Skip to content
Snippets Groups Projects
Commit 75b12ecf authored by Maja Jablonska's avatar Maja Jablonska
Browse files

Modify predictors to fit both AllenNLP predictor class and COMBO predictor class

parent 60d5f107
Branches
Tags
1 merge request!46Merge COMBO 3.0 into master
Show whitespace changes
Inline Side-by-side
combo/modules/graph_parser.py
+ 4
− 0
View file @ 75b12ecf
...@@ -5,6 +5,8 @@ Author: Mateusz Klimaszewski ...@@ -5,6 +5,8 @@ Author: Mateusz Klimaszewski
from typing import List, Optional, Union, Tuple, Dict from typing import List, Optional, Union, Tuple, Dict
from overrides import overrides
from combo import data from combo import data
from combo.config import Registry from combo.config import Registry
from combo.nn import base from combo.nn import base
...@@ -28,6 +30,7 @@ class GraphHeadPredictionModel(Predictor): ...@@ -28,6 +30,7 @@ class GraphHeadPredictionModel(Predictor):
self.cycle_loss_n = cycle_loss_n self.cycle_loss_n = cycle_loss_n
self.graph_weighting = graph_weighting self.graph_weighting = graph_weighting
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
labels: Optional[Union[torch.Tensor, List[torch.Tensor]]] = None, labels: Optional[Union[torch.Tensor, List[torch.Tensor]]] = None,
...@@ -126,6 +129,7 @@ class GraphDependencyRelationModel(Predictor): ...@@ -126,6 +129,7 @@ class GraphDependencyRelationModel(Predictor):
self.dependency_projection_layer = dependency_projection_layer self.dependency_projection_layer = dependency_projection_layer
self.relation_prediction_layer = relation_prediction_layer self.relation_prediction_layer = relation_prediction_layer
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
mask: Optional[torch.BoolTensor] = None, mask: Optional[torch.BoolTensor] = None,
......
combo/modules/lemma.py
+ 2
− 0
View file @ 75b12ecf
...@@ -2,6 +2,7 @@ from typing import Optional, Dict, List, Union ...@@ -2,6 +2,7 @@ from typing import Optional, Dict, List, Union
import torch import torch
import torch.nn as nn import torch.nn as nn
from overrides import overrides
from combo import data from combo import data
from combo.config import Registry from combo.config import Registry
...@@ -57,6 +58,7 @@ class LemmatizerModel(Predictor): ...@@ -57,6 +58,7 @@ class LemmatizerModel(Predictor):
self.dilated_cnn_encoder = TimeDistributed(dilated_cnn_encoder) self.dilated_cnn_encoder = TimeDistributed(dilated_cnn_encoder)
self.input_projection_layer = input_projection_layer self.input_projection_layer = input_projection_layer
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
mask: Optional[torch.BoolTensor] = None, mask: Optional[torch.BoolTensor] = None,
......
combo/modules/morpho.py
+ 2
− 0
View file @ 75b12ecf
...@@ -4,6 +4,7 @@ Author: Mateusz Klimaszewski ...@@ -4,6 +4,7 @@ Author: Mateusz Klimaszewski
""" """
from typing import Dict, List, Optional, Union from typing import Dict, List, Optional, Union
import torch import torch
from overrides import overrides
from combo import data from combo import data
from combo.config import Registry from combo.config import Registry
...@@ -46,6 +47,7 @@ class MorphologicalFeatures(Predictor): ...@@ -46,6 +47,7 @@ class MorphologicalFeatures(Predictor):
dropout=dropout) dropout=dropout)
self.slices = slices self.slices = slices
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
mask: Optional[torch.BoolTensor] = None, mask: Optional[torch.BoolTensor] = None,
......
