com.johnsnowlabs.nlp.annotators.cv.MLLamaForMultimodal.scala Maven / Gradle / Ivy
/*
* Copyright 2017-2024 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.
*/
package com.johnsnowlabs.nlp.annotators.cv
import com.johnsnowlabs.ml.ai.util.Generation.GenerationConfig
import com.johnsnowlabs.ml.ai.MLLama
import com.johnsnowlabs.ml.onnx.OnnxWrapper.DecoderWrappers
import com.johnsnowlabs.ml.util.LoadExternalModel.{
loadJsonStringAsset,
loadTextAsset,
modelSanityCheck,
notSupportedEngineError
}
import com.johnsnowlabs.nlp.annotators.cv.feature_extractor.Preprocessor
import com.johnsnowlabs.ml.util.Openvino
import com.johnsnowlabs.nlp.AnnotatorType.{DOCUMENT, IMAGE}
import com.johnsnowlabs.nlp._
import org.json4s.{DefaultFormats, JValue}
import org.json4s.jackson.JsonMethods.parse
import com.johnsnowlabs.ml.openvino.{OpenvinoWrapper, ReadOpenvinoModel, WriteOpenvinoModel}
import com.johnsnowlabs.ml.openvino.OpenvinoWrapper.MLLamaWrappers
import com.johnsnowlabs.nlp.serialization.{MapFeature, StructFeature}
import org.apache.spark.broadcast.Broadcast
import org.apache.spark.ml.param.{IntArrayParam, IntParam}
import org.apache.spark.ml.util.Identifiable
import org.apache.spark.sql.SparkSession
/** MLLamaForMultimodal can load LLAMA 3.2 Vision models for visual question answering. The model
* consists of a vision encoder, a text encoder as well as a text decoder. The vision encoder
* will encode the input image, the text encoder will encode the input question together with the
* encoding of the image, and the text decoder will output the answer to the question.
*
* The Llama 3.2-Vision collection of multimodal large language models (LLMs) is a collection of
* pretrained and instruction-tuned image reasoning generative models in 11B and 90B sizes (text
* + images in / text out). The Llama 3.2-Vision instruction-tuned models are optimized for
* visual recognition, image reasoning, captioning, and answering general questions about an
* image. The models outperform many of the available open source and closed multimodal models on
* common industry benchmarks.
*
* Pretrained models can be loaded with `pretrained` of the companion object:
* {{{
* val visualQA = MLLamaForMultimodal.pretrained()
* .setInputCols("image_assembler")
* .setOutputCol("answer")
* }}}
* The default model is `"llama_3_2_11b_vision_instruct_int4"`, if no name is provided.
*
* For available pretrained models please see the
* [[https://sparknlp.org/models?task=Question+Answering Models Hub]].
*
* Models from the HuggingFace 🤗 Transformers library are also compatible with Spark NLP 🚀. To
* see which models are compatible and how to import them see
* [[https://github.com/JohnSnowLabs/spark-nlp/discussions/5669]] and to see more extended
* examples, see
* [[https://github.com/JohnSnowLabs/spark-nlp/blob/master/src/test/scala/com/johnsnowlabs/nlp/annotators/cv/MLLamaForMultimodalTest.scala]].
