hiieu/gemma-2-2b-it-lora-vi-en

from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("hiieu/gemma-2-2b-it-lora-vi-en")
tokenizer = tokenizer = AutoTokenizer.from_pretrained("hiieu/gemma-2-2b-it-lora-vi-en")
conversations = [
    [{"role": "user", "content": "Good morning everybody"}],
    [{"role": "user", "content": "Xin chào mọi người"}]
]

batch_input_ids = tokenizer.apply_chat_template(
    conversations,
    add_generation_prompt=True,
    return_tensors="pt",
    padding=True,
    truncation=True
).to(model.device)

outputs = model.generate(
    batch_input_ids,
    max_new_tokens=256,
    do_sample=True,
    temperature=0.6,
    top_p=0.9,
)

responses = outputs[:, batch_input_ids.shape[-1]:]

for response in responses:
    print(tokenizer.decode(response, skip_special_tokens=True))

>>> Chào mọi người
>>> Hello everyone

Uploaded model

• Developed by: hiieu
• License: apache-2.0
• Finetuned from model : unsloth/gemma-2-2b-it-bnb-4bit

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