# Qwen3-VL-Moe [Qwen3-VL](https://huggingface.co/papers/2502.13923) 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 ```python from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration model = Qwen3VLMoeForConditionalGeneration.from_pretrained( "Qwen/Qwen3-VL-Moe", 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[[transformers.Qwen3VLMoeConfig]] #### transformers.Qwen3VLMoeConfig[[transformers.Qwen3VLMoeConfig]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/configuration_qwen3_vl_moe.py#L151) 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 the [Qwen/Qwen3-VL-30B-A3B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) for more information. Example: ```python >>> 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 ``` **Parameters:** text_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the text backbone. vision_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the vision backbone. image_token_id (`int`, *optional*, defaults to `151655`) : The image token index used as a placeholder for input images. video_token_id (`int`, *optional*, defaults to `151656`) : The video token index used as a placeholder for input videos. vision_start_token_id (`int`, *optional*, defaults to `151652`) : Token ID that marks the start of a visual segment in the multimodal input sequence. vision_end_token_id (`int`, *optional*, defaults to `151653`) : Token ID that marks the end of a visual segment in the multimodal input sequence. tie_word_embeddings (`bool`, *optional*, defaults to `False`) : Whether to tie weight embeddings according to model's `tied_weights_keys` mapping. ## Qwen3VLMoeVisionConfig[[transformers.Qwen3VLMoeVisionConfig]] #### transformers.Qwen3VLMoeVisionConfig[[transformers.Qwen3VLMoeVisionConfig]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/configuration_qwen3_vl_moe.py#L121) 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 the [Qwen/Qwen3-VL-30B-A3B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) for more information. **Parameters:** depth (`int`, *optional*, defaults to `27`) : Number of Transformer layers in the vision encoder. hidden_size (`int`, *optional*, defaults to `1152`) : Dimension of the hidden representations. hidden_act (`str`, *optional*, defaults to `gelu_pytorch_tanh`) : The non-linear activation function (function or string) in the decoder. For example, `"gelu"`, `"relu"`, `"silu"`, etc. intermediate_size (`int`, *optional*, defaults to `4304`) : Dimension of the MLP representations. num_heads (`int`, *optional*, defaults to `16`) : Number of attention heads for each attention layer in the Transformer decoder. in_channels (`int`, *optional*, defaults to `3`) : The number of input channels. patch_size (`Union[int, list[int], tuple[int, int]]`, *optional*, defaults to `16`) : The size (resolution) of each patch. spatial_merge_size (`int`, *optional*, defaults to `2`) : The size of the spatial merge window used to reduce the number of visual tokens by merging neighboring patches. temporal_patch_size (`Union[int, list[int], tuple[int, int]]`, *optional*, defaults to `2`) : Temporal patch size used in the 3D patch embedding for video inputs. out_hidden_size (`int`, *optional*, defaults to 3584) : The output hidden size of the vision model. num_position_embeddings (`int`, *optional*, defaults to 2304) : The maximum sequence length that this model might ever be used with deepstack_visual_indexes (`list[int]`, *optional*, defaults to `[8, 16, 24]`) : Indexed of layers for deepstack embeddings. initializer_range (`float`, *optional*, defaults to `0.02`) : The standard deviation of the truncated_normal_initializer for initializing all weight matrices. ## Qwen3VLMoeTextConfig[[transformers.Qwen3VLMoeTextConfig]] #### transformers.Qwen3VLMoeTextConfig[[transformers.Qwen3VLMoeTextConfig]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/configuration_qwen3_vl_moe.py#L29) 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 the [Qwen/Qwen3-VL-30B-A3B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.8.1/en/main_classes/configuration#transformers.PreTrainedConfig) for more information. ```python >>> 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 ``` **Parameters:** vocab_size (`int`, *optional*, defaults to `151936`) : Vocabulary size of the model. Defines the number of different tokens that can be represented by the `input_ids`. 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 decoder. num_attention_heads (`int`, *optional*, defaults to `16`) : Number of attention heads for each attention layer in the Transformer decoder. 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, check out [this paper](https://huggingface.co/papers/2305.13245). If it is not specified, will default to `num_attention_heads`. hidden_act (`str`, *optional*, defaults to `silu`) : The non-linear activation function (function or string) in the decoder. For example, `"gelu"`, `"relu"`, `"silu"`, etc. 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` or when the model is a decoder-only generative model. tie_word_embeddings (`bool`, *optional*, defaults to `True`) : Whether to tie weight embeddings according to model's `tied_weights_keys` mapping. rope_parameters (`Union[~modeling_rope_utils.RopeParameters, dict]`, *optional*) : Dictionary containing the configuration parameters for the RoPE embeddings. The dictionary should contain a value for `rope_theta` and optionally parameters used for scaling in case you want to use RoPE with longer `max_position_embeddings`. attention_bias (`bool`, *optional*, defaults to `False`) : Whether to use a bias in the query, key, value and output projection layers during self-attention. attention_dropout (`Union[float, int]`, *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 MLPs. num_experts_per_tok (`int`, *optional*, defaults to `4`) : Number of experts to route each token to. This is the top-k value for the token-choice routing. num_experts (`int`, *optional*, defaults to `60`) : Number of routed experts in MoE layers. router_aux_loss_coef (`float`, *optional*, defaults to `0.001`) : Auxiliary load balancing loss coefficient. Used to penalize uneven expert routing in MoE models. 