*
* ==Example==
* {{{
* import spark.implicits._
* import com.johnsnowlabs.nlp.base._
* import com.johnsnowlabs.nlp.annotator._
* import org.apache.spark.ml.Pipeline
*
* val imageDF: DataFrame = ResourceHelper.spark.read
* .format("image")
* .option("dropInvalid", value = true)
* .load(imageFolder)
*
* val testDF: DataFrame = imageDF.withColumn("text", lit("<|begin_of_text|><|begin_of_text|><|start_header_id|>user<|end_header_id|>\n\n<|image|>What is unusual on this image?<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n"))
*
* val imageAssembler: ImageAssembler = new ImageAssembler()
* .setInputCol("image")
* .setOutputCol("image_assembler")
*
* val visualQAClassifier = MLLamaForMultimodal.pretrained()
* .setInputCols("image_assembler")
* .setOutputCol("answer")
*
* val pipeline = new Pipeline().setStages(Array(
* imageAssembler,
* visualQAClassifier
* ))
*
* val result = pipeline.fit(testDF).transform(testDF)
*
* result.select("image_assembler.origin", "answer.result").show(false)
* +--------------------------------------+------+
* |origin |result|
* +--------------------------------------+------+
* |[file:///content/images/cat_image.jpg]|[The unusual aspect of this picture is the presence of two cats lying on a pink couch]|
* +--------------------------------------+------+
* }}}
*
* @see
* [[CLIPForZeroShotClassification]] for Zero Shot Image Classifier
* @see
* [[https://sparknlp.org/docs/en/annotators Annotators Main Page]] for a list of transformer
* based classifiers
* @param uid
* required uid for storing annotator to disk
* @groupname anno Annotator types
* @groupdesc anno
* Required input and expected output annotator types
* @groupname Ungrouped Members
* @groupname param Parameters
* @groupname setParam Parameter setters
* @groupname getParam Parameter getters
* @groupname Ungrouped Members
* @groupprio param 1
* @groupprio anno 2
* @groupprio Ungrouped 3
* @groupprio setParam 4
* @groupprio getParam 5
* @groupdesc param
* A list of (hyper-)parameter keys this annotator can take. Users can set and get the
* parameter values through setters and getters, respectively.
*/
class MLLamaForMultimodal(override val uid: String)
extends AnnotatorModel[MLLamaForMultimodal]
with HasBatchedAnnotateImage[MLLamaForMultimodal]
with HasImageFeatureProperties
with WriteOpenvinoModel
with HasGeneratorProperties
with HasEngine {
/** Annotator reference id. Used to identify elements in metadata or to refer to this annotator
* type
*/
def this() = this(Identifiable.randomUID("MLLamaForMultimodal"))
/** Annotator reference id. Used to identify elements in metadata or to refer to this annotator
* type
*/
override val inputAnnotatorTypes: Array[AnnotatorType] = Array(IMAGE)
override val outputAnnotatorType: AnnotatorType = DOCUMENT
/** @group setParam */
def setRandomSeed(value: Int): MLLamaForMultimodal.this.type = {
if (randomSeed.isEmpty) {
this.randomSeed = Some(value)
}
this
}
/** A list of token ids which are ignored in the decoder's output (Default: `Array()`)
*
* @group param
*/
var ignoreTokenIds = new IntArrayParam(
this,
"ignoreTokenIds",
"A list of token ids which are ignored in the decoder's output")
/** @group setParam */
def setIgnoreTokenIds(tokenIds: Array[Int]): MLLamaForMultimodal.this.type = {
set(ignoreTokenIds, tokenIds)
}
/** @group getParam */
def getIgnoreTokenIds: Array[Int] = $(ignoreTokenIds)
/** Vocabulary used to encode the words to ids with bpeTokenizer.encode
*
* @group param
*/
val vocabulary: MapFeature[String, Int] = new MapFeature(this, "vocabulary").setProtected()
/** @group setParam */
def setVocabulary(value: Map[String, Int]): this.type = set(vocabulary, value)
