Time-LLM
Time-LLM is a reprogramming framework to repurpose LLMs for general time series forecasting with the backbone language models kept intact. In other words, it transforms a forecasting task into a “language task” that can be tackled by an off-the-shelf LLM.
1. Auxiliary Functions
source
Normalize
Normalize (num_features:int, eps=1e-05, affine=False, subtract_last=False, non_norm=False)
Normalize
source
ReprogrammingLayer
ReprogrammingLayer (d_model, n_heads, d_keys=None, d_llm=None, attention_dropout=0.1)
ReprogrammingLayer
source
FlattenHead
FlattenHead (n_vars, nf, target_window, head_dropout=0)
FlattenHead
source
PatchEmbedding
PatchEmbedding (d_model, patch_len, stride, dropout)
PatchEmbedding
source
TokenEmbedding
TokenEmbedding (c_in, d_model)
TokenEmbedding
source
ReplicationPad1d
ReplicationPad1d (padding)
ReplicationPad1d
2. Model
source
TimeLLM
TimeLLM (h, input_size, patch_len:int=16, stride:int=8, d_ff:int=128, top_k:int=5, d_llm:int=768, d_model:int=32, n_heads:int=8, enc_in:int=7, dec_in:int=7, llm=None, llm_config=None, llm_tokenizer=None, llm_num_hidden_layers=32, llm_output_attention:bool=True, llm_output_hidden_states:bool=True, prompt_prefix:Optional[str]=None, dropout:float=0.1, stat_exog_list=None, hist_exog_list=None, futr_exog_list=None, loss=MAE(), valid_loss=None, learning_rate:float=0.0001, max_steps:int=5, val_check_steps:int=100, batch_size:int=32, valid_batch_size:Optional[int]=None, windows_batch_size:int=1024, inference_windows_batch_size:int=1024, start_padding_enabled:bool=False, step_size:int=1, num_lr_decays:int=0, early_stop_patience_steps:int=-1, scaler_type:str='identity', num_workers_loader:int=0, drop_last_loader:bool=False, random_seed:int=1, optimizer=None, optimizer_kwargs=None, lr_scheduler=None, lr_scheduler_kwargs=None, **trainer_kwargs)
*TimeLLM
Time-LLM is a reprogramming framework to repurpose an off-the-shelf LLM for time series forecasting.
It trains a reprogramming layer that translates the observed series into a language task. This is fed to the LLM and an output projection layer translates the output back to numerical predictions.
Parameters:
h
: int, Forecast horizon.
input_size
: int,
autorregresive inputs size, y=[1,2,3,4] input_size=2 ->
y_[t-2:t]=[1,2].
patch_len
: int=16, length of patch.
stride
: int=8, stride of patch.
d_ff
: int=128, dimension of
fcn.
top_k
: int=5, top tokens to consider.
d_llm
: int=768,
hidden dimension of LLM.
d_model
: int=32, dimension of model.
n_heads
: int=8, number of heads in attention layer.
enc_in
:
int=7, encoder input size.
dec_in
: int=7, decoder input size.
llm
= None, LLM model to use. If not specified, it will use GPT-2 from
https://huggingface.co/openai-community/gpt2”
llm_config
= None,
configuration of LLM. If not specified, it will use the configuration of
GPT-2 from https://huggingface.co/openai-community/gpt2”
llm_tokenizer
= None, tokenizer of LLM. If not specified, it will use
the GPT-2 tokenizer from
https://huggingface.co/openai-community/gpt2”
llm_num_hidden_layers
= 32, hidden layers in LLM
llm_output_attention
: bool = True, whether to output attention in
encoder.
llm_output_hidden_states
: bool = True, whether to output
hidden states.
prompt_prefix
: str=None, prompt to inform the LLM
about the dataset.
dropout
: float=0.1, dropout rate.
stat_exog_list
: str list, static exogenous columns.
hist_exog_list
: str list, historic exogenous columns.
futr_exog_list
: str list, future exogenous columns.
loss
:
PyTorch module, instantiated train loss class from losses
collection.
valid_loss
: PyTorch module=loss
, instantiated valid loss class from
losses
collection.
learning_rate
: float=1e-3, Learning rate between (0, 1).
max_steps
: int=1000, maximum number of training steps.
num_lr_decays
: int=-1, Number of learning rate decays, evenly
distributed across max_steps.
early_stop_patience_steps
: int=-1,
Number of validation iterations before early stopping.
