sparknlp.annotator.cv.gemma3_for_multimodal#

Classes#

class Gemma3ForMultiModal(classname='com.johnsnowlabs.nlp.annotators.cv.Gemma3ForMultiModal', java_model=None)[source]#

Gemma3ForMultiModal can load Gemma 3 Vision models for visual question answering. The model consists of a vision encoder, a text encoder, a text decoder and a model merger. The vision encoder will encode the input image, the text encoder will encode the input text, the model merger will merge the image and text embeddings, and the text decoder will output the answer.

Gemma 3 is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. It features: - Large 128K context window - Multilingual support in over 140 languages - Multimodal capabilities handling both text and image inputs - Optimized for deployment on limited resources (laptops, desktops, cloud)

Pretrained models can be loaded with pretrained() of the companion object:

>>> visualQA = Gemma3ForMultiModal.pretrained()     ...     .setInputCols(["image_assembler"])     ...     .setOutputCol("answer")

The default model is "gemma3_4b_it_int4", if no name is provided.

For available pretrained models please see the Models Hub.

Input Annotation types	Output Annotation type
IMAGE	DOCUMENT

Parameters:

batchSize

Batch size. Large values allows faster processing but requires more memory, by default 1

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 0.6

topK

The number of highest probability vocabulary tokens to keep for top-k-filtering, by default -1

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, by default 0.9

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 3

beamSize

The Number of beams for beam search, by default 1

maxInputLength

Maximum length of input text, by default 4096

Examples

>>> import sparknlp
>>> from sparknlp.base import *
>>> from sparknlp.annotator import *
>>> from pyspark.ml import Pipeline
>>> from pyspark.sql.functions import lit
>>> imageDF = spark.read.format("image").load(images_path)
>>> testDF = imageDF.withColumn("text", lit("<bos><start_of_turn>user\nYou are a helpful assistant.\n\n<start_of_image>Describe this image in detail.<end_of_turn>\n<start_of_turn>model\n"))
>>> imageAssembler = ImageAssembler() \
...     .setInputCol("image") \
...     .setOutputCol("image_assembler")
>>> visualQA = Gemma3ForMultiModal.pretrained() \
...     .setInputCols("image_assembler") \
...     .setOutputCol("answer")
>>> pipeline = Pipeline().setStages([
...     imageAssembler,
...     visualQA
... ])
>>> result = pipeline.fit(testDF).transform(testDF)
>>> result.select("image_assembler.origin", "answer.result").show(truncate=False)

name = 'Gemma3ForMultiModal'[source]#

inputAnnotatorTypes[source]#

outputAnnotatorType = 'document'[source]#

configProtoBytes[source]#

minOutputLength[source]#

maxOutputLength[source]#

doSample[source]#

temperature[source]#

topK[source]#

topP[source]#

repetitionPenalty[source]#

noRepeatNgramSize[source]#

ignoreTokenIds[source]#

beamSize[source]#

maxInputLength[source]#

setMaxSentenceSize(value)[source]#

Sets Maximum sentence length that the annotator will process, by default 50.

Parameters:

valueint

Maximum sentence length that the annotator will process

setIgnoreTokenIds(value)[source]#

A list of token ids which are ignored in the decoder’s output.

Parameters:

valueList[int]

The words to be filtered out

setConfigProtoBytes(b)[source]#

Sets configProto from tensorflow, serialized into byte array.

Parameters:

bList[int]

ConfigProto from tensorflow, serialized into byte array

setMinOutputLength(value)[source]#

Sets minimum length of the sequence to be generated.

Parameters:

valueint

Minimum length of the sequence to be generated

setMaxOutputLength(value)[source]#

Sets maximum length of output text.

Parameters:

valueint

Maximum length of output text

setDoSample(value)[source]#

Sets whether or not to use sampling, use greedy decoding otherwise.

Parameters:

valuebool

Whether or not to use sampling; use greedy decoding otherwise

setTemperature(value)[source]#

Sets the value used to module the next token probabilities.

Parameters:

valuefloat

The value used to module the next token probabilities

setTopK(value)[source]#

Sets the number of highest probability vocabulary tokens to keep for top-k-filtering.

Parameters:

valueint

Number of highest probability vocabulary tokens to keep

setTopP(value)[source]#

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:

valuefloat

Cumulative probability for vocabulary tokens

setRepetitionPenalty(value)[source]#

Sets the parameter for repetition penalty. 1.0 means no penalty.

Parameters:

valuefloat

The repetition penalty

References

See Ctrl: A Conditional Transformer Language Model For Controllable Generation for more details.

setNoRepeatNgramSize(value)[source]#

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:

valueint

N-gram size can only occur once

setBeamSize(value)[source]#

Sets the number of beam size for beam search, by default 4.

Parameters:

valueint

Number of beam size for beam search

static loadSavedModel(folder, spark_session, use_openvino=False)[source]#

Loads a locally saved model.

Parameters:

folderstr

Folder of the saved model

spark_sessionpyspark.sql.SparkSession

The current SparkSession

Returns:

Gemma3ForMultiModal

The restored model

static pretrained(name='gemma3_4b_it_int4', lang='en', remote_loc=None)[source]#

Downloads and loads a pretrained model.

Parameters:

namestr, optional

Name of the pretrained model, by default “gemma3_4b_it_int4”

langstr, optional

Language of the pretrained model, by default “en”

remote_locstr, optional

Optional remote address of the resource, by default None. Will use Spark NLPs repositories otherwise.

Returns:

Gemma3ForMultiModal

The restored model