Skip to content
Computer vision

smolvlm2-chartqa-sft-v2

LoRA-адаптер для вопросов по графикам

1 min read

Это LoRA-адаптер для HuggingFaceTB/SmolVLM2-2.2B-Instruct, дообученный отвечать на вопросы по изображениям графиков в рамках ChartQA. Часть задания курса Hugging Face smol, Unit 3. Обучение — SFT через TRL SFTTrainer, fp16 на одной Colab T4, около 1 ч 23 мин.

Ключевые особенности

  • Дообучен на первых 1 000 примерах train-сплита HuggingFaceM4/ChartQA.
  • LoRA: r=16, alpha=32, dropout 0.05; q/k/v/o/gate/up/down проекции.
  • Loss считается только на токенах ответа (completion_only_loss=True), 1 эпоха, 63 шага.
  • Собственная проверка на 150 вопросах test: 56.7% с адаптером против 49.3% у базовой модели.
  • Прирост скромный: адаптер может ошибаться при чтении графиков.

Пример

from peft import PeftModel
from transformers import AutoProcessor, AutoModelForImageTextToText
import torch

base = AutoModelForImageTextToText.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct", dtype=torch.float16, device_map="auto")
model = PeftModel.from_pretrained(base, "yas1988/smolvlm2-chartqa-sft-v2")
processor = AutoProcessor.from_pretrained("HuggingFaceTB/SmolVLM2-2.2B-Instruct")
ПодробнееБенчмарк: 85/150 (56.7%) с адаптером vs 74/150 (49.3%) без; разброс около ±4 п.п.
Обучение: lr 2e-4, batch 1 × grad accumulation 16; loss 0.57 → ~0.17, mean token accuracy 0.94.
Ограничения: всего 1 000 примеров, небольшая модель, возможны ошибки в чтении графиков.
Это не официальный скор ChartQA, а собственная проверка автора.
Discuss Telegram 0 42

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # yas1988/smolvlm2-chartqa-sft-v2
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download yas1988/smolvlm2-chartqa-sft-v2 --local-dir ~/models/smolvlm2-chartqa-sft-v2
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of smolvlm2-chartqa-sft-v2 (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

yas1988/smolvlm2-chartqa-sft-v2
Author
yas1988
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

420

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments