float, optional*): Only used with ‘llama3’. Scaling factor applied to high frequency components of the RoPE
  • head_dim (int, optional) — The dimension of the head. If not specified, will default to hidden_size // num_attention_heads.

This is the configuration class to store the configuration of a Qwen3VLMoeTextModel. It is used to instantiate a Qwen3-VL-MOE model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of Qwen3-VL-30B-A3B-Instruct Qwen/Qwen3-VL-30B-A3B-Instruct.

Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.

>>> from transformers import Qwen3VLMoeForConditionalGeneration, Qwen3VLMoeConfig

>>> # Initializing a Qwen3VLMoe style configuration
>>> configuration = Qwen3VLMoeConfig()

>>> # Initializing a model from the Qwen3-VL-30B-A3B style configuration
>>> model = Qwen3VLMoeForConditionalGeneration(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config

Qwen3VLMoeVisionModel

class transformers.Qwen3VLMoeVisionModel

< >

( config *inputs **kwargs )

forward

< >

( hidden_states: Tensor grid_thw: Tensor **kwargs ) → torch.Tensor

Parameters

  • hidden_states (torch.Tensor of shape (seq_len, hidden_size)) — The final hidden states of the model.
  • grid_thw (torch.Tensor of shape (num_images_or_videos, 3)) — The temporal, height and width of feature shape of each image in LLM.

Returns

torch.Tensor

hidden_states.

Qwen3VLMoeTextModel

class transformers.Qwen3VLMoeTextModel

< >

( config: Qwen3VLMoeTextConfig )

Parameters

  • config (Qwen3VLMoeTextConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

Text part of Qwen3VLMoe, not a pure text-only model, as DeepStack integrates visual features into the early hidden states.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: typing.Optional[torch.LongTensor] = None attention_mask: typing.Optional[torch.Tensor] = None position_ids: typing.Optional[torch.LongTensor] = None past_key_values: typing.Optional[transformers.cache_utils.Cache] = None inputs_embeds: typing.Optional[torch.FloatTensor] = None use_cache: typing.Optional[bool] = None cache_position: typing.Optional[torch.LongTensor] = None visual_pos_masks: typing.Optional[torch.Tensor] = None deepstack_visual_embeds: typing.Optional[list[torch.Tensor]] = None **kwargs: typing_extensions.Unpack[transformers.modeling_flash_attention_utils.FlashAttentionKwargs] ) → transformers.modeling_outputs.BaseModelOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • use_cache (bool, optional) — If set to True, past_key_values key value states are returned and can be used to speed up decoding (see past_key_values).
  • cache_position (torch.LongTensor of shape (sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily to position_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length.
  • visual_pos_masks (torch.Tensor of shape (batch_size, seqlen), optional) — The mask of the visual positions.
  • deepstack_visual_embeds (list[torch.Tensor], optional) — The deepstack visual embeddings. The shape is (num_layers, visual_seqlen, embed_dim). The feature is extracted from the different visual encoder layers, and fed to the decoder hidden states. It’s from the paper DeepStack(https://arxiv.org/abs/2406.04334).

Returns

transformers.modeling_outputs.BaseModelOutputWithPast or tuple(torch.FloatTensor)

A transformers.modeling_outputs.BaseModelOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Qwen3VLMoeConfig) and inputs.

  • last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size)) — Sequence of hidden-states at the output of the last layer of the model.

    If past_key_values is used only the last hidden-state of the sequences of shape (batch_size, 1, hidden_size) is output.

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if config.is_encoder_decoder=True in the cross-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple(torch.FloatTensor), optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple(torch.FloatTensor), optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

The Qwen3VLMoeTextModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Qwen3VLMoeModel

class transformers.Qwen3VLMoeModel

< >

( config )

Parameters

  • config (Qwen3VLMoeModel) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.

The bare Qwen3 Vl Moe Model outputting raw hidden-states without any specific head on top.

This model inherits from PreTrainedModel. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.)

