| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| """MAMBA2 configuration""" |
|
|
| import math |
|
|
| from transformers.configuration_utils import PretrainedConfig |
|
|
|
|
| class Mamba2Config(PretrainedConfig): |
| """ |
| This is the configuration class to store the configuration of a [`Mamba2Model`]. It is used to instantiate a MAMBA2 |
| 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 MAMBA2 |
| [state-spaces/mamba2-2.8b](https://huggingface.co/state-spaces/mamba2-2.8b) architecture. |
| |
| Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the |
| documentation from [`PretrainedConfig`] for more information. |
| |
| |
| Args: |
| head_dim (`int`, *optional*, defaults to 64): |
| Dimension of each head. |
| vocab_size (`int`, *optional*, defaults to 32768): |
| Vocabulary size of the MAMBA2 model. Defines the number of different tokens that can be represented by the |
| `inputs_ids` passed when calling [`Mamba2Model`]. |
| hidden_size (`int`, *optional*, defaults to 2048): |
| Dimensionality of the embeddings and hidden states. |
| state_size (`int`, *optional*, defaults to 128): shape of the state space latents. |
| num_hidden_layers (`int`, *optional*, defaults to 48): |
| Number of hidden layers in the model. |
| norm_eps (`float`, *optional*, defaults to 1e-05): |
| The epsilon to use in the layer normalization layers. |
| pad_token_id (`int`, *optional*, defaults to 0): |
| Padding token id. |
| bos_token_id (`int`, *optional*, defaults to 1): |
| The id of the beginning of sentence token in the vocabulary. |
| eos_token_id (`int`, *optional*, defaults to 2): |
| The id of the end of sentence token in the vocabulary. |
| expand (`int`, *optional*, defaults to 2): Expanding factor used to determine the intermediate size. |
| conv_kernel (`int`, *optional*, defaults to 4): Size of the convolution kernel. |
| n_groups (`int`, *optional*, defaults to 1): |
| Number of groups for the evolution matrices of mamba 2. |
| use_bias (`bool`, *optional*, defaults to `False`): |
| Whether or not to use bias in ["in_proj", "out_proj"] of the mixer block |
| use_conv_bias (`bool`, *optional*, defaults to `True`): |
| Whether or not to use bias in the convolution layer of the mixer block. |
| hidden_act (`str`, *optional*, defaults to `"silu"`): |
| The non-linear activation function (function or string) in the decoder. |
| initializer_range (`float`, *optional*, defaults to 0.02): |
| The standard deviation of the truncated_normal_initializer for initializing all weight matrices. |
| residual_in_fp32 (`bool`, *optional*, defaults to `True`): |
| Whether or not residuals should be in `float32`. |
| If set to `False` residuals will keep the same `dtype` as the rest of the model |
| time_step_rank (`Union[int,str]`, *optional*, defaults to `"auto"`): |
| Rank of the discretization projection matrix. |
| `"auto"` means that it will default to `math.ceil(self.hidden_size / 16)` |
| time_step_min (`float`, *optional*, defaults to 0.001): |
| Minimum `time_step` used to bound `dt_proj.bias`. |
| time_step_max (`float`, *optional*, defaults to 0.1): |
| Maximum `time_step` used to bound `dt_proj.bias`. |
| time_step_floor (`float`, *optional*, defaults to 0.0001): |
| Minimum clamping value of the `dt_proj.bias` layer initialization. |
| time_step_limit (`tuple`, *optional*, defaults to `(0.0, inf)`): |
| Accepted range of time step values. |
| rescale_prenorm_residual (`bool`, *optional*, defaults to `True`): |
| Whether or not to rescale `out_proj` weights when initializing. |
| use_cache (`bool`, *optional*, defaults to `True`): |
| Whether or not the cache should be used. |
| rms_norm (`bool`, *optional*, defaults to `True`): |
| Whether to use RMS norm or not. |
| chunk_size (`int`, *optional*, defaults to 256): |
| Size of the chunks that will comprise the sequence. |
| tie_word_embeddings (`bool`, *optional*, defaults to `False`): |
| Whether to tie word embeddings or not. |
| """ |
|
|
| model_type = "mamba2" |
|
|
| def __init__( |
| self, |
| head_dim: int = 64, |
| vocab_size: int = 32000, |
| hidden_size: int = 2048, |
| state_size: int = 128, |
| num_hidden_layers: int = 48, |
| norm_eps: float = 1e-5, |
| pad_token_id: int = 0, |
| bos_token_id: int = 1, |
| eos_token_id: int = 2, |
| expand: int = 2, |
| conv_kernel: int = 4, |
| n_groups: int = 1, |
| use_bias: bool = False, |
| use_conv_bias: bool = True, |
| hidden_act: str = "silu", |
| initializer_range: float = 0.02, |
| residual_in_fp32: bool = True, |
| time_step_rank: str = "auto", |
| time_step_min: float = 0.001, |
| time_step_max: float = 0.1, |
| time_step_floor: float = 1e-4, |
| time_step_limit=(0.0, float("inf")), |
| rescale_prenorm_residual: bool = True, |
| use_cache: bool = True, |
| rms_norm: bool = True, |
| chunk_size: int = 256, |
| fuse_norm: bool = True, |
| fuse_cross_entropy: bool = True, |
| tie_word_embeddings: bool = False, |
| **kwargs, |
| ): |
| self.vocab_size = vocab_size |
| self.hidden_size = hidden_size |
| self.state_size = state_size |
| self.num_hidden_layers = num_hidden_layers |
| self.norm_eps = norm_eps |
| self.conv_kernel = conv_kernel |
| self.expand = expand |
|
|
| self.bos_token_id = bos_token_id |
| self.eos_token_id = eos_token_id |
| self.pad_token_id = pad_token_id |
| self.use_bias = use_bias |
| self.use_conv_bias = use_conv_bias |
| self.hidden_act = hidden_act |
| self.initializer_range = initializer_range |
| self.time_step_rank = ( |
| math.ceil(self.hidden_size / 16) |
| if time_step_rank == "auto" |
| else time_step_rank |
| ) |
| self.time_step_min = time_step_min |
| self.time_step_max = time_step_max |
| self.time_step_floor = time_step_floor |
| self.rescale_prenorm_residual = rescale_prenorm_residual |
| self.residual_in_fp32 = residual_in_fp32 |
| self.use_cache = use_cache |
| self.n_groups = n_groups |
| self.head_dim = head_dim |
| self.num_heads = int(self.expand * self.hidden_size / self.head_dim) |
| self.rms_norm = rms_norm |
| self.state_size = state_size |
| self.chunk_size = chunk_size |
| self.time_step_limit = time_step_limit |
| self.fuse_norm = fuse_norm |
| self.fuse_cross_entropy = fuse_cross_entropy |
| self.tie_word_embeddings = tie_word_embeddings |
|
|
| super().__init__( |
| bos_token_id=bos_token_id, |
| eos_token_id=eos_token_id, |
| pad_token_id=pad_token_id, |
| tie_word_embeddings=tie_word_embeddings, |
| **kwargs, |
| ) |
|
|