combo/modules/parser.py
+ 3
− 0
View file @ 75b12ecf
...@@ -7,6 +7,7 @@ from typing import Tuple, Dict, Optional, Union, List ...@@ -7,6 +7,7 @@ from typing import Tuple, Dict, Optional, Union, List
import numpy as np import numpy as np
import torch import torch
import torch.nn.functional as F import torch.nn.functional as F
from overrides import overrides
from combo import data from combo import data
from combo.config import Registry from combo.config import Registry
...@@ -28,6 +29,7 @@ class HeadPredictionModel(Predictor): ...@@ -28,6 +29,7 @@ class HeadPredictionModel(Predictor):
self.dependency_projection_layer = dependency_projection_layer self.dependency_projection_layer = dependency_projection_layer
self.cycle_loss_n = cycle_loss_n self.cycle_loss_n = cycle_loss_n
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
mask: Optional[torch.BoolTensor] = None, mask: Optional[torch.BoolTensor] = None,
...@@ -141,6 +143,7 @@ class DependencyRelationModel(Predictor): ...@@ -141,6 +143,7 @@ class DependencyRelationModel(Predictor):
self.dependency_projection_layer = dependency_projection_layer self.dependency_projection_layer = dependency_projection_layer
self.relation_prediction_layer = relation_prediction_layer self.relation_prediction_layer = relation_prediction_layer
@overrides
def forward(self, def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]], x: Union[torch.Tensor, List[torch.Tensor]],
mask: Optional[torch.BoolTensor] = None, mask: Optional[torch.BoolTensor] = None,
......
combo/predict.py
+ 3
− 3
View file @ 75b12ecf
...@@ -13,14 +13,14 @@ from combo.data import tokenizers, Instance, conllu2sentence, tokens2conllu, sen ...@@ -13,14 +13,14 @@ from combo.data import tokenizers, Instance, conllu2sentence, tokens2conllu, sen
from combo.data.dataset_readers.dataset_reader import DatasetReader from combo.data.dataset_readers.dataset_reader import DatasetReader
from combo.data.instance import JsonDict from combo.data.instance import JsonDict
from combo.predictors.predictor import Predictor from combo.predictors import PredictorModel
from combo.utils import download, graph from combo.utils import download, graph
logger = logging.getLogger(__name__) logger = logging.getLogger(__name__)
@Registry.register(Predictor, 'combo') @Registry.register(PredictorModel, 'combo')
class COMBO(Predictor): class COMBO(PredictorModel):
def __init__(self, def __init__(self,
model: models.Model, model: models.Model,
......
combo/predictors/__init__.py
+ 2
− 0
View file @ 75b12ecf
from .predictor import Predictor
from .predictor_model import PredictorModel
combo/predictors/predictor.py
+ 11
− 395
View file @ 75b12ecf
""" """
Adapted from AllenNLP Adapted from COMBO
https://github.com/allenai/allennlp/blob/main/allennlp/predictors/predictor.py https://gitlab.clarin-pl.eu/syntactic-tools/combo/-/blob/master/combo/models/base.py
""" """
from typing import List, Iterator, Dict, Tuple, Any
import logging
import json
import re
from contextlib import contextmanager
import numpy
import torch import torch
from torch.utils.hooks import RemovableHandle from combo.config.from_parameters import FromParameters
from torch import Tensor from typing import Dict, List, Optional, Union
from torch import backends
from combo.common.util import sanitize
from combo.config import FromParameters
from combo.data.batch import Batch
from combo.data.dataset_readers.dataset_reader import DatasetReader
from combo.data.instance import JsonDict, Instance
from combo.modules.model import Model
from combo.nn import util
logger = logging.getLogger(__name__)
class Predictor(FromParameters):
"""
a `Predictor` is a thin wrapper around an AllenNLP model that handles JSON -> JSON predictions
that can be used for serving models through the web API or making predictions in bulk.
"""
def __init__(self, model: Model, dataset_reader: DatasetReader, frozen: bool = True) -> None:
if frozen:
model.eval()
self._model = model
self.dataset_reader = dataset_reader
self.cuda_device = next(self._model.named_parameters())[1].get_device()
self._token_offsets: List[Tensor] = []
def load_line(self, line: str) -> JsonDict:
"""
If your inputs are not in JSON-lines format (e.g. you have a CSV)
you can override this function to parse them correctly.
"""
return json.loads(line)
def dump_line(self, outputs: Any) -> Any:
"""
If you don't want your outputs in JSON-lines format
you can override this function to output them differently.
"""
return json.dumps(outputs) + "\n"
def predict_json(self, inputs: JsonDict) -> Any:
instance = self._json_to_instance(inputs)
return self.predict_instance(instance)
def json_to_labeled_instances(self, inputs: JsonDict) -> List[Instance]:
"""
Converts incoming json to a [`Instance`](../data/instance.md),
runs the model on the newly created instance, and adds labels to the
`Instance`s given by the model's output.
# Returns
`List[instance]`
A list of `Instance`'s.