This model is also a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.

forward

< >

( input_ids: LongTensor = None attention_mask: typing.Optional[torch.Tensor] = None position_ids: typing.Optional[torch.LongTensor] = None past_key_values: typing.Optional[transformers.cache_utils.Cache] = None inputs_embeds: typing.Optional[torch.FloatTensor] = None pixel_values: typing.Optional[torch.Tensor] = None pixel_values_videos: typing.Optional[torch.FloatTensor] = None image_grid_thw: typing.Optional[torch.LongTensor] = None video_grid_thw: typing.Optional[torch.LongTensor] = None cache_position: typing.Optional[torch.LongTensor] = None **kwargs: typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs] ) → transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeModelOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • pixel_values (torch.Tensor of shape (batch_size, num_channels, image_size, image_size), optional) — The tensors corresponding to the input images. Pixel values can be obtained using image_processor_class. See image_processor_class.__call__ for details (processor_class uses image_processor_class for processing images).
  • pixel_values_videos (torch.FloatTensor of shape (batch_size, num_frames, num_channels, frame_size, frame_size), optional) — The tensors corresponding to the input video. Pixel values for videos can be obtained using video_processor_class. See video_processor_class.__call__ for details (processor_class uses video_processor_class for processing videos).
  • image_grid_thw (torch.LongTensor of shape (num_images, 3), optional) — The temporal, height and width of feature shape of each image in LLM.
  • video_grid_thw (torch.LongTensor of shape (num_videos, 3), optional) — The temporal, height and width of feature shape of each video in LLM.
  • cache_position (torch.LongTensor of shape (sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily to position_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length.

Returns

transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeModelOutputWithPast or tuple(torch.FloatTensor)

A transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeModelOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (None) and inputs.

  • last_hidden_state (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional, defaults to None) — Sequence of hidden-states at the output of the last layer of the model.

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple[torch.FloatTensor], optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple[torch.FloatTensor], optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

  • rope_deltas (torch.LongTensor of shape (batch_size, ), optional) — The rope index difference between sequence length and multimodal rope.

The Qwen3VLMoeModel forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Qwen3VLMoeForConditionalGeneration

class transformers.Qwen3VLMoeForConditionalGeneration

< >

( config )

forward

< >

( input_ids: LongTensor = None attention_mask: typing.Optional[torch.Tensor] = None position_ids: typing.Optional[torch.LongTensor] = None past_key_values: typing.Optional[transformers.cache_utils.Cache] = None inputs_embeds: typing.Optional[torch.FloatTensor] = None labels: typing.Optional[torch.LongTensor] = None pixel_values: typing.Optional[torch.Tensor] = None pixel_values_videos: typing.Optional[torch.FloatTensor] = None image_grid_thw: typing.Optional[torch.LongTensor] = None video_grid_thw: typing.Optional[torch.LongTensor] = None cache_position: typing.Optional[torch.LongTensor] = None logits_to_keep: typing.Union[int, torch.Tensor] = 0 **kwargs: typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs] ) → transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeCausalLMOutputWithPast or tuple(torch.FloatTensor)

Parameters

  • input_ids (torch.LongTensor of shape (batch_size, sequence_length)) — Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

    Indices can be obtained using AutoTokenizer. See PreTrainedTokenizer.encode() and PreTrainedTokenizer.call() for details.

    What are input IDs?

  • attention_mask (torch.Tensor of shape (batch_size, sequence_length), optional) — Mask to avoid performing attention on padding token indices. Mask values selected in [0, 1]:

    • 1 for tokens that are not masked,
    • 0 for tokens that are masked.

    What are attention masks?

  • position_ids (torch.LongTensor of shape (batch_size, sequence_length), optional) — Indices of positions of each input sequence tokens in the position embeddings. Selected in the range [0, config.n_positions - 1].

    What are position IDs?

  • past_key_values (~cache_utils.Cache, optional) — Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the past_key_values returned by the model at a previous stage of decoding, when use_cache=True or config.use_cache=True.

    Only Cache instance is allowed as input, see our kv cache guide. If no past_key_values are passed, DynamicCache will be initialized by default.

