import warnings from typing import Literal from transformers import AutoConfig from transformers.models.auto import CONFIG_MAPPING from transformers.configuration_utils import PretrainedConfig from transformers.modeling_rope_utils import rope_config_validation GIGAREMBED_TYPE = "gigarembed" LATENT_ATTENTION_TYPE = "latent_attention" class GigarConfig(PretrainedConfig): r""" This is the configuration class to store the configuration of a [`GigarModel`]. It is used to instantiate an Gigar 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 Gigar-7B. Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the documentation from [`PretrainedConfig`] for more information. Args: vocab_size (`int`, *optional*, defaults to 32000): Vocabulary size of the Gigar model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`GigarModel`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 11008): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 32): 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*): This is the number of key_value heads that should be used to implement Grouped Query Attention. If `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed by meanpooling all the original heads within that group. For more details checkout [this paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `num_attention_heads`. 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 2048): The maximum sequence length that this model might ever be used with. Gigar 1 supports up to 2048 tokens, Gigar 2 up to 4096, CodeLlama up to 16384. initializer_range (`float`, *optional*, defaults to 0.02): The standard deviation of the truncated_normal_initializer for initializing all weight matrices. rms_norm_eps (`float`, *optional*, defaults to 1e-06): The epsilon used by the rms normalization layers. use_cache (`bool`, *optional*, defaults to `True`): Whether or not the model should return the last key/values attentions (not used by all models). Only relevant if `config.is_decoder=True`. pad_token_id (`int`, *optional*): Padding token id. bos_token_id (`int`, *optional*, defaults to 1): Beginning of stream token id. eos_token_id (`int`, *optional*, defaults to 2): End of stream token id. pretraining_tp (`int`, *optional*, defaults to 1): Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this document](https://huggingface.co/docs/transformers/main/perf_train_gpu_many#tensor-parallelism) to understand more about it. This value is necessary to ensure exact reproducibility of the pretraining results. Please refer to [this issue](https://github.com/pytorch/pytorch/issues/76232). tie_word_embeddings (`bool`, *optional*, defaults to `False`): Whether to tie weight embeddings rope_theta (`float`, *optional*, defaults to 10000.0): The base period of the RoPE embeddings. 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', 'gigar3'], 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 'gigar3'. 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 'gigar3'. Scaling factor applied to low frequency components of the RoPE `high_freq_factor` (`float`, *optional*): Only used with 'gigar3'. Scaling factor applied to high frequency components of the RoPE 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. mlp_bias (`bool`, *optional*, defaults to `False`): Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers. head_dim (`int`, *optional*): The attention head dimension. If None, it will default to hidden_size // num_attention_heads ```python >>> from transformers import GigarModel, GigarConfig >>> # Initializing a Gigar gigar-7b style configuration >>> configuration = GigarConfig() >>> # Initializing a model from the gigar-7b style configuration >>> model = GigarModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```""" model_type = "gigar" keys_to_ignore_at_inference = ["past_key_values"] # Default tensor parallel plan for base model `GigarModel` base_model_tp_plan = { "layers.*.self_attn.q_proj": "colwise", "layers.*.self_attn.k_proj": "colwise", "layers.*.self_attn.v_proj": "colwise", "layers.*.self_attn.o_proj": "rowwise", "layers.*.mlp.gate_proj": "colwise", "layers.*.mlp.up_proj": "colwise", "layers.