"""
instance = self._json_to_instance(inputs)
outputs = self._model.forward_on_instance(instance)
new_instances = self.predictions_to_labeled_instances(instance, outputs)
return new_instances
def get_gradients(self, instances: List[Instance]) -> Tuple[Dict[str, Any], Dict[str, Any]]:
"""
Gets the gradients of the loss with respect to the model inputs.
# Parameters
instances : `List[Instance]`
# Returns
`Tuple[Dict[str, Any], Dict[str, Any]]`
The first item is a Dict of gradient entries for each input.
The keys have the form `{grad_input_1: ..., grad_input_2: ... }`
up to the number of inputs given. The second item is the model's output.
# Notes
Takes a `JsonDict` representing the inputs of the model and converts
them to [`Instances`](../data/instance.md)), sends these through
the model [`forward`](../models/model.md#forward) function after registering hooks on the embedding
layer of the model. Calls `backward` on the loss and then removes the
hooks.
"""
# set requires_grad to true for all parameters, but save original values to
# restore them later
original_param_name_to_requires_grad_dict = {}
for param_name, param in self._model.named_parameters():
original_param_name_to_requires_grad_dict[param_name] = param.requires_grad
param.requires_grad = True
embedding_gradients: List[Tensor] = [] class Predictor(torch.nn.Module, FromParameters):
hooks: List[RemovableHandle] = self._register_embedding_gradient_hooks(embedding_gradients) def forward(self,
x: Union[torch.Tensor, List[torch.Tensor]],
dataset = Batch(instances) mask: Optional[torch.BoolTensor] = None,
dataset.index_instances(self._model.vocab) labels: Optional[Union[torch.Tensor, List[torch.Tensor]]] = None,
dataset_tensor_dict = util.move_to_device(dataset.as_tensor_dict(), self.cuda_device) sample_weights: Optional[Union[torch.Tensor, List[torch.Tensor]]] = None) -> Dict[str, torch.Tensor]:
# To bypass "RuntimeError: cudnn RNN backward can only be called in training mode" raise NotImplementedError()
with backends.cudnn.flags(enabled=False):
outputs = self._model.make_output_human_readable(
self._model.forward(**dataset_tensor_dict) # type: ignore
)
loss = outputs["loss"]
# Zero gradients.
# NOTE: this is actually more efficient than calling `self._model.zero_grad()`
# because it avoids a read op when the gradients are first updated below.
for p in self._model.parameters():
p.grad = None
loss.backward()
for hook in hooks:
hook.remove()
grad_dict = dict()
for idx, grad in enumerate(embedding_gradients):
key = "grad_input_" + str(idx + 1)
grad_dict[key] = grad.detach().cpu().numpy()
# restore the original requires_grad values of the parameters
for param_name, param in self._model.named_parameters():
param.requires_grad = original_param_name_to_requires_grad_dict[param_name]
return grad_dict, outputs
def get_interpretable_layer(self) -> torch.nn.Module:
"""
Returns the input/embedding layer of the model.
If the predictor wraps around a non-AllenNLP model,
this function should be overridden to specify the correct input/embedding layer.
For the cases where the input layer _is_ an embedding layer, this should be the
layer 0 of the embedder.
"""
try:
return util.find_embedding_layer(self._model)
except RuntimeError:
raise RuntimeError(
"If the model does not use `TextFieldEmbedder`, please override "
"`get_interpretable_layer` in your predictor to specify the embedding layer."
)
def get_interpretable_text_field_embedder(self) -> torch.nn.Module:
"""
Returns the first `TextFieldEmbedder` of the model.
If the predictor wraps around a non-AllenNLP model,
this function should be overridden to specify the correct embedder.
"""
try:
return util.find_text_field_embedder(self._model)
except RuntimeError:
raise RuntimeError(
"If the model does not use `TextFieldEmbedder`, please override "
"`get_interpretable_text_field_embedder` in your predictor to specify "
"the embedding layer."
)
def _register_embedding_gradient_hooks(self, embedding_gradients):
"""
Registers a backward hook on the embedding layer of the model. Used to save the gradients
of the embeddings for use in get_gradients()
When there are multiple inputs (e.g., a passage and question), the hook
will be called multiple times. We append all the embeddings gradients
to a list.
We additionally add a hook on the _forward_ pass of the model's `TextFieldEmbedder` to save
token offsets, if there are any. Having token offsets means that you're using a mismatched
token indexer, so we need to aggregate the gradients across wordpieces in a token. We do
that with a simple sum.
