ai4med-id/medgemma-1.5-4b-it-litertlm

MedGemma 1.5 4B IT — LiteRT LM format

A LiteRT-converted version of google/medgemma-1.5-4b-it.

Quantization settings

• Text model (tf_lite_prefill_decode) quantization recipe: dynamic_wi8_afp32
• Vision encoder quantization recipe: weight_only_wi8_afp32
• Quantization status: quantized

What is in this repository

• litertlm_medgemma-1.5-4b-it-fp8.litertlm: LiteRT-LM bundle for MedGemma 1.5 4B IT.
• The bundle is intended for LiteRT-compatible runtimes that support .litertlm artifacts.

Quick download

from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="ai4med-id/medgemma-1.5-4b-it-litertlm",
    filename="litertlm_medgemma-1.5-4b-it-fp8.litertlm",
)
print(path)

Conversion context

• Base model: google/medgemma-1.5-4b-it
• Conversion target format: LiteRT-LM (.litertlm)
• Prefill lengths: 128, 256, 512
• Cache length: 4096

Intended use

• Research and development for medical-domain multimodal assistants.
• Prototyping and experimentation on LiteRT-supported environments.

Limitations and safety

• This is not a medical device and must not be used as a substitute for professional clinical judgment.
• Model outputs can be incorrect, incomplete, or unsafe without human review.
• Evaluate thoroughly for your deployment context before any real-world use.

License and upstream terms

Please follow the license and usage terms of the upstream model: google/medgemma-1.5-4b-it.