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. pad_token_id (`int`, *optional*) : Token id used for padding in the vocabulary. bos_token_id (`int`, *optional*) : Token id used for beginning-of-stream in the vocabulary. eos_token_id (`Union[int, list[int]]`, *optional*) : Token id used for end-of-stream in the vocabulary. head_dim (`int`, *optional*) : The attention head dimension. If None, it will default to hidden_size // num_attention_heads ## Qwen3VLMoeVisionModel[[transformers.Qwen3VLMoeVisionModel]] #### transformers.Qwen3VLMoeVisionModel[[transformers.Qwen3VLMoeVisionModel]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L599) forwardtransformers.Qwen3VLMoeVisionModel.forwardhttps://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L749[{"name": "hidden_states", "val": ": Tensor"}, {"name": "grid_thw", "val": ": Tensor"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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.0`torch.Tensor`hidden_states. **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[[transformers.Qwen3VLMoeTextModel]] #### transformers.Qwen3VLMoeTextModel[[transformers.Qwen3VLMoeTextModel]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L905) 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](/docs/transformers/v5.8.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Qwen3VLMoeTextModel.forwardhttps://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L926[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "visual_pos_masks", "val": ": torch.Tensor | None = None"}, {"name": "deepstack_visual_embeds", "val": ": list[torch.Tensor] | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.modeling_flash_attention_utils.FlashAttentionKwargs]"}]- **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](/docs/transformers/v5.8.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.1/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.1/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.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`). - **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).0`MoeModelOutputWithPast` or `tuple(torch.FloatTensor)`A `MoeModelOutputWithPast` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs. The [Qwen3VLMoeTextModel](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.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. - **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. - **past_key_values** (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a [Cache](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). 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. - **router_logits** (`tuple(torch.FloatTensor)`, *optional*, returned when `output_router_probs=True` and `config.add_router_probs=True` is passed or when `config.output_router_probs=True`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, sequence_length, num_experts)`. Raw router logtis (post-softmax) that are computed by MoE routers, these terms are used to compute the auxiliary loss for Mixture of Experts models. **Parameters:** config ([Qwen3VLMoeTextConfig](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.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()](/docs/transformers/v5.8.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``MoeModelOutputWithPast` or `tuple(torch.FloatTensor)`` A `MoeModelOutputWithPast` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs. ## Qwen3VLMoeModel[[transformers.Qwen3VLMoeModel]] #### transformers.Qwen3VLMoeModel[[transformers.Qwen3VLMoeModel]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1082) The bare Qwen3 Vl Moe Model outputting raw hidden-states without any specific head on top. This model inherits from [PreTrainedModel](/docs/transformers/v5.8.1/en/main_classes/model#transformers.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](https://pytorch.org/docs/stable/nn.html#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. forwardtransformers.Qwen3VLMoeModel.forwardhttps://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1387[{"name": "input_ids", "val": ": LongTensor = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "pixel_values_videos", "val": ": torch.FloatTensor | None = None"}, {"name": "image_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "video_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "mm_token_type_ids", "val": ": torch.IntTensor | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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](/docs/transformers/v5.8.1/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.1/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.1/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#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?](../glossary#attention-mask) - **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?](../glossary#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](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.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. - **mm_token_type_ids** (`torch.IntTensor` of shape `(batch_size, sequence_length)`, *optional*) -- Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2). Multimodal token type ids can be obtained using [AutoProcessor](/docs/transformers/v5.8.1/en/model_doc/auto#transformers.AutoProcessor). See [ProcessorMixin.__call__()](/docs/transformers/v5.8.1/en/model_doc/align#transformers.AlignProcessor.__call__) for details.0`Qwen3VLMoeModelOutputWithPast` or `tuple(torch.FloatTensor)`A `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. The [Qwen3VLMoeModel](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.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. - **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](/docs/transformers/v5.8.1/en/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). 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. - **router_logits** (`tuple[torch.FloatTensor]`, *optional*, returned when `output_router_logits=True` is passed or when `config.add_router_probs=True`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, sequence_length, num_experts)`. Router logits of the model, useful to compute the auxiliary loss for Mixture of Experts models. **Parameters:** config ([Qwen3VLMoeModel](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.