"""
def hook_layers(module, grad_in, grad_out):
grads = grad_out[0]
if self._token_offsets:
# If you have a mismatched indexer with multiple TextFields, it's quite possible
# that the order we deal with the gradients is wrong. We'll just take items from
# the list one at a time, and try to aggregate the gradients. If we got the order
# wrong, we should crash, so you'll know about it. If you get an error because of
# that, open an issue on github, and we'll see what we can do. The intersection of
# multiple TextFields and mismatched indexers is pretty small (currently empty, that
# I know of), so we'll ignore this corner case until it's needed.
offsets = self._token_offsets.pop(0)
span_grads, span_mask = util.batched_span_select(grads.contiguous(), offsets)
span_mask = span_mask.unsqueeze(-1)
span_grads *= span_mask # zero out paddings
span_grads_sum = span_grads.sum(2)
span_grads_len = span_mask.sum(2)
# Shape: (batch_size, num_orig_tokens, embedding_size)
grads = span_grads_sum / torch.clamp_min(span_grads_len, 1)
# All the places where the span length is zero, write in zeros.
grads[(span_grads_len == 0).expand(grads.shape)] = 0
embedding_gradients.append(grads)
def get_token_offsets(module, inputs, outputs):
offsets = util.get_token_offsets_from_text_field_inputs(inputs)
if offsets is not None:
self._token_offsets.append(offsets)
hooks = []
text_field_embedder = self.get_interpretable_text_field_embedder()
hooks.append(text_field_embedder.register_forward_hook(get_token_offsets))
embedding_layer = self.get_interpretable_layer()
hooks.append(embedding_layer.register_backward_hook(hook_layers))
return hooks
@contextmanager
def capture_model_internals(self, module_regex: str = ".*") -> Iterator[dict]:
"""
Context manager that captures the internal-module outputs of
this predictor's model. The idea is that you could use it as follows:
```
with predictor.capture_model_internals() as internals:
outputs = predictor.predict_json(inputs)
return {**outputs, "model_internals": internals}
```
"""
results = {}
hooks = []
# First we'll register hooks to add the outputs of each module to the results dict.
def add_output(idx: int):
def _add_output(mod, _, outputs):
results[idx] = {"name": str(mod), "output": sanitize(outputs)}
return _add_output
regex = re.compile(module_regex)
for idx, (name, module) in enumerate(self._model.named_modules()):
if regex.fullmatch(name) and module != self._model:
hook = module.register_forward_hook(add_output(idx))
hooks.append(hook)
# If you capture the return value of the context manager, you get the results dict.
yield results
# And then when you exit the context we remove all the hooks.
for hook in hooks:
hook.remove()
def predict_instance(self, instance: Instance) -> Any:
outputs = self._model.forward_on_instance(instance)
return sanitize(outputs)
def predictions_to_labeled_instances(
self, instance: Instance, outputs: Dict[str, numpy.ndarray]
) -> List[Instance]:
"""
This function takes a model's outputs for an Instance, and it labels that instance according
to the output. For example, in classification this function labels the instance according
to the class with the highest probability. This function is used to to compute gradients
of what the model predicted. The return type is a list because in some tasks there are
multiple predictions in the output (e.g., in NER a model predicts multiple spans). In this
case, each instance in the returned list of Instances contains an individual
entity prediction as the label.
"""
raise RuntimeError("implement this method for model interpretations or attacks")
def _json_to_instance(self, json_dict: JsonDict) -> Instance:
"""
Converts a JSON object into an [`Instance`](../data/instance.md)
and a `JsonDict` of information which the `Predictor` should pass through,
such as tokenised inputs.
"""
raise NotImplementedError
def predict_batch_json(self, inputs: List[JsonDict]) -> List[Any]:
instances = self._batch_json_to_instances(inputs)
return self.predict_batch_instance(instances)
def predict_batch_instance(self, instances: List[Instance]) -> List[Any]:
outputs = self._model.forward_on_instances(instances)
return sanitize(outputs)
def _batch_json_to_instances(self, json_dicts: List[JsonDict]) -> List[Instance]:
"""
Converts a list of JSON objects into a list of `Instance`s.
By default, this expects that a "batch" consists of a list of JSON blobs which would
individually be predicted by `predict_json`. In order to use this method for
batch prediction, `_json_to_instance` should be implemented by the subclass, or
if the instances have some dependency on each other, this method should be overridden
directly.