/** Holding merges.txt coming from RoBERTa model
*
* @group param
*/
val merges: MapFeature[(String, String), Int] = new MapFeature(this, "merges").setProtected()
/** @group setParam */
def setMerges(value: Map[(String, String), Int]): this.type = set(merges, value)
/** Additional tokens to be added to the vocabulary
*
* @group param
*/
val addedTokens: MapFeature[String, Int] = new MapFeature(this, "addedTokens").setProtected()
/** @group setParam */
def setAddedTokens(value: Map[String, Int]): this.type = set(addedTokens, value)
/** Stop tokens to terminate the generation
*
* @group param
*/
override val stopTokenIds =
new IntArrayParam(this, "stopTokenIds", "Stop tokens to terminate the generation")
/** @group setParam */
override def setStopTokenIds(value: Array[Int]): this.type = {
set(stopTokenIds, value)
}
/** @group getParam */
override def getStopTokenIds: Array[Int] = $(stopTokenIds)
private var _model: Option[Broadcast[MLLama]] = None
val generationConfig: StructFeature[GenerationConfig] =
new StructFeature(this, "generationConfig").setProtected()
def setGenerationConfig(value: GenerationConfig): this.type =
set(generationConfig, value)
def getGenerationConfig: GenerationConfig = $$(generationConfig)
val imageToken =
new IntParam(this, "imageToken", "Token id for image embeddings")
/** @group setParam */
def setImageToken(value: Int): this.type = set(imageToken, value)
/** @group getParam */
def getImageToken: Int = $(imageToken)
val maxImageTiles =
new IntParam(this, "maxImageTiles", "Maximum number of image tiles")
/** @group setParam */
def setMaxImageTiles(value: Int): this.type = set(maxImageTiles, value)
/** @group getParam */
def getMaxImageTiles: Int = $(maxImageTiles)
val numVisionTokens =
new IntParam(this, "numVisionTokens", "Number of vision tokens")
/** @group setParam */
def setNumVisionTokens(value: Int): this.type = set(numVisionTokens, value)
/** @group getParam */
def getNumVisionTokens: Int = $(numVisionTokens)
val paddingConstant =
new IntParam(this, "paddingConstant", "Padding constant for the model. Default is 0")
/** @group setParam */
def setPaddingConstant(value: Int): this.type = set(paddingConstant, value)
/** @group getParam */
def getPaddingConstant: Int = $(paddingConstant)
/** @group setParam */
def setModelIfNotSet(
spark: SparkSession,
preprocessor: Preprocessor,
onnxWrappers: Option[DecoderWrappers],
openvinoWrapper: Option[MLLamaWrappers]): this.type = {
if (_model.isEmpty) {
_model = Some(
spark.sparkContext.broadcast(
new MLLama(
onnxWrappers,
openvinoWrapper,
$$(merges),
$$(vocabulary),
$$(addedTokens),
preprocessor,
generationConfig = getGenerationConfig,
imageToken = getImageToken,
maxImageTiles = getMaxImageTiles,
numVisionTokens = getNumVisionTokens,
paddingConstant = getPaddingConstant)))
}
this
}
/** @group getParam */
def getModelIfNotSet: MLLama = _model.get.value
setDefault(
minOutputLength -> 0,
maxOutputLength -> 20,
doSample -> false,
temperature -> 0.6,
topK -> -1,
topP -> 0.9,
repetitionPenalty -> 1.0,
noRepeatNgramSize -> 3,
ignoreTokenIds -> Array(),
batchSize -> 1,
beamSize -> 1,
maxInputLength -> 4096,
stopTokenIds -> Array(128001, 128008, 128009),
imageToken -> 128256,
maxImageTiles -> 576,
numVisionTokens -> 1601,
paddingConstant -> 0)
/** takes a document and annotations and produces new annotations of this annotator's annotation
* type
*
* @param batchedAnnotations
* Annotations in batches that correspond to inputAnnotationCols generated by previous
* annotators if any
* @return
* any number of annotations processed for every batch of input annotations. Not necessary
* one to one relationship
*/
