---
title: lora_v5_0414_Qwen3-VL-32B-Instruct
canonical_url: "https://www.modelscope.cn/datasets/revuener/lora_v5_0414_Qwen3-VL-32B-Instruct"
md_url: "https://www.modelscope.cn/datasets/revuener/lora_v5_0414_Qwen3-VL-32B-Instruct.md"
repository: revuener/lora_v5_0414_Qwen3-VL-32B-Instruct
chinese_name: "lora_v5_0414_50万个样本512~1024"
last_updated: 2026-04-15
license: other
storage_size: "7.0 GB"
downloads: 30
stars: 0
---

# lora_v5_0414_Qwen3-VL-32B-Instruct

> lora_v5_0414_Qwen3-VL-32B-Instruct - revuener 在 ModelScope 开源的数据集。train2100-01-01-01-01

revuener/lora_v5_0414_Qwen3-VL-32B-Instruct 是 ModelScope 魔搭社区上的数据集，存储大小 7.0 GB，采用 other 许可。

- **Repository**: revuener/lora_v5_0414_Qwen3-VL-32B-Instruct
- **License**: other
- **Storage size**: 7.0 GB
- **Downloads**: 30
- **Stars**: 0
- **Last updated**: 2026-04-15

Source: https://www.modelscope.cn/datasets/revuener/lora_v5_0414_Qwen3-VL-32B-Instruct

---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# train_2100-01-01-01-01

This model is a fine-tuned version of [/root/autodl-tmp/models/Qwen/Qwen3-VL-32B-Instruct](https://huggingface.co//root/autodl-tmp/models/Qwen/Qwen3-VL-32B-Instruct) on the v4_messages_512, the v4_messages_1024, the v4_messages_class_choice_512, the v4_messages_class_choice_1024, the v4_messages_coord_choice_512, the v4_messages_coord_choice_1024, the v4_train_class_512, the v4_train_class_1024, the v4_train_coord_512 and the v4_train_coord_1024 datasets.

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-06
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- training_steps: 3922

### Training results



### Framework versions

- PEFT 0.18.1
- Transformers 5.2.0
- Pytorch 2.10.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2