"""
instances = []
for json_dict in json_dicts:
instances.append(self._json_to_instance(json_dict))
return instances
#
# @classmethod
# def from_path(
# cls,
# archive_path: Union[str, Path],
# predictor_name: str = None,
# cuda_device: int = -1,
# dataset_reader_to_load: str = "validation",
# frozen: bool = True,
# import_plugins: bool = True,
# overrides: Union[str, Dict[str, Any]] = "",
# ) -> "Predictor":
# """
# Instantiate a `Predictor` from an archive path.
#
# If you need more detailed configuration options, such as overrides,
# please use `from_archive`.
#
# # Parameters
#
# archive_path : `Union[str, Path]`
# The path to the archive.
# predictor_name : `str`, optional (default=`None`)
# Name that the predictor is registered as, or None to use the
# predictor associated with the model.
# cuda_device : `int`, optional (default=`-1`)
# If `cuda_device` is >= 0, the model will be loaded onto the
# corresponding GPU. Otherwise it will be loaded onto the CPU.
# dataset_reader_to_load : `str`, optional (default=`"validation"`)
# Which dataset reader to load from the archive, either "train" or
# "validation".
# frozen : `bool`, optional (default=`True`)
# If we should call `model.eval()` when building the predictor.
# import_plugins : `bool`, optional (default=`True`)
# If `True`, we attempt to import plugins before loading the predictor.
# This comes with additional overhead, but means you don't need to explicitly
# import the modules that your predictor depends on as long as those modules
# can be found by `allennlp.common.plugins.import_plugins()`.
# overrides : `Union[str, Dict[str, Any]]`, optional (default = `""`)
# JSON overrides to apply to the unarchived `Params` object.
#
# # Returns
#
# `Predictor`
# A Predictor instance.
# """
# if import_plugins:
# plugins.import_plugins()
# return Predictor.from_archive(
# load_archive(archive_path, cuda_device=cuda_device, overrides=overrides),
# predictor_name,
# dataset_reader_to_load=dataset_reader_to_load,
# frozen=frozen,
# )
#
# @classmethod
# def from_archive(
# cls,
# archive: Archive,
# predictor_name: str = None,
# dataset_reader_to_load: str = "validation",
# frozen: bool = True,
# ) -> "Predictor":
# """
# Instantiate a `Predictor` from an [`Archive`](../models/archival.md);
# that is, from the result of training a model. Optionally specify which `Predictor`
# subclass; otherwise, we try to find a corresponding predictor in `DEFAULT_PREDICTORS`, or if
# one is not found, the base class (i.e. `Predictor`) will be used. Optionally specify
# which [`DatasetReader`](../data/dataset_readers/dataset_reader.md) should be loaded;
# otherwise, the validation one will be used if it exists followed by the training dataset reader.
# Optionally specify if the loaded model should be frozen, meaning `model.eval()` will be called.
# """
# # Duplicate the config so that the config inside the archive doesn't get consumed
# config = archive.config.duplicate()
#
# if not predictor_name:
# model_type = config.get("model").get("type")
# model_class, _ = Model.resolve_class_name(model_type)
# predictor_name = model_class.default_predictor
# predictor_class: Type[Predictor] = (
# Predictor.by_name(predictor_name) if predictor_name is not None else cls # type: ignore
# )
#
# if dataset_reader_to_load == "validation":
# dataset_reader = archive.validation_dataset_reader
# else:
# dataset_reader = archive.dataset_reader
#
# model = archive.model
# if frozen:
# model.eval()
#
# return predictor_class(model, dataset_reader)
combo/predictors/predictor_model.py 0 → 100644
+ 401
− 0
View file @ 75b12ecf
"""
Adapted from AllenNLP
https://github.com/allenai/allennlp/blob/main/allennlp/predictors/predictor.py
"""
from typing import List, Iterator, Dict, Tuple, Any
import logging
import json
import re
from contextlib import contextmanager
import numpy
import torch
from torch.utils.hooks import RemovableHandle
from torch import Tensor
from torch import backends
from combo.common.util import sanitize
from combo.config import FromParameters
from combo.data.batch import Batch
from combo.data.dataset_readers.dataset_reader import DatasetReader
from combo.data.instance import JsonDict, Instance
from combo.modules.model import Model
from combo.nn import util
logger = logging.getLogger(__name__)
class PredictorModel(FromParameters):
"""
a `Predictor` is a thin wrapper around an AllenNLP model that handles JSON -> JSON predictions
that can be used for serving models through the web API or making predictions in bulk.