override def batchAnnotate(
batchedAnnotations: Seq[Array[AnnotationImage]]): Seq[Seq[Annotation]] = {
batchedAnnotations
// .filter { annotationImages =>
// annotationImages.exists(_.text.nonEmpty)
// }
.map { cleanAnnotationImages =>
val validImages = cleanAnnotationImages.filter(_.result.nonEmpty)
val questionAnnotations = extractInputAnnotation(validImages)
getModelIfNotSet.predict(
questionAnnotations,
validImages.toSeq,
batchSize = $(batchSize),
minOutputLength = $(minOutputLength),
maxOutputLength = $(maxOutputLength),
doSample = $(doSample),
temperature = $(temperature),
topK = $(topK),
topP = $(topP),
repetitionPenalty = $(repetitionPenalty),
noRepeatNgramSize = $(noRepeatNgramSize),
randomSeed = this.randomSeed,
ignoreTokenIds = $(ignoreTokenIds),
beamSize = $(beamSize),
maxInputLength = $(maxInputLength))
}
}
private def extractInputAnnotation(
annotationImages: Array[AnnotationImage]): Seq[Annotation] = {
val questions = annotationImages.map(annotationImage => {
val imageText =
if (annotationImage.text.nonEmpty) annotationImage.text
else
"""<|begin_of_text|><|begin_of_text|><|start_header_id|>user<|end_header_id|>\n""" +
"""\n<|image|>This is an image<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n""".stripMargin // default question
Annotation(imageText)
})
questions
}
override def onWrite(path: String, spark: SparkSession): Unit = {
super.onWrite(path, spark)
getEngine match {
case Openvino.name =>
val wrappers = getModelIfNotSet.openvinoWrapper
writeOpenvinoModels(
path,
spark,
Seq(
(
wrappers.get.languageModel,
"llm_int4_asym_r10_gs64_max_activation_variance_awq_scale_all_layers.xml")),
MLLamaForMultimodal.suffix)
writeOpenvinoModels(
path,
spark,
Seq((wrappers.get.visionEmbeddingsModel, "openvino_vision_encoder.xml")),
MLLamaForMultimodal.suffix)
writeOpenvinoModels(
path,
spark,
Seq((wrappers.get.reshapeModel, "openvino_reshape_model.xml")),
MLLamaForMultimodal.suffix)
case _ =>
throw new Exception(notSupportedEngineError)
}
}
}
trait ReadablePretrainedMLLamaForMultimodal
extends ParamsAndFeaturesReadable[MLLamaForMultimodal]
with HasPretrained[MLLamaForMultimodal] {
override val defaultModelName: Some[String] = Some("llama_3_2_11b_vision_instruct_int4")
/** Java compliant-overrides */
override def pretrained(): MLLamaForMultimodal = super.pretrained()
override def pretrained(name: String): MLLamaForMultimodal =
super.pretrained(name)
override def pretrained(name: String, lang: String): MLLamaForMultimodal =
super.pretrained(name, lang)
override def pretrained(name: String, lang: String, remoteLoc: String): MLLamaForMultimodal =
super.pretrained(name, lang, remoteLoc)
}
trait ReadMLLamaForMultimodalDLModel extends ReadOpenvinoModel {
this: ParamsAndFeaturesReadable[MLLamaForMultimodal] =>
val suffix: String = "_mllama"
override val openvinoFile: String = "mllama_openvino"
def readModel(instance: MLLamaForMultimodal, path: String, spark: SparkSession): Unit = {
instance.getEngine match {
case Openvino.name =>
val languageModelWrappers =
readOpenvinoModels(
path,
spark,
Seq("llm_int4_asym_r10_gs64_max_activation_variance_awq_scale_all_layers.xml"),
suffix)
val visionEmbeddingsModelWrappers =
readOpenvinoModels(path, spark, Seq("openvino_vision_encoder.xml"), suffix)
val reshapeModelWrapper =
readOpenvinoModels(path, spark, Seq("openvino_reshape_model.xml"), suffix)
val ovWrapper = MLLamaWrappers(
languageModel = languageModelWrappers(
"llm_int4_asym_r10_gs64_max_activation_variance_awq_scale_all_layers.xml"),
visionEmbeddingsModel = visionEmbeddingsModelWrappers("openvino_vision_encoder.xml"),
reshapeModel = reshapeModelWrapper("openvino_reshape_model.xml"))
