qwen_tweet_generator_pro

This model is a fine-tuned version of Qwen/Qwen3-14B for German tweet generation.

Model Description

• Developed by: Giordano De Marzo
• Model type: Causal Language Model
• Language(s): German
• License: Apache 2.0
• Finetuned from model: Qwen/Qwen3-14B

Intended Uses

This model is designed to generate German tweets in a specific style learned from training data.

Direct Use

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("giordano-dm/qwen_tweet_generator_pro")
base_model = AutoModelForCausalLM.from_pretrained('Qwen/Qwen3-14B')
model = PeftModel.from_pretrained(base_model, 'giordano-dm/qwen_tweet_generator_pro')

# Generate tweet
messages = [
    {"role": "user", "content": "Schreib einen Tweet."}
]

formatted_text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=False
)

inputs = tokenizer(formatted_text, return_tensors="pt")
outputs = model.generate(
    **inputs,
    max_new_tokens=100,
    temperature=0.8,
    do_sample=True
)

response = tokenizer.decode(outputs[0][len(inputs.input_ids[0]):], skip_special_tokens=True)
print(response)

Training Details

Training Data

• Dataset: Custom German tweet dataset
• Data type: Social media posts
• Language: German

Training Procedure

• Training regime: LoRA fine-tuning
• Base model: Qwen/Qwen3-14B
• Training framework: Transformers + PEFT
• Hardware: GPU

Training Hyperparameters

• Training regime: LoRA (r=16, alpha=32, dropout=0.1)
• Learning rate: 1e-4
• Batch size: 8
• Epochs: 3

Limitations and Biases

• This model is trained on specific social media data and may reflect biases present in the training data
• Generated content should be reviewed before publication
• The model may not be suitable for all contexts or audiences

Ethical Considerations

• Users should be aware of potential biases in generated content
• Generated tweets should be fact-checked before sharing
• Consider the social impact of automated content generation