# Copyright 2024 The GLM & ZhipuAI team and HuggingFace Inc. team. All rights reserved. # # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from ...configuration_utils import PreTrainedConfig from ...modeling_rope_utils import RopeParameters class GlmConfig(PreTrainedConfig): r""" This is the configuration class to store the configuration of a [`GlmModel`]. It is used to instantiate an Glm 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 Glm-4-9b-chat. e.g. [THUDM/glm-4-9b-chat](https://huggingface.co/THUDM/glm-4-9b-chat) 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 151552): Vocabulary size of the Glm model. Defines the number of different tokens that can be represented by the `inputs_ids` passed when calling [`GlmModel`] hidden_size (`int`, *optional*, defaults to 4096): Dimension of the hidden representations. intermediate_size (`int`, *optional*, defaults to 13696): Dimension of the MLP representations. num_hidden_layers (`int`, *optional*, defaults to 40): 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 2): 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`. head_dim (`int`, *optional*, defaults to 128): The attention head dimension. hidden_act (`str` or `function`, *optional*, defaults to `"silu"`): The legacy activation function. It is overwritten by the `hidden_activation`. attention_dropout (`float`, *optional*, defaults to 0.0): The dropout ratio for the attention probabilities. max_position_embeddings (`int`, *optional*, defaults to 131072): 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 1.5625e-07): 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 to tie weight embeddings rope_parameters (`RopeParameters`, *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`. pad_token_id (`int`, *optional*, defaults to 151329): Padding token id. eos_token_id (`int` | `list`, *optional*, defaults to `[151329, 151336, 151338]`): End of stream token id. bos_token_id (`int`, *optional*): Beginning of stream token id. attention_bias (`bool`, defaults to `False`, *optional*, defaults to `True`): Whether to use a bias in the query, key, value and output projection layers during self-attention. ```python >>> from transformers import GlmModel, GlmConfig >>> # Initializing a Glm glm-4-9b-chat style configuration >>> configuration = GlmConfig() >>> # Initializing a model from the glm-4-9b-chat style configuration >>> model = GlmModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```""" model_type = "glm" keys_to_ignore_at_inference = ["past_key_values"] 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_up_proj": "colwise_gather_output", # we need to replicate here due to the `chunk` operation "layers.*.mlp.down_proj": "rowwise_split_input", # input is replicated due to the `chunk` operation } base_model_pp_plan = { "embed_tokens": (["input_ids"], ["inputs_embeds"]), "layers": (["hidden_states", "attention_mask"], ["hidden_states"]), "norm": (["hidden_states"], ["hidden_states"]), } def __init__( self, vocab_size: int | None = 151552, hidden_size: int | None = 4096, intermediate_size: int | None = 13696, num_hidden_layers: int | None = 40, num_attention_heads: int | None = 32, num_key_value_heads: int | None = 2, head_dim: int | None = 128, hidden_act: str | None = "silu", attention_dropout: float | None = 0.0, max_position_embeddings: int | None = 131072, initializer_range: float | None = 0.02, rms_norm_eps: float | None = 0.00000015625, use_cache: bool | None = True, tie_word_embeddings: bool | None = False, rope_parameters: RopeParameters | dict[str, RopeParameters] | None = None, pad_token_id: int | None = 151329, eos_token_id: list[int] | None = [151329, 151336, 151338], bos_token_id: int | None = None, attention_bias: bool | None = True, **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 self.head_dim = head_dim 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.use_cache = use_cache self.attention_bias = attention_bias self.attention_dropout = attention_dropout self.rope_parameters = rope_parameters kwargs.setdefault("partial_rotary_factor", 0.5) # assign default for BC self.tie_word_embeddings = tie_word_embeddings self.pad_token_id = pad_token_id self.bos_token_id = bos_token_id self.eos_token_id = eos_token_id super().__init__(**kwargs) __all__ = ["GlmConfig"]