val preprocessor = Preprocessor(
do_normalize = true,
do_resize = true,
"LLAVAFeatureExtractor",
instance.getImageMean,
instance.getImageStd,
instance.getResample,
instance.getSize)
instance.setModelIfNotSet(spark, preprocessor, None, Some(ovWrapper))
case _ => {
throw new Exception(notSupportedEngineError)
}
}
}
addReader(readModel)
def loadSavedModel(
modelPath: String,
spark: SparkSession,
useOpenvino: Boolean = false): MLLamaForMultimodal = {
implicit val formats: DefaultFormats.type = DefaultFormats // for json4
val (localModelPath, detectedEngine) =
modelSanityCheck(
modelPath,
isDecoder = false,
custom = Some(
List(
"llm_int4_asym_r10_gs64_max_activation_variance_awq_scale_all_layers",
"openvino_vision_encoder",
"openvino_reshape_model")))
val modelConfig: JValue =
parse(loadJsonStringAsset(localModelPath, "config.json"))
val preprocessorConfigJsonContent =
loadJsonStringAsset(localModelPath, "preprocessor_config.json")
val preprocessorConfig = Preprocessor.loadPreprocessorConfig(preprocessorConfigJsonContent)
val parsedPreprocessorConfig: JValue = parse(preprocessorConfigJsonContent)
val beginSuppressTokens: Array[Int] =
(modelConfig \ "begin_suppress_tokens").extract[Array[Int]]
val suppressTokenIds: Array[Int] =
(modelConfig \ "suppress_tokens").extract[Array[Int]]
val forcedDecoderIds: Array[(Int, Int)] =
(modelConfig \ "forced_decoder_ids").extract[Array[Array[Int]]].map {
case idxWithTokenId: Array[Int] if idxWithTokenId.length == 2 =>
(idxWithTokenId(0), idxWithTokenId(1))
case _ =>
throw new Exception(
"Could not extract forced_decoder_ids. Should be a list of tuples with 2 entries.")
}
val maxImageTiles = (parsedPreprocessorConfig \ "max_image_tiles").extract[Int]
def arrayOrNone[T](array: Array[T]): Option[Array[T]] =
if (array.nonEmpty) Some(array) else None
val generationConfig: JValue =
parse(loadJsonStringAsset(localModelPath, "generation_config.json"))
val bosTokenId = (generationConfig \ "bos_token_id").extract[Int]
val eosTokenIdArray = (generationConfig \ "eos_token_id").extract[Array[Int]]
val eosTokenId = eosTokenIdArray.head
val padTokenId = (generationConfig \ "pad_token_id").extract[Int]
val vocabSize = (modelConfig \ "text_config" \ "vocab_size").extract[Int]
val imageToken = (modelConfig \ "image_token_index").extract[Int]
val imageSize = (modelConfig \ "vision_config" \ "image_size").extract[Int]
val patchSize = (modelConfig \ "vision_config" \ "patch_size").extract[Int]
val numVisionTokens = Math
.pow(imageSize / patchSize, 2)
.toInt + 1
// val numVisionTokens = Math
// .pow(
// ((modelConfig \ "vision_config" \ "image_size")
// .extract[Int] / (modelConfig \ "vision_config" \ "patch_size").extract[Int]).toInt,
// 2)
// .toInt + 1
// Check if tokenizer.json exists
val tokenizerPath = s"$localModelPath/assets/tokenizer.json"
val tokenizerExists = new java.io.File(tokenizerPath).exists()
val (vocabs, addedTokens, bytePairs) = if (tokenizerExists) {
val tokenizerConfig: JValue = parse(loadJsonStringAsset(localModelPath, "tokenizer.json"))
// extract vocab from tokenizer.json ( model -> vocab)
var vocabs: Map[String, Int] =
(tokenizerConfig \ "model" \ "vocab").extract[Map[String, Int]]
// extract merges from tokenizer.json ( model -> merges)
val bytePairs = (tokenizerConfig \ "model" \ "merges")
.extract[List[Array[String]]]
.filter(w => w.length == 2)
.map { case Array(c1, c2) => (c1, c2) }
.zipWithIndex
.toMap
// extract added_tokens from tokenizer.json (added_tokens)
// "added_tokens": [
// {
// "id": 128000,
// "content": "<|begin_of_text|>",
// "single_word": false,
// "lstrip": false,
// "rstrip": false,
// "normalized": false,
// "special": true
// }, ...
