Instructions to use sanchitahuja205/xelm-gemma-4b-slavic-freeze with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sanchitahuja205/xelm-gemma-4b-slavic-freeze with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="sanchitahuja205/xelm-gemma-4b-slavic-freeze")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("sanchitahuja205/xelm-gemma-4b-slavic-freeze")
model = AutoModelForMultimodalLM.from_pretrained("sanchitahuja205/xelm-gemma-4b-slavic-freeze")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use sanchitahuja205/xelm-gemma-4b-slavic-freeze with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sanchitahuja205/xelm-gemma-4b-slavic-freeze"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sanchitahuja205/xelm-gemma-4b-slavic-freeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/sanchitahuja205/xelm-gemma-4b-slavic-freeze

SGLang

How to use sanchitahuja205/xelm-gemma-4b-slavic-freeze with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "sanchitahuja205/xelm-gemma-4b-slavic-freeze" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sanchitahuja205/xelm-gemma-4b-slavic-freeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "sanchitahuja205/xelm-gemma-4b-slavic-freeze" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sanchitahuja205/xelm-gemma-4b-slavic-freeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use sanchitahuja205/xelm-gemma-4b-slavic-freeze with Docker Model Runner:
```
docker model run hf.co/sanchitahuja205/xelm-gemma-4b-slavic-freeze
```

xelm-gemma-4b-slavic-freeze

Layer-freezing strategy: middle transformer layers are frozen at the base Gemma-3-4B weights; only the first and last layers are updated during CPT. Mitigates catastrophic forgetting of general capabilities.

Base model: google/gemma-3-4b-pt
Strategy: freeze
Language family: Slavic
Code: https://github.com/sanchit-ahuja/scaling-multilingual-experts

Loading

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("sanchitahuja205/xelm-gemma-4b-slavic-freeze")
tokenizer = AutoTokenizer.from_pretrained("sanchitahuja205/xelm-gemma-4b-slavic-freeze")

Training recipe

The exact training recipe lives in configs/yaml/train_gemma_freeze.yaml in the code repo. The resolved config used for this specific run is also included in this model repo as training_config.yaml — load it with pyrallis to reproduce the run bit-for-bit:

python train.py --config_path configs/yaml/train_gemma_freeze.yaml

Citation

@misc{ahuja2026parameteralignmentmitigatescatastrophic,
      title={Parameter Alignment Mitigates Catastrophic Forgetting in Multilingual Expert Language Models},
      author={Sanchit Ahuja and Terra Blevins},
      year={2026},
      eprint={2606.00284},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2606.00284},
}