Source code for sparknlp.annotator.seq2seq.phi2_transformer

#  Copyright 2017-2022 John Snow Labs
#
#  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.
"""Contains classes for the Phi2Transformer."""

from sparknlp.common import *


[docs]class Phi2Transformer(AnnotatorModel, HasBatchedAnnotate, HasEngine): """Phi-2: Textbooks Are All You Need. Phi-2 is a Transformer with 2.7 billion parameters. It was trained using the same data sources as Phi-1.5, augmented with a new data source that consists of various NLP synthetic texts and filtered websites (for safety and educational value). When assessed against benchmarks testing common sense, language understanding, and logical reasoning, Phi-2 showcased a nearly state-of-the-art performance among models with less than 13 billion parameters. Phi-2 hasn't been fine-tuned through reinforcement learning from human feedback. The intention behind crafting this open-source model is to provide the research community with a non-restricted small model to explore vital safety challenges, such as reducing toxicity, understanding societal biases, enhancing controllability, and more. Pretrained models can be loaded with :meth:`.pretrained` of the companion object: >>> phi2 = Phi2Transformer.pretrained() \\ ... .setInputCols(["document"]) \\ ... .setOutputCol("generation") The default model is ``"llam2-7b"``, if no name is provided. For available pretrained models please see the `Models Hub <https://sparknlp.org/models?q=phi2>`__. ====================== ====================== Input Annotation types Output Annotation type ====================== ====================== ``DOCUMENT`` ``DOCUMENT`` ====================== ====================== Parameters ---------- configProtoBytes ConfigProto from tensorflow, serialized into byte array. minOutputLength Minimum length of the sequence to be generated, by default 0 maxOutputLength Maximum length of output text, by default 20 doSample Whether or not to use sampling; use greedy decoding otherwise, by default False temperature The value used to module the next token probabilities, by default 1.0 topK The number of highest probability vocabulary tokens to keep for top-k-filtering, by default 50 topP Top cumulative probability for vocabulary tokens, by default 1.0 If set to float < 1, only the most probable tokens with probabilities that add up to ``topP`` or higher are kept for generation. repetitionPenalty The parameter for repetition penalty, 1.0 means no penalty. , by default 1.0 noRepeatNgramSize If set to int > 0, all ngrams of that size can only occur once, by default 0 ignoreTokenIds A list of token ids which are ignored in the decoder's output, by default [] Notes ----- This is a very computationally expensive module especially on larger sequence. The use of an accelerator such as GPU is recommended. References ---------- - `Phi-2: Textbooks Are All You Need. <https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/>`__ - https://huggingface.co/microsoft/phi-2 **Paper Abstract:** *In this work, we develop and release Llama 2, a collection of pretrained and fine-tuned large language models (LLMs) ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama 2-Chat, are optimized for dialogue use cases. Our models outperform open-source chat models on most benchmarks we tested, and based on our human evaluations for helpfulness and safety, may be a suitable substitute for closed-source models. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama 2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs.* Examples -------- >>> import sparknlp >>> from sparknlp.base import * >>> from sparknlp.annotator import * >>> from pyspark.ml import Pipeline >>> documentAssembler = DocumentAssembler() \\ ... .setInputCol("text") \\ ... .setOutputCol("documents") >>> phi2 = Phi2Transformer.pretrained("phi2") \\ ... .setInputCols(["documents"]) \\ ... .setMaxOutputLength(50) \\ ... .setOutputCol("generation") >>> pipeline = Pipeline().setStages([documentAssembler, phi2]) >>> data = spark.createDataFrame([["My name is Leonardo."]]).toDF("text") >>> result = pipeline.fit(data).transform(data) >>> result.select("summaries.generation").show(truncate=False) +----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ |result | +----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ |[My name is Leonardo . I am a student of the University of California, Berkeley. I am interested in the field of Artificial Intelligence and its applications in the real world. I have a strong | | passion for learning and am always looking for ways to improve my knowledge and skills] | -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ """ name = "Phi2Transformer" inputAnnotatorTypes = [AnnotatorType.DOCUMENT] outputAnnotatorType = AnnotatorType.DOCUMENT configProtoBytes = Param(Params._dummy(), "configProtoBytes", "ConfigProto from tensorflow, serialized into byte array. Get with config_proto.SerializeToString()", TypeConverters.toListInt) minOutputLength = Param(Params._dummy(), "minOutputLength", "Minimum length of the sequence to be generated", typeConverter=TypeConverters.toInt) maxOutputLength = Param(Params._dummy(), "maxOutputLength", "Maximum length of output text", typeConverter=TypeConverters.toInt) doSample = Param(Params._dummy(), "doSample", "Whether or not to use sampling; use greedy decoding otherwise", typeConverter=TypeConverters.toBoolean) temperature = Param(Params._dummy(), "temperature", "The value used to module the next token probabilities", typeConverter=TypeConverters.toFloat) topK = Param(Params._dummy(), "topK", "The number of highest probability vocabulary tokens to keep for top-k-filtering", typeConverter=TypeConverters.toInt) topP = Param(Params._dummy(), "topP", "If set to float < 1, only the most probable tokens with probabilities that add up to ``top_p`` or higher are kept for generation", typeConverter=TypeConverters.toFloat) repetitionPenalty = Param(Params._dummy(), "repetitionPenalty", "The parameter for repetition penalty. 