MMoshtaghi/Pixtral-12B-2409-LoRAAdpt-General

Uploaded model

• Developed by: MMoshtaghi
• License: apache-2.0
• Finetuned from model : unsloth/pixtral-12b-2409-unsloth-bnb-4bit
• Finetuned on dataset: unsloth/llava-instruct-mix-vsft-mini
• PEFT method : Quantized LoRA

Quick start

from datasets import load_dataset
from unsloth import FastVisionModel

model, tokenizer = FastVisionModel.from_pretrained(
    model_name = "MMoshtaghi/Pixtral-12B-2409-LoRAAdpt-General",
    load_in_4bit = True,
)
FastVisionModel.for_inference(model) # Enable for inference!

dataset = load_dataset("unsloth/llava-instruct-mix-vsft-mini", split = "train")
image = dataset[2]["images"][0]
instruction = "Is there something interesting about this image?"

messages = [
    {"role": "user", "content": [
        {"type": "image"},
        {"type": "text", "text": instruction}
    ]}
]
input_text = tokenizer.apply_chat_template(messages, add_generation_prompt = True)
inputs = tokenizer(
    image,
    input_text,
    add_special_tokens = False,
    return_tensors = "pt",
).to("cuda")

from transformers import TextStreamer
text_streamer = TextStreamer(tokenizer, skip_prompt = True)
_ = model.generate(**inputs, streamer = text_streamer, max_new_tokens = 64,
                   use_cache = True, temperature = 1.5, min_p = 0.1)

Framework versions

• TRL: 0.13.0
• Transformers: 4.47.1
• Pytorch: 2.5.1+cu121
• Datasets: 3.2.0
• Tokenizers: 0.21.0
• Unsloth: 2025.1.5

Training procedure

(Log-in required!)

Citations

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