xrusnack/lora_model

Uploaded model

• Finetuned from model : unsloth/gemma-2-9b-bnb-4bit

This gemma2 model was trained 2x faster with Unsloth and Huggingface's TRL library.

Training

The gpt-4o-mini model was used to summarize 100 of the text examples in this dataset https://huggingface.co/datasets/vojtam/czech_books_descriptions The lora model was trained on these summaries.

Example of Inference:

alpaca_prompt = "### Text: {} ### Summary: {}"
FastLanguageModel.for_inference(model)
inputs = tokenizer(
[
    alpaca_prompt.format(
        "", # text to summarize
        "", # output - leave this blank for generation!
    )
], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 64, use_cache = True)
tokenizer.batch_decode(outputs)