    The model will output the same cache format that is fed as input.

    If past_key_values are used, the user is expected to input only unprocessed input_ids (those that don’t have their past key value states given to this model) of shape (batch_size, unprocessed_length) instead of all input_ids of shape (batch_size, sequence_length).

  • inputs_embeds (torch.FloatTensor of shape (batch_size, sequence_length, hidden_size), optional) — Optionally, instead of passing input_ids you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert input_ids indices into associated vectors than the model’s internal embedding lookup matrix.
  • labels (torch.LongTensor of shape (batch_size, sequence_length), optional) — Labels for computing the masked language modeling loss. Indices should either be in [0, ..., config.vocab_size] or -100 (see input_ids docstring). Tokens with indices set to -100 are ignored (masked), the loss is only computed for the tokens with labels in [0, ..., config.vocab_size].
  • pixel_values (torch.Tensor of shape (batch_size, num_channels, image_size, image_size), optional) — The tensors corresponding to the input images. Pixel values can be obtained using image_processor_class. See image_processor_class.__call__ for details (Qwen3VLProcessor uses image_processor_class for processing images).
  • pixel_values_videos (torch.FloatTensor of shape (batch_size, num_frames, num_channels, frame_size, frame_size), optional) — The tensors corresponding to the input video. Pixel values for videos can be obtained using Qwen3VLVideoProcessor. See Qwen3VLVideoProcessor.__call__() for details (Qwen3VLProcessor uses Qwen3VLVideoProcessor for processing videos).
  • image_grid_thw (torch.LongTensor of shape (num_images, 3), optional) — The temporal, height and width of feature shape of each image in LLM.
  • video_grid_thw (torch.LongTensor of shape (num_videos, 3), optional) — The temporal, height and width of feature shape of each video in LLM.
  • cache_position (torch.LongTensor of shape (sequence_length), optional) — Indices depicting the position of the input sequence tokens in the sequence. Contrarily to position_ids, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length.
  • logits_to_keep (Union[int, torch.Tensor], defaults to 0) — If an int, compute logits for the last logits_to_keep tokens. If 0, calculate logits for all input_ids (special case). Only last token logits are needed for generation, and calculating them only for that token can save memory, which becomes pretty significant for long sequences or large vocabulary size. If a torch.Tensor, must be 1D corresponding to the indices to keep in the sequence length dimension. This is useful when using packed tensor format (single dimension for batch and sequence length).

Returns

transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeCausalLMOutputWithPast or tuple(torch.FloatTensor)

A transformers.models.qwen3_vl_moe.modeling_qwen3_vl_moe.Qwen3VLMoeCausalLMOutputWithPast or a tuple of torch.FloatTensor (if return_dict=False is passed or when config.return_dict=False) comprising various elements depending on the configuration (Qwen3VLMoeConfig) and inputs.

  • loss (torch.FloatTensor of shape (1,), optional, returned when labels is provided) — Language modeling loss (for next-token prediction).

  • logits (torch.FloatTensor of shape (batch_size, sequence_length, config.vocab_size)) — Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).

  • past_key_values (Cache, optional, returned when use_cache=True is passed or when config.use_cache=True) — It is a Cache instance. For more details, see our kv cache guide.

    Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see past_key_values input) to speed up sequential decoding.

  • hidden_states (tuple[torch.FloatTensor], optional, returned when output_hidden_states=True is passed or when config.output_hidden_states=True) — Tuple of torch.FloatTensor (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape (batch_size, sequence_length, hidden_size).

    Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.

  • attentions (tuple[torch.FloatTensor], optional, returned when output_attentions=True is passed or when config.output_attentions=True) — Tuple of torch.FloatTensor (one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length).

    Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads.

  • rope_deltas (torch.LongTensor of shape (batch_size, ), optional) — The rope index difference between sequence length and multimodal rope.

The Qwen3VLMoeForConditionalGeneration forward method, overrides the __call__ special method.

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.

Example: TODO: Add example

< > Update on GitHub

←Qwen3VL SAM2→