Hugging Face's logo

Transformers documentation

Ling 3.0

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v5.19.0).
Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started

This model was contributed to Hugging Face Transformers on 2026-10-13.

Ling 3.0

Overview

Ling 3.0 Flash is a hybrid Mixture-of-Experts language model from InclusionAI. Its decoder groups five Kimi Delta Attention (KDA) layers with one Multi-head Latent Attention (MLA) layer. The MLA blocks use interleaved rotary embeddings and a head-wise output gate, while the feed-forward blocks combine routed and shared experts.

The native Transformers architecture is exposed as BailingHybridForCausalLM and uses the checkpoint model type bailing_hybrid. Native support means the model can be loaded without trust_remote_code=True.

from transformers import AutoModelForCausalLM, AutoTokenizer


model_id = "inclusionAI/Ling-3.0-Flash"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", dtype="auto")

messages = [{"role": "user", "content": "Introduce the Ling model family."}]
inputs = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_tensors="pt", return_dict=True
).to(model.device)
output = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(output[0][inputs.input_ids.shape[1] :], skip_special_tokens=True))

The KDA layers include a pure PyTorch fallback. If flash-linear-attention and causal-conv1d are installed, their optimized kernels are used automatically on CUDA. Installing kernels and passing use_kernels=True to from_pretrained can enable compatible Hub kernels when available.

BailingHybridConfig

class transformers.BailingHybridConfig

< >

( transformers_version: str | None = Nonearchitectures: list[str] | None = Noneoutput_hidden_states: bool | None = Falsereturn_dict: bool | None = Truedtype: str | torch.dtype | None = Nonechunk_size_feed_forward: int = 0is_encoder_decoder: bool = Falseid2label: dict[int, str] | dict[str, str] | None = Nonelabel2id: dict[str, int] | dict[str, str] | None = Noneproblem_type: Literal['regression', 'single_label_classification', 'multi_label_classification'] | None = Nonevocab_size: int = 157184hidden_size: int = 2560intermediate_size: int = 6144moe_intermediate_size: int = 768num_hidden_layers: int = 42num_attention_heads: int = 32num_key_value_heads: int | None = 32n_shared_experts: int = 1output_router_logits: bool = Falserouted_scaling_factor: float = 2.5kv_lora_rank: int = 512qk_rope_head_dim: int = 64v_head_dim: int = 128qk_nope_head_dim: int = 128n_group: int = 8topk_group: int = 4num_experts_per_tok: int = 8first_k_dense_replace: int = 2norm_topk_prob: bool = Truehidden_act: str = 'silu'max_position_embeddings: int = 262144initializer_range: float = 0.02rms_norm_eps: float = 1e-06use_cache: bool = Truepad_token_id: int | None = 156892bos_token_id: int | None = Noneeos_token_id: int | list[int] | None = 156895pretraining_tp: int | None = 1tie_word_embeddings: bool = Falserope_parameters: transformers.modeling_rope_utils.RopeParameters | dict | None = Noneattention_bias: bool = Falseattention_dropout: float = 0.0num_mtp_layers: int = 1num_local_experts: int = 512rope_theta: float | int = 6000000.0head_dim: int = 128short_conv_kernel_size: int = 4layer_group_size: int = 6kda_safe_gate: bool = Truekda_lower_bound: float = -5.0layer_types: list[str] | None = Nonemlp_layer_types: list[str] | None = Nonemtp_loss_scaling_factor: float | int = 0.0 )

