Instructions to use HiTZ/eu_Qwen3-14B-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HiTZ/eu_Qwen3-14B-Base with Transformers:

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

pipe = pipeline("text-generation", model="HiTZ/eu_Qwen3-14B-Base")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("HiTZ/eu_Qwen3-14B-Base")
model = AutoModelForCausalLM.from_pretrained("HiTZ/eu_Qwen3-14B-Base", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use HiTZ/eu_Qwen3-14B-Base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HiTZ/eu_Qwen3-14B-Base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HiTZ/eu_Qwen3-14B-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/HiTZ/eu_Qwen3-14B-Base

SGLang

How to use HiTZ/eu_Qwen3-14B-Base 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 "HiTZ/eu_Qwen3-14B-Base" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HiTZ/eu_Qwen3-14B-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "HiTZ/eu_Qwen3-14B-Base" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HiTZ/eu_Qwen3-14B-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use HiTZ/eu_Qwen3-14B-Base with Docker Model Runner:
```
docker model run hf.co/HiTZ/eu_Qwen3-14B-Base
```

eu_Qwen3-14B-Base

File size: 2,607 Bytes

---
library_name: transformers
pipeline_tag: text-generation
language:
- eu
license: apache-2.0
base_model:
- Qwen/Qwen3-14B-Base
---

# HiTZ/eu_Qwen3-14B-Base

This is a **Basque (eu) language-specific base language model** trained by the HiTZ Research Center, starting from **Qwen3-14B-Base** and further pretrained on curated Basque data.

This model is released as a **base model**, intended for further fine-tuning or adaptation (e.g., instruction tuning, domain adaptation).

---

## Training Data

To train language-specific base LLMs, we followed the methodology proposed by [Etxaniz et al. (2024)](https://aclanthology.org/2024.acl-long.799/), originally developed for Basque, and extended it to other low-resource languages. To enable fair comparisons across languages, we limited the corpus size for each language to roughly the same number of tokens. We also included a small English subset to mitigate catastrophic forgetting.

### Corpus composition

| Language | Documents | Tokens (Qwen3) |
|----------|-----------|---------------:|
| Basque (eu) | 4.2M | ~3.5B |
| English (en) | 0.5M | ~0.3B |

Token counts vary slightly depending on the tokenizer, but remain comparable in overall size.

### Data sources

Basque data was obtained from the Latxa corpus, which consists primarily of large-scale web-crawled content, news articles, and encyclopedic text.  
The English subset was sampled from the FineWeb corpus.

---

## Model Training

- Sequence length: 8,196 tokens  
- Effective batch size: 256 sequences  
- Tokens per optimization step: ~2M  
- Learning rate schedule: cosine decay with 10% warm-up  
- Peak learning rate: 1e-5  

Training was conducted on the CINECA Leonardo high-performance computing cluster using Fully Sharded Data Parallel (FSDP) across 32 nodes, each equipped with 4 NVIDIA A100 GPUs (64 GB).

---

## Getting Started

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "HiTZ/eu_Qwen3-14B-Base"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

inputs = tokenizer("Kaixo!", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Acknowledgements

This work has been partially supported by the Basque Government (Research group funding IT1570-22 and IKER-GAITU project), the Spanish Ministry for Digital Transformation and of Civil Service, and the EU-funded NextGenerationEU Recovery, Transformation and Resilience Plan (ILENIA project, 2022/TL22/00215335; and ALIA project).