// ]
val addedTokens = (tokenizerConfig \ "added_tokens")
.extract[List[Map[String, Any]]]
.map { token =>
val id = token("id").asInstanceOf[BigInt].intValue()
val content = token("content").asInstanceOf[String]
(content, id)
}
.toMap
// update vocab with added tokens
addedTokens.foreach { case (content, id) =>
vocabs += (content -> id)
}
(vocabs, addedTokens, bytePairs)
} else {
val vocabs = loadTextAsset(localModelPath, "vocab.txt").zipWithIndex.toMap
val addedTokens = loadTextAsset(localModelPath, "added_tokens.txt").zipWithIndex.toMap
val bytePairs = loadTextAsset(localModelPath, "merges.txt")
.map(_.split(" "))
.filter(w => w.length == 2)
.map { case Array(c1, c2) => (c1, c2) }
.zipWithIndex
.toMap
(vocabs, addedTokens, bytePairs)
}
// val vocabSize = vocabs.size
val annotatorModel = new MLLamaForMultimodal()
.setGenerationConfig(
GenerationConfig(
bosTokenId,
padTokenId,
eosTokenId,
vocabSize,
arrayOrNone(beginSuppressTokens),
arrayOrNone(suppressTokenIds),
arrayOrNone(forcedDecoderIds)))
.setVocabulary(vocabs)
.setMerges(bytePairs)
.setAddedTokens(addedTokens)
.setImageToken(imageToken)
.setMaxImageTiles(maxImageTiles)
.setNumVisionTokens(numVisionTokens)
.setSize(preprocessorConfig.size)
.setImageMean(preprocessorConfig.image_mean)
.setImageStd(preprocessorConfig.image_std)
.setResample(preprocessorConfig.resample)
val modelEngine =
if (useOpenvino)
Openvino.name
else
detectedEngine
annotatorModel.set(annotatorModel.engine, modelEngine)
detectedEngine match {
case Openvino.name =>
val visionWrapper =
OpenvinoWrapper.read(
spark,
localModelPath,
zipped = false,
useBundle = true,
detectedEngine = detectedEngine,
modelName = "openvino_vision_encoder")
val reshapeWrapper =
OpenvinoWrapper.read(
spark,
localModelPath,
zipped = false,
useBundle = true,
detectedEngine = detectedEngine,
modelName = "openvino_reshape_model")
val languageModelWrapper =
OpenvinoWrapper.read(
spark,
localModelPath,
zipped = false,
useBundle = true,
detectedEngine = detectedEngine,
modelName = "llm_int4_asym_r10_gs64_max_activation_variance_awq_scale_all_layers")
val openvinoWrapper = MLLamaWrappers(
languageModel = languageModelWrapper,
visionEmbeddingsModel = visionWrapper,
reshapeModel = reshapeWrapper)
annotatorModel.setModelIfNotSet(spark, preprocessorConfig, None, Some(openvinoWrapper))
case _ =>
throw new Exception(notSupportedEngineError)
}
annotatorModel
}
}
object MLLamaForMultimodal
extends ReadablePretrainedMLLamaForMultimodal
with ReadMLLamaForMultimodalDLModel
© 2015 - 2025 Weber Informatics LLC | Privacy Policy