Parameters

  • vocab_size (int, optional, defaults to 157184) — 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 2560) — Dimension of the hidden representations.
  • intermediate_size (int, optional, defaults to 6144) — Dimension of the MLP representations.
  • moe_intermediate_size (int, optional, defaults to 768) — Intermediate size of the routed expert MLPs.
  • num_hidden_layers (int, optional, defaults to 42) — Number of hidden layers in the Transformer decoder.
  • num_attention_heads (int, optional, defaults to 32) — Number of attention heads for each attention layer in the Transformer decoder.
  • num_key_value_heads (int, optional, defaults to 32) — 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. If it is not specified, will default to num_attention_heads.
  • n_shared_experts (int, optional, defaults to 1) — Number of shared experts.
  • output_router_logits (bool, optional, defaults to False) — Whether or not the router logits should be returned by the model. Enabling this will also allow the model to output the auxiliary loss, including load balancing loss and router z-loss.
  • routed_scaling_factor (float, optional, defaults to 2.5) — Scaling factor or routed experts.
  • kv_lora_rank (int, optional, defaults to 512) — Rank of the LoRA matrices for key and value projections.
  • qk_rope_head_dim (int, optional, defaults to 64) — Dimension of the query/key heads that use rotary position embeddings.
  • v_head_dim (int, optional, defaults to 128) — Dimension of the value heads.
  • qk_nope_head_dim (int, optional, defaults to 128) — Dimension of the query/key heads that don’t use rotary position embeddings.
  • n_group (int, optional, defaults to 8) — Number of expert groups used by the group-limited router.
  • topk_group (int, optional, defaults to 4) — Number of selected groups for each token (for each token, ensuring the selected experts is only within topk_group groups).
  • num_experts_per_tok (int, optional, defaults to 8) — Number of experts to route each token to. This is the top-k value for the token-choice routing.
  • first_k_dense_replace (int, optional, defaults to 2) — Number of initial decoder layers that use a dense MLP instead of routed experts.
  • norm_topk_prob (bool, optional, defaults to True) — Whether to normalize the weights of the routed experts.
  • 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 262144) — 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.
  • pad_token_id (int, optional, defaults to 156892) — 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, defaults to 156895) — Token id used for end-of-stream in the vocabulary.
  • pretraining_tp (int, optional, defaults to 1) — Experimental feature. Tensor parallelism rank used during pretraining. Please refer to this document to understand more about it. This value is necessary to ensure exact reproducibility of the pretraining results. Please refer to this issue.
  • tie_word_embeddings (bool, optional, defaults to False) — 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 (float, optional, defaults to 0.0) — The dropout ratio for the attention probabilities.
  • num_mtp_layers (int, optional, defaults to 1) — Number of Multi-Token Prediction (MTP) modules available to append after the base transformer model. When 0, the model behaves as a standard decoder. When >0, each extra module can predict one additional future token at inference time (speculative decoding via generate(..., use_mtp=True)).
  • num_local_experts (int, optional, defaults to 512) — Number of local experts on each device. num_experts should be divisible by num_local_experts.
  • rope_theta (float, optional, defaults to 6000000.0) — Base period of the rotary position embeddings used by MLA layers.
  • head_dim (int, optional, defaults to 128) — The attention head dimension. If None, it will default to hidden_size // num_attention_heads
  • short_conv_kernel_size (int, optional, defaults to 4) — Kernel size of the short depthwise convolutions in KDA layers.
  • layer_group_size (int, optional, defaults to 6) — Number of layers in each hybrid attention group. The last layer in every group uses full MLA and the other layers use KDA.
  • kda_safe_gate (bool, optional, defaults to True) — Whether to use the bounded KDA forget gate.
  • kda_lower_bound (float, optional, defaults to -5.0) — Lower bound of the KDA forget gate in log space.
  • layer_types (list[str], optional) — A list that explicitly maps each layer index with its layer type. If not provided, it will be automatically generated based on config values.
  • mlp_layer_types (list[str], optional) — Per-layer MLP types. Derived from first_k_dense_replace when omitted.
  • mtp_loss_scaling_factor (float, optional, defaults to 0.0) — Scaling factor used for the auxiliary multi-token-prediction loss during pretraining.

This is the configuration class to store the configuration of a BailingHybridModel. It is used to instantiate a Bailing Hybrid 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 inclusionAI/Ling-3.0-Flash

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

BailingHybridModel

class transformers.BailingHybridModel

< >

( config: BailingHybridConfig )

Parameters

  • config (BailingHybridConfig) — 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 Bailing Hybrid 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: typing.Optional[torch.LongTensor] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Noneuse_cache: bool | None = None**kwargs: *<class 'transformers.utils.generic.TransformersKwargs'> ) → MoeModelOutputWithPast 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).

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 (BailingHybridConfig) and inputs.

The BailingHybridModel 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 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.

  • 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 logits (post-softmax) that are computed by MoE routers, these terms are used to compute the auxiliary loss for Mixture of Experts models.

BailingHybridForCausalLM

class transformers.BailingHybridForCausalLM

< >

( config )

Parameters

  • config (BailingHybridForCausalLM) — 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 Bailing Hybrid Model for causal language modeling.

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] = Noneattention_mask: typing.Optional[torch.Tensor] = Noneposition_ids: typing.Optional[torch.LongTensor] = Nonepast_key_values: transformers.cache_utils.Cache | None = Noneinputs_embeds: typing.Optional[torch.FloatTensor] = Nonelabels: typing.Optional[torch.LongTensor] = Noneuse_cache: bool | None = Nonelogits_to_keep: typing.Union[int, torch.Tensor] = 0**kwargs: *<class 'transformers.utils.generic.TransformersKwargs'> ) → MoeCausalLMOutputWithPast 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.
  • 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].
  • 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).
  • logits_to_keep (Union[int, torch.Tensor], optional, 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

MoeCausalLMOutputWithPast or tuple(torch.FloatTensor)

A MoeCausalLMOutputWithPast 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 (BailingHybridConfig) and inputs.

The BailingHybridForCausalLM 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.

  • 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).

  • aux_loss (torch.FloatTensor, optional, returned when output_router_logits=True is passed or when config.output_router_logits=True, and the model trains its router with a load-balancing loss) — Load-balancing auxiliary loss for the sparse modules. Models that balance their experts with a router bias instead (DeepSeek-V3 and the architectures derived from it) return None here while still returning router_logits.

  • router_logits (tuple(torch.FloatTensor), optional, returned when output_router_logits=True is passed or when config.output_router_logits=True) — Tuple of torch.FloatTensor (one for each sparse layer) of shape (batch_size * sequence_length, num_experts).

    Raw router logits computed by the MoE routers. They can be used to compute a load-balancing loss or to inspect how tokens are dispatched to experts.

  • 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.

Example:

>>> from transformers import AutoTokenizer, BailingHybridForCausalLM

>>> model = BailingHybridForCausalLM.from_pretrained("meta-bailing_hybrid/BailingHybrid-2-7b-hf")
>>> tokenizer = AutoTokenizer.from_pretrained("meta-bailing_hybrid/BailingHybrid-2-7b-hf")

>>> prompt = "Hey, are you conscious? Can you talk to me?"
>>> inputs = tokenizer(prompt, return_tensors="pt")

>>> # Generate
>>> generate_ids = model.generate(inputs.input_ids, max_length=30)
>>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
"Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
Update on GitHub

←LFM2Moe LLaMA→