*.mlp.down_proj": "rowwise", } def __init__( self, vocab_size=32000, hidden_size=4096, intermediate_size=11008, num_hidden_layers=32, num_attention_heads=32, num_key_value_heads=None, hidden_act="silu", max_position_embeddings=2048, initializer_range=0.02, rms_norm_eps=1e-6, use_cache=True, pad_token_id=None, bos_token_id=1, eos_token_id=2, pretraining_tp=1, tie_word_embeddings=False, rope_theta=10000.0, rope_scaling=None, attention_bias=False, attention_dropout=0.0, mlp_bias=False, head_dim=None, apply_qk_norm=False, mla_config=None, **kwargs, ): super().__init__( pad_token_id=pad_token_id, bos_token_id=bos_token_id, eos_token_id=eos_token_id, tie_word_embeddings=tie_word_embeddings, **kwargs, ) self.vocab_size = vocab_size self.max_position_embeddings = max_position_embeddings self.hidden_size = hidden_size self.intermediate_size = intermediate_size self.num_hidden_layers = num_hidden_layers self.num_attention_heads = num_attention_heads # for backward compatibility if num_key_value_heads is None: num_key_value_heads = num_attention_heads self.num_key_value_heads = num_key_value_heads self.hidden_act = hidden_act self.initializer_range = initializer_range self.rms_norm_eps = rms_norm_eps self.pretraining_tp = pretraining_tp self.use_cache = use_cache self.rope_theta = rope_theta self.rope_scaling = rope_scaling self.attention_bias = attention_bias self.attention_dropout = attention_dropout self.mlp_bias = mlp_bias self.head_dim = head_dim if head_dim is not None else self.hidden_size // self.num_attention_heads # Validate the correctness of rotary position embeddings parameters # BC: if there is a 'type' field, copy it it to 'rope_type'. if self.rope_scaling is not None and "type" in self.rope_scaling: self.rope_scaling["rope_type"] = self.rope_scaling["type"] rope_config_validation(self) self.apply_qk_norm = apply_qk_norm self.mla_config = mla_config self._validate_mla_config() def _validate_mla_config(self): if self.mla_config is None: warnings.warn("MLA config is None!") return EXPECTED_KEYS = [ "qk_nope_head_dim", "qk_rope_head_dim", "v_head_dim", "kv_lora_rank", "q_lora_rank", ] if not all((key in self.mla_config for key in EXPECTED_KEYS)): raise ValueError( f"MLA config is expected to have the following keys {EXPECTED_KEYS} but got {self.mla_config.keys()}." ) if self.mla_config["qk_nope_head_dim"] + self.mla_config["qk_rope_head_dim"] != self.mla_config["v_head_dim"]: err_msg = ( f"QK and V head dims do not match! Got {self.mla_config['qk_nope_head_dim']} + {self.mla_config['qk_rope_head_dim']} " f"= {self.mla_config['qk_rope_head_dim'] + self.mla_config['qk_nope_head_dim']} and {self.mla_config['v_head_dim']}." ) raise ValueError(err_msg) class GigarEmbedConfig(PretrainedConfig): model_type = "gigarembed" is_composition = False def __init__( self, latent_attention_config=None, text_config=None, padding_side: Literal["right", "left"]="right", add_pad_token: bool=True, is_mask_instruction: bool = True, add_eos: bool=True, mask_type: str="b", **kwargs, ): if isinstance(latent_attention_config, dict): latent_attention_config["model_type"] = ( latent_attention_config["model_type"] if "model_type" in latent_attention_config else LATENT_ATTENTION_TYPE ) latent_attention_config = CONFIG_MAPPING[latent_attention_config["model_type"]](**latent_attention_config) self.latent_attention_config = latent_attention_config if isinstance(text_config, dict): text_config = GigarConfig(**text_config) elif text_config is None: text_config = None self.text_config = text_config self.padding_side = padding_side self.is_mask_instruction = is_mask_instruction self.add_pad_token = add_pad_token self.add_eos = add_eos self.mask_type = mask_type if "hidden_size" in kwargs: self.hidden_size = kwargs["hidden_size"] super().__init__(**kwargs) class LatentAttentionConfig(PretrainedConfig): model_type = LATENT_ATTENTION_TYPE is_composition = False _name_or_path = "latent_attention" def __init__( self, num_latents_value: int, num_cross_heads: int, hidden_dim: int, latent_dim: int, cross_dim_head: int, mult: int, **kwargs, ): self.num_latents_value = num_latents_value self.num_cross_heads = num_cross_heads self.hidden_dim = hidden_dim self.latent_dim = latent_dim self.cross_dim_head = cross_dim_head self.mult = mult super().__init__(**kwargs) AutoConfig.register(GIGAREMBED_TYPE, GigarEmbedConfig) AutoConfig.register(LATENT_ATTENTION_TYPE, LatentAttentionConfig) GigarEmbedConfig.register_for_auto_class() LatentAttentionConfig.register_for_auto_class()