► KerasHub: Pretrained Models / API documentation / Model Architectures / Mistral3 / Mistral3CausalLMPreprocessor layer

Mistral3CausalLMPreprocessor layer

[source]

Mistral3CausalLMPreprocessor class

keras_hub.models.Mistral3CausalLMPreprocessor(
    tokenizer,
    image_converter=None,
    sequence_length=1024,
    add_start_token=True,
    add_end_token=True,
    spatial_merge_size=2,
    **kwargs
)

Mistral3 Causal LM preprocessor.

This preprocessing layer is meant for use with keras_hub.models.Mistral3CausalLM. It takes in batches of prompts and (optionally, per-prompt) images and returns outputs in a (x, y, sample_weight) format, where the y label is the next token id in the x sequence.

x for call()/generate_preprocess() should be a dict with "prompts" (and optionally "responses") and, for multimodal inputs, an "images" key. Images are matched to prompts by their "[IMG]" placeholder occurrences, consumed in order — "images" can be any reasonable nesting (a single image, a batched array, flat or per-prompt-grouped lists), as long as the total image count matches the total placeholder count. Omitting "images" (or passing x as a plain string/list of strings) preprocesses as plain text, matching HF's Mistral3ForConditionalGeneration, which also supports text-only calls.

For use with generation, the layer also exposes two methods generate_preprocess() and generate_postprocess(). When this preprocessor is attached to a keras_hub.models.Mistral3CausalLM instance, these methods will be called implicitly in generate(). They can also be called standalone (e.g. to precompute preprocessing inputs for generation in a separate process).

Arguments

  • tokenizer: A keras_hub.models.Mistral3Tokenizer instance.
  • image_converter: A keras_hub.layers.Mistral3ImageConverter instance or None.
  • sequence_length: The length of the packed inputs.
  • add_start_token: If True, the preprocessor will prepend the tokenizer start token to each input sequence. Default is True.
  • add_end_token: If True, the preprocessor will append the tokenizer end token to each input sequence. Default is True.
  • spatial_merge_size: int. The multimodal projector's spatial merge size, used to compute how many image placeholder tokens each image expands to. Defaults to 2.

Call arguments

  • x: A dict with "prompts" and, optionally, "images" keys.
  • y: Label data. Should always be None as the layer generates labels.
  • sample_weight: Label weights. Should always be None as the layer generates label weights.
  • sequence_length: Pass to override the configured sequence_length of the layer.

[source]

from_preset method

Mistral3CausalLMPreprocessor.from_preset(
    preset, config_file="preprocessor.json", **kwargs
)

Instantiate a keras_hub.models.Preprocessor from a model preset.

A preset is a directory of configs, weights and other file assets used to save and load a pre-trained model. The preset can be passed as one of:

  1. a built-in preset identifier like 'bert_base_en'
  2. a Kaggle Models handle like 'kaggle://user/bert/keras/bert_base_en'
  3. a Hugging Face handle like 'hf://user/bert_base_en'
  4. a path to a local preset directory like './bert_base_en'

For any Preprocessor subclass, you can run cls.presets.keys() to list all built-in presets available on the class.

As there are usually multiple preprocessing classes for a given model, this method should be called on a specific subclass like keras_hub.models.BertTextClassifierPreprocessor.from_preset().

Arguments

  • preset: string. A built-in preset identifier, a Kaggle Models handle, a Hugging Face handle, or a path to a local directory.

Examples

# Load a preprocessor for Gemma generation.
preprocessor = keras_hub.models.CausalLMPreprocessor.from_preset(
    "gemma_2b_en",
)

# Load a preprocessor for Bert classification.
preprocessor = keras_hub.models.TextClassifierPreprocessor.from_preset(
    "bert_base_en",
)
Preset Parameters Description
ministral_3_3b_base_2512_en 3.85B 3 billion parameter, 26-layer, pretrained Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_3b_instruct_2512_en 3.85B 3 billion parameter, 26-layer, instruction-tuned Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_3b_reasoning_2512_en 3.85B 3 billion parameter, 26-layer, reasoning-tuned Ministral 3 model with a Pixtral vision encoder for image input.
shieldstral_1.0_3b_en 3.85B 3 billion parameter, 26-layer Shieldstral model with a Pixtral vision encoder. The model checks text and image content against a safety policy written in natural language. The model returns one safety score for each input.
ministral_3_8b_base_2512_en 8.92B 8 billion parameter, 34-layer, pretrained Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_8b_instruct_2512_en 8.92B 8 billion parameter, 34-layer, instruction-tuned Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_8b_reasoning_2512_en 8.92B 8 billion parameter, 34-layer, reasoning-tuned Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_14b_base_2512_en 13.95B 14 billion parameter, 40-layer, pretrained Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_14b_instruct_2512_en 13.95B 14 billion parameter, 40-layer, instruction-tuned Ministral 3 model with a Pixtral vision encoder for image input.
ministral_3_14b_reasoning_2512_en 13.95B 14 billion parameter, 40-layer, reasoning-tuned Ministral 3 model with a Pixtral vision encoder for image input.
mistral_small_3.1_24b_base_2503_en 24.01B 24 billion parameter, 40-layer, pretrained Mistral3 model with a Pixtral vision encoder for image input.
mistral_small_3.1_24b_instruct_2503_en 24.01B 24 billion parameter, 40-layer, instruction-tuned Mistral3 model with a Pixtral vision encoder for image input.
mistral_small_3.2_24b_instruct_2506_en 24.01B 24 billion parameter, 40-layer, instruction-tuned Mistral3 model with a Pixtral vision encoder for image input. An updated version of mistral_small_3.1_24b_instruct_2503_en with improved instruction-following and reduced repetition.
devstral_small_2_24b_instruct_2512_en 24.01B 24 billion parameter, 40-layer, instruction-tuned Devstral Small 2 model for agentic software engineering, with a Pixtral vision encoder for image input.
magistral_small_2509_en 24.01B 24 billion parameter, 40-layer, reasoning-tuned Magistral Small model with a Pixtral vision encoder for image input.

tokenizer property

keras_hub.models.Mistral3CausalLMPreprocessor.tokenizer

The tokenizer used to tokenize strings.


image_converter property

keras_hub.models.Mistral3CausalLMPreprocessor.image_converter

The image converter used to preprocess image data.