1.0 means no penalty. See `this paper <https://arxiv.org/pdf/1909.05858.pdf>`__ for more details", typeConverter=TypeConverters.toFloat) noRepeatNgramSize = Param(Params._dummy(), "noRepeatNgramSize", "If set to int > 0, all ngrams of that size can only occur once", typeConverter=TypeConverters.toInt) ignoreTokenIds = Param(Params._dummy(), "ignoreTokenIds", "A list of token ids which are ignored in the decoder's output", typeConverter=TypeConverters.toListInt)
[docs] def setIgnoreTokenIds(self, value): """A list of token ids which are ignored in the decoder's output. Parameters ---------- value : List[int] The words to be filtered out """ return self._set(ignoreTokenIds=value)
[docs] def setConfigProtoBytes(self, b): """Sets configProto from tensorflow, serialized into byte array. Parameters ---------- b : List[int] ConfigProto from tensorflow, serialized into byte array """ return self._set(configProtoBytes=b)
[docs] def setMinOutputLength(self, value): """Sets minimum length of the sequence to be generated. Parameters ---------- value : int Minimum length of the sequence to be generated """ return self._set(minOutputLength=value)
[docs] def setMaxOutputLength(self, value): """Sets maximum length of output text. Parameters ---------- value : int Maximum length of output text """ return self._set(maxOutputLength=value)
[docs] def setDoSample(self, value): """Sets whether or not to use sampling, use greedy decoding otherwise. Parameters ---------- value : bool Whether or not to use sampling; use greedy decoding otherwise """ return self._set(doSample=value)
[docs] def setTemperature(self, value): """Sets the value used to module the next token probabilities. Parameters ---------- value : float The value used to module the next token probabilities """ return self._set(temperature=value)
[docs] def setTopK(self, value): """Sets the number of highest probability vocabulary tokens to keep for top-k-filtering. Parameters ---------- value : int Number of highest probability vocabulary tokens to keep """ return self._set(topK=value)
[docs] def setTopP(self, value): """Sets the top cumulative probability for vocabulary tokens. If set to float < 1, only the most probable tokens with probabilities that add up to ``topP`` or higher are kept for generation. Parameters ---------- value : float Cumulative probability for vocabulary tokens """ return self._set(topP=value)
[docs] def setRepetitionPenalty(self, value): """Sets the parameter for repetition penalty. 1.0 means no penalty. Parameters ---------- value : float The repetition penalty References ---------- See `Ctrl: A Conditional Transformer Language Model For Controllable Generation <https://arxiv.org/pdf/1909.05858.pdf>`__ for more details. """ return self._set(repetitionPenalty=value)
[docs] def setNoRepeatNgramSize(self, value): """Sets size of n-grams that can only occur once. If set to int > 0, all ngrams of that size can only occur once. Parameters ---------- value : int N-gram size can only occur once """ return self._set(noRepeatNgramSize=value)
@keyword_only def __init__(self, classname="com.johnsnowlabs.nlp.annotators.seq2seq.Phi2Transformer", java_model=None): super(Phi2Transformer, self).__init__(classname=classname, java_model=java_model) self._setDefault(minOutputLength=0, maxOutputLength=20, doSample=False, temperature=0.6, topK=50, topP=0.9, repetitionPenalty=1.0, noRepeatNgramSize=0, ignoreTokenIds=[], batchSize=1) @staticmethod
[docs] def loadSavedModel(folder, spark_session, use_openvino=False): """Loads a locally saved model. Parameters ---------- folder : str Folder of the saved model spark_session : pyspark.sql.SparkSession The current SparkSession Returns ------- Phi2Transformer The restored model """ from sparknlp.internal import _Phi2Loader jModel = _Phi2Loader(folder, spark_session._jsparkSession, use_openvino)._java_obj return Phi2Transformer(java_model=jModel)
@staticmethod
[docs] def pretrained(name="phi2", lang="en", remote_loc=None): """Downloads and loads a pretrained model. Parameters ---------- name : str, optional Name of the pretrained model, by default "phi2" lang : str, optional Language of the pretrained model, by default "en" remote_loc : str, optional Optional remote address of the resource, by default None. Will use Spark NLPs repositories otherwise. Returns ------- Phi2Transformer The restored model """ from sparknlp.pretrained import ResourceDownloader return ResourceDownloader.downloadModel(Phi2Transformer, name, lang, remote_loc)