Transformers documentation
Mamba 2
Mamba 2
Mamba 2 is based on the state space duality (SSD) framework which connects structured state space models (SSMs) and attention variants. It uses a more efficient SSD algorithm that is 2-8x faster than Mamba and modifies the architecture to enable tensor parallelism and a grouped-value attention (GVA) head structure.
You can find all the original Mamba 2 checkpoints under the State Space Models organization, but the examples shown below use mistralai/Mamba-Codestral-7B-v0.1 because a Hugging Face implementation isn’t supported yet for the original checkpoints.
Click on the Mamba models in the right sidebar for more examples of how to apply Mamba to different language tasks.
The example below demonstrates how to generate text with Pipeline, AutoModel, and from the command line.
hfoptions id=“usage”>
<hfoption id="Pipeline">import torch
from transformers import pipeline
pipeline = pipeline(
task="text-generation",
model="mistralai/Mamba-Codestral-7B-v0.1",
torch_dtype=torch.bfloat16,
device=0
)
pipeline("Plants create energy through a process known as")
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mamba-Codestral-7B-v0.1")
model = AutoModelForCausalLM.from_pretrained("mistralai/Mamba-Codestral-7B-v0.1", torch_dtype=torch.bfloat16, device_map="auto")
input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to("cuda")
output = model.generate(**input_ids)
print(tokenizer.decode(output[0], skip_special_tokens=True))
echo -e "Plants create energy through a process known as" | transformers-cli run --task text-generation --model mistralai/Mamba-Codestral-7B-v0.1 --device 0
Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the Quantization overview for more available quantization backends.
The example below uses torchao to only quantize the weights to 4-bit integers.
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig
quantization_config = TorchAoConfig("int4_weight_only", group_size=128)
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mamba-Codestral-7B-v0.1")
model = AutoModelForCausalLM.from_pretrained("mistralai/Mamba-Codestral-7B-v0.1", torch_dtype=torch.bfloat16, quantization_config=quantization_config, device_map="auto")
input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to("cuda")
output = model.generate(**input_ids)
print(tokenizer.decode(output[0], skip_special_tokens=True))
Notes
Codestral Mamba has
groups=8
which are similar to the number of kv heads in an attention-based model.Codestral Mamba has two different forward passes,
torch_forward
orcuda_kernels_forward
, and their results are expected to be slightly different.torch_forward
without compilation is 3-4x faster thancuda_kernels_forward
.cuda_kernels_forward
uses the original CUDA kernels if they’re available in your environment. It is slower during prefill because it requires a “warmup run” due to the higher CPU overhead (see these comments for more details).
There are no positional embeddings in this model, but there is an
attention_mask
and a specific logic to mask out hidden states in two places in the case of batched generation (see this comment for more details). This (and the addition of the reimplemented Mamba 2 kernels) results in a slight discrepancy between batched and cached generation.The SSM algorithm heavily relies on tensor contractions, which have matmul equivalents but the order of operations is slightly different. This makes the difference greater at smaller precisions.
Hidden states that correspond to padding tokens is shutdown in 2 places and is mostly tested with left-padding. Right-padding propagates noise down the line and is not guaranteed to yield satisfactory results.
tokenizer.padding_side = "left"
ensures you are using the correct padding side.The example below demonstrates how to fine-tune Mamba 2 with PEFT.
from trl import SFTTrainer
from peft import LoraConfig
from transformers import AutoTokenizer, Mamba2ForCausalLM, TrainingArguments
model_id = 'mistralai/Mamba-Codestral-7B-v0.1'
tokenizer = AutoTokenizer.from_pretrained(model_id, revision='refs/pr/9', from_slow=True, legacy=False)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "left" #enforce padding side left
model = Mamba2ForCausalLM.from_pretrained(model_id, revision='refs/pr/9')
dataset = load_dataset("Abirate/english_quotes", split="train")
# Without CUDA kernels, batch size of 2 occupies one 80GB device
# but precision can be reduced.
# Experiments and trials welcome!
training_args = TrainingArguments(
output_dir="./results",
num_train_epochs=3,
per_device_train_batch_size=2,
logging_dir='./logs',
logging_steps=10,
learning_rate=2e-3
)
lora_config = LoraConfig(
r=8,
target_modules=["embeddings", "in_proj", "out_proj"],
task_type="CAUSAL_LM",
bias="none"
)
trainer = SFTTrainer(
model=model,
tokenizer=tokenizer,
args=training_args,
peft_config=lora_config,
train_dataset=dataset,
dataset_text_field="quote",
)
trainer.train()
Mamba2Config
class transformers.Mamba2Config
< source >( num_heads = 128 head_dim = 64 vocab_size = 32768 hidden_size = 4096 state_size = 128 num_hidden_layers = 64 layer_norm_epsilon = 1e-05 pad_token_id = 1 bos_token_id = 0 eos_token_id = 2 expand = 2 conv_kernel = 4 n_groups = 8 use_bias = False use_conv_bias = True hidden_act = 'silu' initializer_range = 0.1 residual_in_fp32 = True time_step_rank = 'auto' time_step_min = 0.001 time_step_max = 0.1 time_step_floor = 0.0001 time_step_limit = (0.0, inf) rescale_prenorm_residual = False use_cache = True rms_norm = True chunk_size = 256 tie_word_embeddings = False **kwargs )
Parameters
- num_heads (
int
, optional, defaults to 128) — Number of heads for the evolution matrices of mamba 2. - 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 theinputs_ids
passed when calling Mamba2Model. - hidden_size (
int
, optional, defaults to 4096) — 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 64) — Number of hidden layers in the model. - layer_norm_epsilon (
float
, optional, defaults to 1e-05) — The epsilon to use in the layer normalization layers. - pad_token_id (
int
, optional, defaults to 1) — Padding token id. - bos_token_id (
int
, optional, defaults to 0) — 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 8) — Number of groups for the evolution matrices of mamba 2. - use_bias (
bool
, optional, defaults toFalse
) — Whether or not to use bias in [“in_proj”, “out_proj”] of the mixer block - use_conv_bias (
bool
, optional, defaults toTrue
) — 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 0.1) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices. - residual_in_fp32 (
bool
, optional, defaults toTrue
) — Whether or not residuals should be infloat32
. If set toFalse
residuals will keep the samedtype
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 tomath.ceil(self.hidden_size / 16)
- time_step_min (
float
, optional, defaults to 0.001) — Minimumtime_step
used to bounddt_proj.bias
. - time_step_max (
float
, optional, defaults to 0.1) — Maximumtime_step
used to bounddt_proj.bias
. - time_step_floor (
float
, optional, defaults to 0.0001) — Minimum clamping value of thedt_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 toFalse
) — Whether or not to rescaleout_proj
weights when initializing. - use_cache (
bool
, optional, defaults toTrue
) — Whether or not the cache should be used. - rms_norm (
bool
, optional, defaults toTrue
) — 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 toFalse
) — Whether to tie word embeddings or not.
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 architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
Example:
>>> from transformers import Mamba2Config, Mamba2Model
>>> # Initializing a Mamba2 configuration
>>> configuration = Mamba2Config()
>>> # Initializing a model (with random weights) from the configuration
>>> model = Mamba2Model(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
Mamba2Model
class transformers.Mamba2Model
< source >( config )
Parameters
- config (Mamba2Model) — 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 Mamba2 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
< source >( input_ids: typing.Optional[torch.LongTensor] = None inputs_embeds: typing.Optional[torch.LongTensor] = None cache_params: typing.Optional[transformers.models.mamba2.modeling_mamba2.Mamba2Cache] = None use_cache: typing.Optional[bool] = None output_hidden_states: typing.Optional[bool] = None return_dict: typing.Optional[bool] = None cache_position: typing.Optional[torch.LongTensor] = None attention_mask: typing.Optional[torch.Tensor] = None **kwargs ) → transformers.models.mamba2.modeling_mamba2.Mamba2Output
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.
- inputs_embeds (
torch.LongTensor
of shape(batch_size, sequence_length, hidden_size)
, optional) — Optionally, instead of passinginput_ids
you can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_ids
indices into associated vectors than the model’s internal embedding lookup matrix. - cache_params (
~models.mamba2.modeling_mamba2.Mamba2Cache
, optional) — If passed along, the model uses the previous state in all the blocks (which will give the output for theinput_ids
provided as if the model addstate_input_ids + input_ids
as context). - use_cache (
bool
, optional) — If set toTrue
, thecache_params
is returned and can be used to quickly generate the next logits. - output_hidden_states (
bool
, optional) — Whether or not to return the hidden states of all layers. Seehidden_states
under returned tensors for more detail. - return_dict (
bool
, optional) — Whether or not to return a ModelOutput instead of a plain tuple. - cache_position (
torch.LongTensor
of shape(batch_size,)
, optional) — The position of the current input in the cache. This is used to ensure that the cache is correctly updated. Ifcache_params
is passed,cache_position
should also be passed. - 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.
Returns
transformers.models.mamba2.modeling_mamba2.Mamba2Output
or tuple(torch.FloatTensor)
A transformers.models.mamba2.modeling_mamba2.Mamba2Output
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 (Mamba2Config) and inputs.
-
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. -
cache_params (
Mamba2Cache
) — The state of the model at the last time step. Can be used in a forward method with the nextinput_ids
to avoid providing the oldinput_ids
.Includes both the State space model state matrices after the selective scan, and the Convolutional states
-
hidden_states (
tuple(torch.FloatTensor)
, optional, returned whenoutput_hidden_states=True
is passed or whenconfig.output_hidden_states=True
) — Tuple oftorch.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.
The Mamba2Model 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.
Mamba2LMHeadModel
class transformers.Mamba2ForCausalLM
< source >( config )
Parameters
- config (Mamba2ForCausalLM) — 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 MAMBA2 Model transformer with a language modeling head on top (linear layer with weights not tied to the input embeddings).
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
< source >( input_ids: typing.Optional[torch.LongTensor] = None inputs_embeds: typing.Optional[torch.FloatTensor] = None cache_params: typing.Optional[transformers.models.mamba2.modeling_mamba2.Mamba2Cache] = None labels: typing.Optional[torch.LongTensor] = None output_hidden_states: typing.Optional[bool] = None return_dict: typing.Optional[bool] = None use_cache: typing.Optional[bool] = None cache_position: typing.Optional[torch.Tensor] = None attention_mask: typing.Optional[torch.Tensor] = None **kwargs ) → transformers.models.mamba2.modeling_mamba2.Mamba2CausalLMOutput
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.
- inputs_embeds (
torch.FloatTensor
of shape(batch_size, sequence_length, hidden_size)
, optional) — Optionally, instead of passinginput_ids
you can choose to directly pass an embedded representation. This is useful if you want more control over how to convertinput_ids
indices into associated vectors than the model’s internal embedding lookup matrix. - cache_params (
~models.mamba2.modeling_mamba2.Mamba2Cache
, optional) — If passed along, the model uses the previous state in all the blocks (which will give the output for theinput_ids
provided as if the model addstate_input_ids + input_ids
as context). - labels (
torch.LongTensor
of shape(batch_size, sequence_length)
, optional) — Labels for language modeling. Note that the labels are shifted inside the model, i.e. you can setlabels = input_ids
Indices are selected in[-100, 0, ..., config.vocab_size]
All labels set to-100
are ignored (masked), the loss is only computed for labels in[0, ..., config.vocab_size]
- output_hidden_states (
bool
, optional) — Whether or not to return the hidden states of all layers. Seehidden_states
under returned tensors for more detail. - return_dict (
bool
, optional) — Whether or not to return a ModelOutput instead of a plain tuple. - use_cache (
bool
, optional) — If set toTrue
, thecache_params
is returned and can be used to quickly generate the next logits. - cache_position (
torch.Tensor
of shape(batch_size,)
, optional) — The position of the current input in the cache. This is used to ensure that the cache is correctly updated. Ifcache_params
is passed,cache_position
should also be passed. - 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.
Returns
transformers.models.mamba2.modeling_mamba2.Mamba2CausalLMOutput
or tuple(torch.FloatTensor)
A transformers.models.mamba2.modeling_mamba2.Mamba2CausalLMOutput
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 (Mamba2Config) and inputs.
-
loss (
torch.FloatTensor
of shape(1,)
, optional, returned whenlabels
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). -
cache_params (
Mamba2Cache
) — The state of the model at the last time step. Can be used in a forward method with the nextinput_ids
to avoid providing the oldinput_ids
.Includes both the State space model state matrices after the selective scan, and the Convolutional states
-
hidden_states (
tuple(torch.FloatTensor)
, optional, returned whenoutput_hidden_states=True
is passed or whenconfig.output_hidden_states=True
) — Tuple oftorch.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.
The Mamba2ForCausalLM 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.