"""
def __init__(self, model: Model, dataset_reader: DatasetReader, frozen: bool = True) -> None:
if frozen:
model.eval()
self._model = model
self.dataset_reader = dataset_reader
self.cuda_device = next(self._model.named_parameters())[1].get_device()
self._token_offsets: List[Tensor] = []
def load_line(self, line: str) -> JsonDict:
"""
If your inputs are not in JSON-lines format (e.g. you have a CSV)
you can override this function to parse them correctly.
"""
return json.loads(line)
def dump_line(self, outputs: Any) -> Any:
"""
If you don't want your outputs in JSON-lines format
you can override this function to output them differently.
"""
return json.dumps(outputs) + "\n"
def predict_json(self, inputs: JsonDict) -> Any:
instance = self._json_to_instance(inputs)
return self.predict_instance(instance)
def json_to_labeled_instances(self, inputs: JsonDict) -> List[Instance]:
"""
Converts incoming json to a [`Instance`](../data/instance.md),
runs the model on the newly created instance, and adds labels to the
`Instance`s given by the model's output.
# Returns
`List[instance]`
A list of `Instance`'s.
"""
instance = self._json_to_instance(inputs)
outputs = self._model.forward_on_instance(instance)
new_instances = self.predictions_to_labeled_instances(instance, outputs)
return new_instances
def get_gradients(self, instances: List[Instance]) -> Tuple[Dict[str, Any], Dict[str, Any]]:
"""
Gets the gradients of the loss with respect to the model inputs.
# Parameters
instances : `List[Instance]`
# Returns
`Tuple[Dict[str, Any], Dict[str, Any]]`
The first item is a Dict of gradient entries for each input.
The keys have the form `{grad_input_1: ..., grad_input_2: ... }`
up to the number of inputs given. The second item is the model's output.
# Notes
Takes a `JsonDict` representing the inputs of the model and converts
them to [`Instances`](../data/instance.md)), sends these through
the model [`forward`](../models/model.md#forward) function after registering hooks on the embedding
layer of the model. Calls `backward` on the loss and then removes the
hooks.
"""
# set requires_grad to true for all parameters, but save original values to
# restore them later
original_param_name_to_requires_grad_dict = {}
for param_name, param in self._model.named_parameters():
original_param_name_to_requires_grad_dict[param_name] = param.requires_grad
param.requires_grad = True
embedding_gradients: List[Tensor] = []
hooks: List[RemovableHandle] = self._register_embedding_gradient_hooks(embedding_gradients)
dataset = Batch(instances)
dataset.index_instances(self._model.vocab)
dataset_tensor_dict = util.move_to_device(dataset.as_tensor_dict(), self.cuda_device)
# To bypass "RuntimeError: cudnn RNN backward can only be called in training mode"
with backends.cudnn.flags(enabled=False):
outputs = self._model.make_output_human_readable(
self._model.forward(**dataset_tensor_dict) # type: ignore
)
loss = outputs["loss"]
# Zero gradients.
# NOTE: this is actually more efficient than calling `self._model.zero_grad()`
# because it avoids a read op when the gradients are first updated below.
for p in self._model.parameters():
p.grad = None
loss.backward()
for hook in hooks:
hook.remove()
grad_dict = dict()
for idx, grad in enumerate(embedding_gradients):
key = "grad_input_" + str(idx + 1)
grad_dict[key] = grad.detach().cpu().numpy()
# restore the original requires_grad values of the parameters
for param_name, param in self._model.named_parameters():
param.requires_grad = original_param_name_to_requires_grad_dict[param_name]
return grad_dict, outputs
def get_interpretable_layer(self) -> torch.nn.Module:
"""
Returns the input/embedding layer of the model.
If the predictor wraps around a non-AllenNLP model,
this function should be overridden to specify the correct input/embedding layer.
For the cases where the input layer _is_ an embedding layer, this should be the
layer 0 of the embedder.
"""
try:
return util.find_embedding_layer(self._model)
except RuntimeError:
raise RuntimeError(
"If the model does not use `TextFieldEmbedder`, please override "
"`get_interpretable_layer` in your predictor to specify the embedding layer."
)
def get_interpretable_text_field_embedder(self) -> torch.nn.Module:
"""
Returns the first `TextFieldEmbedder` of the model.
If the predictor wraps around a non-AllenNLP model,
this function should be overridden to specify the correct embedder.