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()](/docs/transformers/v5.8.1/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``Qwen3VLMoeModelOutputWithPast` or `tuple(torch.FloatTensor)`` A `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. #### get_video_features[[transformers.Qwen3VLMoeModel.get_video_features]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1255) - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`, *optional*) -- Last layer hidden-state after a pooling operation on the spatial dimensions. - **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. - **deepstack_features** (`List[torch.FloatTensor]`, *optional*) -- List of hidden-states (feature maps) from deepstack layers. **Parameters:** pixel_values_videos (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input videos. video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*) : The temporal, height and width of feature shape of each video in LLM. **Returns:** ``BaseModelOutputWithDeepstackFeatures` or `tuple(torch.FloatTensor)`` A `BaseModelOutputWithDeepstackFeatures` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs. #### get_image_features[[transformers.Qwen3VLMoeModel.get_image_features]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1272) - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`, *optional*) -- Last layer hidden-state after a pooling operation on the spatial dimensions. - **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. - **deepstack_features** (`List[torch.FloatTensor]`, *optional*) -- List of hidden-states (feature maps) from deepstack layers. **Parameters:** pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input images. image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*) : The temporal, height and width of feature shape of each image in LLM. **Returns:** ``BaseModelOutputWithDeepstackFeatures` or `tuple(torch.FloatTensor)`` A `BaseModelOutputWithDeepstackFeatures` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs. ## Qwen3VLMoeForConditionalGeneration[[transformers.Qwen3VLMoeForConditionalGeneration]] #### transformers.Qwen3VLMoeForConditionalGeneration[[transformers.Qwen3VLMoeForConditionalGeneration]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1576) forwardtransformers.Qwen3VLMoeForConditionalGeneration.forwardhttps://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1627[{"name": "input_ids", "val": ": LongTensor = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "pixel_values_videos", "val": ": torch.FloatTensor | None = None"}, {"name": "image_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "video_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "mm_token_type_ids", "val": ": torch.IntTensor | None = None"}, {"name": "logits_to_keep", "val": ": int | torch.Tensor = 0"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}] 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]`. 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. Example: ```python >>> from PIL import Image >>> import httpx >>> from io import BytesIO >>> from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration >>> model = Qwen3VLMoeForConditionalGeneration.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct", dtype="auto", device_map="auto") >>> processor = AutoProcessor.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct") >>> messages = [ { "role": "user", "content": [ { "type": "image", "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg", }, {"type": "text", "text": "Describe this image in short."}, ], } ] >>> # Preparation for inference >>> inputs = processor.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_dict=True, return_tensors="pt" ) >>> inputs = inputs.to(model.device) >>> # Generate >>> 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) ] >>> processor.batch_decode(generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "A woman in a plaid shirt sits on a sandy beach at sunset, smiling as she gives a high-five to a yellow Labrador Retriever wearing a harness. The ocean waves roll in the background." ``` #### get_video_features[[transformers.Qwen3VLMoeForConditionalGeneration.get_video_features]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1595) - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`, *optional*) -- Last layer hidden-state after a pooling operation on the spatial dimensions. - **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. - **deepstack_features** (`List[torch.FloatTensor]`, *optional*) -- List of hidden-states (feature maps) from deepstack layers. Example: ```python >>> from PIL import Image >>> from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration >>> model = Qwen3VLMoeForConditionalGeneration.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct") >>> processor = AutoProcessor.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct") >>> messages = [ ... { ... "role": "user", "content": [ ... {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"}, ... {"type": "text", "text": "Where is the cat standing?"}, ... ] ... }, ... ] >>> inputs = processor.apply_chat_template( ... messages, ... tokenize=True, ... return_dict=True, ... return_tensors="pt", ... add_generation_prompt=True ... ) >>> # Generate >>> generate_ids = model.generate(**inputs) >>> processor.batch_decode(generate_ids, skip_special_tokens=True)[0] ``` **Parameters:** pixel_values_videos (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input videos. video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*) : The temporal, height and width of feature shape of each video in LLM. **Returns:** ``BaseModelOutputWithDeepstackFeatures` or `tuple(torch.FloatTensor)`` A `BaseModelOutputWithDeepstackFeatures` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs. #### get_image_features[[transformers.Qwen3VLMoeForConditionalGeneration.get_image_features]] [Source](https://github.com/huggingface/transformers/blob/v5.8.1/src/transformers/models/qwen3_vl_moe/modeling_qwen3_vl_moe.py#L1612) - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`, *optional*) -- Last layer hidden-state after a pooling operation on the spatial dimensions. - **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. - **deepstack_features** (`List[torch.FloatTensor]`, *optional*) -- List of hidden-states (feature maps) from deepstack layers. Example: ```python >>> from PIL import Image >>> from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration >>> model = Qwen3VLMoeForConditionalGeneration.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct") >>> processor = AutoProcessor.from_pretrained("Qwen/Qwen3-VL-30B-A3B-Instruct") >>> messages = [ ... { ... "role": "user", "content": [ ... {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"}, ... {"type": "text", "text": "Where is the cat standing?"}, ... ] ... }, ... ] >>> inputs = processor.apply_chat_template( ... messages, ... tokenize=True, ... return_dict=True, ... return_tensors="pt", ... add_generation_prompt=True ... ) >>> # Generate >>> generate_ids = model.generate(**inputs) >>> processor.batch_decode(generate_ids, skip_special_tokens=True)[0] ``` **Parameters:** pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input images. image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*) : The temporal, height and width of feature shape of each image in LLM. **Returns:** ``BaseModelOutputWithDeepstackFeatures` or `tuple(torch.FloatTensor)`` A `BaseModelOutputWithDeepstackFeatures` 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](/docs/transformers/v5.8.1/en/model_doc/qwen3_vl_moe#transformers.Qwen3VLMoeConfig)) and inputs.