# Overview

This is the converted tokenizer for [tencent/KaLM-Embedding-Gemma3-12B-2511](https://huggingface.co/tencent/KaLM-Embedding-Gemma3-12B-2511/) to make it compatible with `transformers>=5.0.0` (and `sentence-transformers>=5.3.0`).

To load the model with `sentence-transfomrers` you can use:
```python
import sentence_transformers.sentence_transformer.modules as st_modules
from sentence_transformers import (
    SentenceTransformer,
)

transformer_module = st_modules.Transformer(
    "tencent/KaLM-Embedding-Gemma3-12B-2511",
    tokenizer_name_or_path="minhnguyent546/KaLM-Embedding-Gemma3-12B-2511-tokenizer-for-transformers-v5",
    model_kwargs={
        "torch_dtype": "bfloat16",
        "attn_implementation": "flash_attention_2",
    },
)

pooling_module = st_modules.Pooling(
    transformer_module.get_embedding_dimension(),
    pooling_mode="lasttoken",
    include_prompt=True,
)

normalize_module = st_modules.Normalize()
model = SentenceTransformer(modules=[transformer_module, pooling_module, normalize_module])
```