"""
try:
return util.find_text_field_embedder(self._model)
except RuntimeError:
raise RuntimeError(
"If the model does not use `TextFieldEmbedder`, please override "
"`get_interpretable_text_field_embedder` in your predictor to specify "
"the embedding layer."
)
def _register_embedding_gradient_hooks(self, embedding_gradients):
"""
Registers a backward hook on the embedding layer of the model. Used to save the gradients
of the embeddings for use in get_gradients()
When there are multiple inputs (e.g., a passage and question), the hook
will be called multiple times. We append all the embeddings gradients
to a list.
We additionally add a hook on the _forward_ pass of the model's `TextFieldEmbedder` to save
token offsets, if there are any. Having token offsets means that you're using a mismatched
token indexer, so we need to aggregate the gradients across wordpieces in a token. We do
that with a simple sum.
"""
def hook_layers(module, grad_in, grad_out):
grads = grad_out[0]
if self._token_offsets:
# If you have a mismatched indexer with multiple TextFields, it's quite possible
# that the order we deal with the gradients is wrong. We'll just take items from
# the list one at a time, and try to aggregate the gradients. If we got the order
# wrong, we should crash, so you'll know about it. If you get an error because of
# that, open an issue on github, and we'll see what we can do. The intersection of
# multiple TextFields and mismatched indexers is pretty small (currently empty, that
# I know of), so we'll ignore this corner case until it's needed.
offsets = self._token_offsets.pop(0)
span_grads, span_mask = util.batched_span_select(grads.contiguous(), offsets)
span_mask = span_mask.unsqueeze(-1)
span_grads *= span_mask # zero out paddings
span_grads_sum = span_grads.sum(2)
span_grads_len = span_mask.sum(2)
# Shape: (batch_size, num_orig_tokens, embedding_size)
grads = span_grads_sum / torch.clamp_min(span_grads_len, 1)
# All the places where the span length is zero, write in zeros.
grads[(span_grads_len == 0).expand(grads.shape)] = 0
embedding_gradients.append(grads)
def get_token_offsets(module, inputs, outputs):
offsets = util.get_token_offsets_from_text_field_inputs(inputs)
if offsets is not None:
self._token_offsets.append(offsets)
hooks = []
text_field_embedder = self.get_interpretable_text_field_embedder()
hooks.append(text_field_embedder.register_forward_hook(get_token_offsets))
embedding_layer = self.get_interpretable_layer()
hooks.append(embedding_layer.register_backward_hook(hook_layers))
return hooks
@contextmanager
def capture_model_internals(self, module_regex: str = ".*") -> Iterator[dict]:
"""
Context manager that captures the internal-module outputs of
this predictor's model. The idea is that you could use it as follows:
```
with predictor.capture_model_internals() as internals:
outputs = predictor.predict_json(inputs)
return {**outputs, "model_internals": internals}
```
"""
results = {}
hooks = []
# First we'll register hooks to add the outputs of each module to the results dict.
def add_output(idx: int):
def _add_output(mod, _, outputs):
results[idx] = {"name": str(mod), "output": sanitize(outputs)}
return _add_output
regex = re.compile(module_regex)
for idx, (name, module) in enumerate(self._model.named_modules()):
if regex.fullmatch(name) and module != self._model:
hook = module.register_forward_hook(add_output(idx))
hooks.append(hook)
# If you capture the return value of the context manager, you get the results dict.
yield results
# And then when you exit the context we remove all the hooks.
for hook in hooks:
hook.remove()
def predict_instance(self, instance: Instance) -> Any:
outputs = self._model.forward_on_instance(instance)
return sanitize(outputs)
def predictions_to_labeled_instances(
self, instance: Instance, outputs: Dict[str, numpy.ndarray]
) -> List[Instance]:
"""
This function takes a model's outputs for an Instance, and it labels that instance according
to the output. For example, in classification this function labels the instance according
to the class with the highest probability. This function is used to to compute gradients
of what the model predicted. The return type is a list because in some tasks there are
multiple predictions in the output (e.g., in NER a model predicts multiple spans). In this
case, each instance in the returned list of Instances contains an individual
entity prediction as the label.
"""
raise RuntimeError("implement this method for model interpretations or attacks")
def _json_to_instance(self, json_dict: JsonDict) -> Instance:
"""
Converts a JSON object into an [`Instance`](../data/instance.md)
and a `JsonDict` of information which the `Predictor` should pass through,
such as tokenised inputs.