val_check_steps
: int=100, Number of training steps between every
validation loss check.
batch_size
: int=32, number of different
series in each batch.
valid_batch_size
: int=None, number of
different series in each validation and test batch, if None uses
batch_size.
windows_batch_size
: int=1024, number of windows to
sample in each training batch, default uses all.
inference_windows_batch_size
: int=1024, number of windows to sample in
each inference batch.
start_padding_enabled
: bool=False, if True,
the model will pad the time series with zeros at the beginning, by input
size.
step_size
: int=1, step size between each window of temporal
data.
scaler_type
: str=‘identity’, type of scaler for temporal
inputs normalization see temporal
scalers.
random_seed
: int, random_seed for pytorch initializer and numpy
generators.
num_workers_loader
: int=os.cpu_count(), workers to be
used by TimeSeriesDataLoader
.
drop_last_loader
: bool=False, if
True TimeSeriesDataLoader
drops last non-full batch.
alias
: str,
optional, Custom name of the model.
optimizer
: Subclass of
‘torch.optim.Optimizer’, optional, user specified optimizer instead of
the default choice (Adam).
optimizer_kwargs
: dict, optional, list
of parameters used by the user specified optimizer
.
lr_scheduler
: Subclass of ‘torch.optim.lr_scheduler.LRScheduler’,
optional, user specified lr_scheduler instead of the default choice
(StepLR).
lr_scheduler_kwargs
: dict, optional, list of parameters
used by the user specified lr_scheduler
.
**trainer_kwargs
: int,
keyword trainer arguments inherited from PyTorch Lighning’s
trainer.
TimeLLM.fit
TimeLLM.fit (dataset, val_size=0, test_size=0, random_seed=None, distributed_config=None)
*Fit.
The fit
method, optimizes the neural network’s weights using the
initialization parameters (learning_rate
, windows_batch_size
, …) and
the loss
function as defined during the initialization. Within fit
we use a PyTorch Lightning Trainer
that inherits the initialization’s
self.trainer_kwargs
, to customize its inputs, see PL’s trainer
arguments.
The method is designed to be compatible with SKLearn-like classes and in particular to be compatible with the StatsForecast library.
By default the model
is not saving training checkpoints to protect
disk memory, to get them change enable_checkpointing=True
in
__init__
.
Parameters:
dataset
: NeuralForecast’s
TimeSeriesDataset
,
see
documentation.
val_size
: int, validation size for temporal cross-validation.
random_seed
: int=None, random_seed for pytorch initializer and numpy
generators, overwrites model.__init__’s.
test_size
: int, test
size for temporal cross-validation.
*
TimeLLM.predict
TimeLLM.predict (dataset, test_size=None, step_size=1, random_seed=None, **data_module_kwargs)
*Predict.
Neural network prediction with PL’s Trainer
execution of
predict_step
.
Parameters:
dataset
: NeuralForecast’s
TimeSeriesDataset
,
see
documentation.
test_size
: int=None, test size for temporal cross-validation.
step_size
: int=1, Step size between each window.
random_seed
:
int=None, random_seed for pytorch initializer and numpy generators,
overwrites model.__init__’s.
**data_module_kwargs
: PL’s
TimeSeriesDataModule args, see
documentation.*
Usage example
from neuralforecast import NeuralForecast
from neuralforecast.models import TimeLLM
from neuralforecast.utils import AirPassengersPanel, augment_calendar_df
from transformers import GPT2Config, GPT2Model, GPT2Tokenizer
AirPassengersPanel, calendar_cols = augment_calendar_df(df=AirPassengersPanel, freq='M')
Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test
gpt2_config = GPT2Config.from_pretrained('openai-community/gpt2')
gpt2 = GPT2Model.from_pretrained('openai-community/gpt2', config=gpt2_config)
gpt2_tokenizer = GPT2Tokenizer.from_pretrained('openai-community/gpt2')
prompt_prefix = "The dataset contains data on monthly air passengers. There is a yearly seasonality"
timellm = TimeLLM(h=12,
input_size=36,
llm=gpt2,
llm_config=gpt2_config,
llm_tokenizer=gpt2_tokenizer,
prompt_prefix=prompt_prefix,
batch_size=24,
windows_batch_size=24)
nf = NeuralForecast(
models=[timellm],
freq='M'
)
nf.fit(df=Y_train_df, val_size=12)
forecasts = nf.predict(futr_df=Y_test_df)