MiniCPM-V 4.6 Fine-Tuning Tutorial (ms-swift)

Model and Task Overview

This section uses the counting task allenai/pixmo-count as a fine-tuning example.

Training task:

• Input an image and a counting question
• The target assistant response first lists the coordinates of each target object in x y format, for example, <point>573 489</point>, and then outputs the final count, such as 0, 3, or 10.

Environment Setup

Minimum runnable installation steps

conda create -n "MiniCPM-V-4.6-Counting" python=3.10 -y
conda activate "MiniCPM-V-4.6-Counting"

pip install torch==2.8.0 torchvision==0.23.0

pip install \
  transformers==5.7.0 accelerate==1.13.0 \
  deepspeed==0.18.3 peft==0.18.1 trl==0.24.0 \
  wandb ninja einops safetensors tokenizers sentencepiece

MAX_JOBS=32 NVCC_THREADS=4 pip install --no-build-isolation flash-attn==2.8.3
git clone https://github.com/modelscope/ms-swift.git
cd ms-swift
pip install -e .
pip install transformers==5.7.0

Note: MiniCPM-V 4.6 is officially supported by transformers>=5.7.0. Please make sure the final version of transformers in your environment is larger or equal to 5.7.0.

Reference dependency versions

python                        3.10.0
accelerate                    1.13.0
deepspeed                     0.18.3
flash_attn                    2.8.3
ms_swift                      latest official code
torch                         2.8.0
torchvision                   0.23.0
transformers                  5.7.0

2.2 Data Preparation

Download the dataset from allenai/pixmo-count and convert it into the ms-swift format.

Data format reference: json { "messages": [ { "content": "<image>\nCarefully observe the image. Are there any people in the image? If yes, please list their respective coordinates and provide the total count. If no, answer 0.", "role": "user" }, { "content": "<think>\n\n</think>\n\nThe respective coordinates of people: <point>236 469</point>, <point>307 232</point>, <point>362 434</point>, <point>485 521</point>, <point>487 340</point>, <point>615 386</point>, <point>735 441</point>, <point>870 615</point>. So the total count is 8.", "role": "assistant" } ], "images": [ "/path/to/images/*.jpg" ], "source_file": "pixmo-count", "orig_index": 1, "channel": "pixmo-count" } - For the Counting task, adding supervision on point prediction can improve fine-tuning performance. Therefore, we recommend concatenating the points coordinates from the dataset into the assistant response. - Since MiniCPM-V 4.6 normalizes image coordinates to 0~1000, the point coordinates also need to be transformed as follows: python def expected_norm(x_px: float, y_px: float, width: int, height: int) -> Tuple[int, int]: return int((x_px / width) * 1000.0), int((y_px / height) * 1000.0)

Launch Training

After configuring the model path, training set path, validation set path, and output directory, run the following script to start training.

run_swift.sh
#!/bin/bash
set -euo pipefail

export CUDA_VISIBLE_DEVICES="${CUDA_VISIBLE_DEVICES:-0,1,2,3,4,5,6,7}"
export NPROC_PER_NODE="${NPROC_PER_NODE:-8}"
export MASTER_PORT="${MASTER_PORT:-29632}"

export WANDB_API_KEY="${WANDB_API_KEY:-}"
export WANDB_PROJECT="${WANDB_PROJECT:-MiniCPMV46-Counting}"
export WANDB_RUN_NAME="${WANDB_RUN_NAME:-mcpmv46_count}"
export WANDB_NAME="${WANDB_NAME:-mcpmv46_count}"

export DOWNSAMPLE_MODE="${DOWNSAMPLE_MODE:-4x}"

SWIFT_BIN="${SWIFT_BIN:-swift}"
MODEL_PATH="${MODEL_PATH:-/path/to/minicpm-v-4_6}"

TRAIN_DATA="${TRAIN_DATA:-/path/to/task_dataset/train/pixmo_count_train_with_channel}"
VALID_DATA="${VALID_DATA:-/path/to/task_dataset/val/validation-00000-of-00001-swift.parquet}"

DEEPSPEED_CONFIG="${DEEPSPEED_CONFIG:-zero2}"
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
OUTPUT_DIR="${OUTPUT_DIR:-path/to/outdir}"

${SWIFT_BIN} sft \
  --model "${MODEL_PATH}" \
  --model_type minicpmv4_6 \
  --template minicpmv4_6 \
  --add_non_thinking_prefix true \
  --run_name "${WANDB_RUN_NAME}" \
  --dataset "${TRAIN_DATA}" \
  --val_dataset "${VALID_DATA}" \
  --deepspeed "${DEEPSPEED_CONFIG}" \
  --tuner_type full \
  --torch_dtype bfloat16 \
  --freeze_vit False \
  --packing false \
  --max_length 4096 \
  --num_train_epochs 4 \
  --per_device_train_batch_size 1 \
  --gradient_accumulation_steps 16 \
  --learning_rate 5e-6 \
  --warmup_ratio 0.05 \
  --logging_steps 1 \
  --save_steps 132 \
  --eval_strategy steps \
  --eval_steps 80 \
  --save_total_limit 30 \
  --load_from_cache_file false \
  --dataset_num_proc 16 \
  --dataloader_num_workers 16 \
  --enable_channel_loss True \
  --attn_impl flash_attn \
  --loss_scale ignore_empty_think \
  --output_dir "${OUTPUT_DIR}" \
  --report_to wandb

Key parameter notes

• Training supports two visual token compression ratios, 16x and 4x, which are controlled by export DOWNSAMPLE_MODE="${DOWNSAMPLE_MODE:-4x}".
• The current version of transformers still has issues with packed training for the Qwen3.5 series. For now, please use --packing false. This document will be updated once an official fix is available.
• --add_non_thinking_prefix true can append <think>\n\n</think>\n\n to training prompts, achieve no-thinking training.

Training Curves

https://wandb.ai/majy24-tsinghua-university/MiniCPMV46-Counting/reports/ms-swift---VmlldzoxNjgxMDk0Ng

Evaluation Results

Evaluation metrics explanation:

The table below shows the results under two visual token compression ratios:

[1]: MiniCPM-V 4.6 is the original model without fine-tuning, so only one Acc@0 result (Acc@0 Top1) is available and Acc@0 Avg.Top3 cannot be computed.

Output example:

```text Q: Carefully observe the image. Are there any airplanes in the image? If yes, please list their respective coordinates and provide the total count. If no, answer 0.

A: The respective coordinates of airplanes: 310 370, 365 277, 388 486, 405 185, 437 368, 474 611, 503 250, 527 451, 535 818, 597 331. So the total count is 10. ```