"""
raise NotImplementedError
def predict_batch_json(self, inputs: List[JsonDict]) -> List[Any]:
instances = self._batch_json_to_instances(inputs)
return self.predict_batch_instance(instances)
def predict_batch_instance(self, instances: List[Instance]) -> List[Any]:
outputs = self._model.forward_on_instances(instances)
return sanitize(outputs)
def _batch_json_to_instances(self, json_dicts: List[JsonDict]) -> List[Instance]:
"""
Converts a list of JSON objects into a list of `Instance`s.
By default, this expects that a "batch" consists of a list of JSON blobs which would
individually be predicted by `predict_json`. In order to use this method for
batch prediction, `_json_to_instance` should be implemented by the subclass, or
if the instances have some dependency on each other, this method should be overridden
directly.
"""
instances = []
for json_dict in json_dicts:
instances.append(self._json_to_instance(json_dict))
return instances
#
# @classmethod
# def from_path(
# cls,
# archive_path: Union[str, Path],
# predictor_name: str = None,
# cuda_device: int = -1,
# dataset_reader_to_load: str = "validation",
# frozen: bool = True,
# import_plugins: bool = True,
# overrides: Union[str, Dict[str, Any]] = "",
# ) -> "Predictor":
# """
# Instantiate a `Predictor` from an archive path.
#
# If you need more detailed configuration options, such as overrides,
# please use `from_archive`.
#
# # Parameters
#
# archive_path : `Union[str, Path]`
# The path to the archive.
# predictor_name : `str`, optional (default=`None`)
# Name that the predictor is registered as, or None to use the
# predictor associated with the model.
# cuda_device : `int`, optional (default=`-1`)
# If `cuda_device` is >= 0, the model will be loaded onto the
# corresponding GPU. Otherwise it will be loaded onto the CPU.
# dataset_reader_to_load : `str`, optional (default=`"validation"`)
# Which dataset reader to load from the archive, either "train" or
# "validation".
# frozen : `bool`, optional (default=`True`)
# If we should call `model.eval()` when building the predictor.
# import_plugins : `bool`, optional (default=`True`)
# If `True`, we attempt to import plugins before loading the predictor.
# This comes with additional overhead, but means you don't need to explicitly
# import the modules that your predictor depends on as long as those modules
# can be found by `allennlp.common.plugins.import_plugins()`.
# overrides : `Union[str, Dict[str, Any]]`, optional (default = `""`)
# JSON overrides to apply to the unarchived `Params` object.
#
# # Returns
#
# `Predictor`
# A Predictor instance.
# """
# if import_plugins:
# plugins.import_plugins()
# return Predictor.from_archive(
# load_archive(archive_path, cuda_device=cuda_device, overrides=overrides),
# predictor_name,
# dataset_reader_to_load=dataset_reader_to_load,
# frozen=frozen,
# )
#
# @classmethod
# def from_archive(
# cls,
# archive: Archive,
# predictor_name: str = None,
# dataset_reader_to_load: str = "validation",
# frozen: bool = True,
# ) -> "Predictor":
# """
# Instantiate a `Predictor` from an [`Archive`](../models/archival.md);
# that is, from the result of training a model. Optionally specify which `Predictor`
# subclass; otherwise, we try to find a corresponding predictor in `DEFAULT_PREDICTORS`, or if
# one is not found, the base class (i.e. `Predictor`) will be used. Optionally specify
# which [`DatasetReader`](../data/dataset_readers/dataset_reader.md) should be loaded;
# otherwise, the validation one will be used if it exists followed by the training dataset reader.
# Optionally specify if the loaded model should be frozen, meaning `model.eval()` will be called.
# """
# # Duplicate the config so that the config inside the archive doesn't get consumed
# config = archive.config.duplicate()
#
# if not predictor_name:
# model_type = config.get("model").get("type")
# model_class, _ = Model.resolve_class_name(model_type)
# predictor_name = model_class.default_predictor
# predictor_class: Type[Predictor] = (
# Predictor.by_name(predictor_name) if predictor_name is not None else cls # type: ignore
# )
#
# if dataset_reader_to_load == "validation":
# dataset_reader = archive.validation_dataset_reader
# else:
# dataset_reader = archive.dataset_reader
#
# model = archive.model
# if frozen:
# model.eval()
#
# return predictor_class(model, dataset_reader)
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment