mirror of
https://www.modelscope.cn/Tencent-Hunyuan/HunyuanVideo-1.5.git
synced 2026-04-02 05:42:53 +08:00
update README (batch 1/1)
This commit is contained in:
Cherrytest
30
README.md
30
README.md
@ -57,7 +57,8 @@ HunyuanVideo-1.5 is a video generation model that delivers top-tier quality with
|
||||
</p>
|
||||
|
||||
## 🔥🔥🔥 News
|
||||
* 🚀 Dec 05, 2025: **New Release**: We now release the [480p I2V step-distilled model](https://huggingface.co/tencent/HunyuanVideo-1.5/tree/main/transformer/480p_i2v_step_distilled), which generates videos in 8 or 12 steps (recommended)! On RTX 4090, end-to-end generation time is reduced by 75%, and a single RTX 4090 can generate videos within 75 seconds. The step-distilled model maintains comparable quality to the original model while achieving significant speedup. See [Step Distillation Comparison](./assets/step_distillation_comparison.md) for detailed quality comparisons. For even faster generation, you can also try 4 steps (faster speed with slightly reduced quality). **To enable the step-distilled model, run `generate.py` with the `--enable_step_distill` parameter.** See [Usage](#-usage) for detailed usage instructions. 🔥🔥🔥🆕
|
||||
* 🚀 Dec 09, 2025: LoRA tuning script is released, enjoy it! 🔥🔥🔥🆕
|
||||
* 🚀 Dec 05, 2025: **New Release**: We now release the [480p I2V step-distilled model](https://huggingface.co/tencent/HunyuanVideo-1.5/tree/main/transformer/480p_i2v_step_distilled), which generates videos in 8 or 12 steps (recommended)! On RTX 4090, end-to-end generation time is reduced by 75%, and a single RTX 4090 can generate videos within **75 seconds**. The step-distilled model maintains comparable quality to the original model while achieving significant speedup. See [Step Distillation Comparison](./assets/step_distillation_comparison.md) for detailed quality comparisons. For even faster generation, you can also try 4 steps (faster speed with slightly reduced quality). **To enable the step-distilled model, run `generate.py` with the `--enable_step_distill` parameter.** See [Usage](#-usage) for detailed usage instructions. 🔥🔥🔥🆕
|
||||
* 📚 Dec 05, 2025: **Training Code Released**: We now open-source the training code for HunyuanVideo-1.5! The training script (`train.py`) provides a full training pipeline with support for distributed training, FSDP, context parallel, gradient checkpointing, and more. HunyuanVideo-1.5 is trained using the Muon optimizer, which we have open-sourced in the [Training](#-training) section. **If you would like to continue training our model or fine-tune it with LoRA, please use the Muon optimizer.** See [Training](#-training) section for detailed usage instructions. 🔥🔥🔥🆕
|
||||
* 🎉 **Diffusers Support**: HunyuanVideo-1.5 is now available on Hugging Face Diffusers! Check out [Diffusers collection](https://huggingface.co/collections/hunyuanvideo-community/hunyuanvideo-15) for easy integration. 🔥🔥🔥🆕
|
||||
* 🚀 Nov 27, 2025: We now support cache inference (deepcache, teacache, taylorcache), achieving significant speedup! Pull the latest code to try it. 🔥🔥🔥🆕
|
||||
@ -498,6 +499,11 @@ torchrun --nproc_per_node=8 train.py \
|
||||
| `--i2v_prob` | Probability of i2v task for video data | 0.3 |
|
||||
| `--use_muon` | Use Muon optimizer | true |
|
||||
| `--resume_from_checkpoint` | Resume from checkpoint directory | None |
|
||||
| `--use_lora` | Enable LoRA fine-tuning | false |
|
||||
| `--lora_r` | LoRA rank | 8 |
|
||||
| `--lora_alpha` | LoRA alpha scaling parameter | 16 |
|
||||
| `--lora_dropout` | LoRA dropout rate | 0.0 |
|
||||
| `--pretrained_lora_path` | Path to pretrained LoRA adapter | None |
|
||||
|
||||
#### 4. Monitor Training
|
||||
|
||||
@ -514,6 +520,28 @@ python train.py \
|
||||
--resume_from_checkpoint ./outputs/checkpoint-1000
|
||||
```
|
||||
|
||||
#### 6. LoRA Fine-tuning
|
||||
|
||||
To enable LoRA fine-tuning, add `--use_lora` to your training command. LoRA adapters will be saved in the checkpoint directory under `lora/`:
|
||||
|
||||
```bash
|
||||
torchrun --nproc_per_node=8 train.py \
|
||||
--pretrained_model_root ./ckpts \
|
||||
--use_lora \
|
||||
--lora_r 8 \
|
||||
--lora_alpha 16 \
|
||||
--learning_rate 1e-4 \
|
||||
--output_dir ./outputs
|
||||
```
|
||||
|
||||
To load a pretrained LoRA adapter, use `--pretrained_lora_path`:
|
||||
```bash
|
||||
torchrun --nproc_per_node=8 train.py \
|
||||
--pretrained_model_root ./ckpts \
|
||||
--use_lora \
|
||||
--pretrained_lora_path ./outputs/checkpoint-1000/lora/default
|
||||
```
|
||||
|
||||
|
||||
## 📊 Evaluation
|
||||
|
||||
|
||||
30
README_CN.md
30
README_CN.md
@ -40,7 +40,8 @@ HunyuanVideo-1.5作为一款轻量级视频生成模型，仅需83亿参数即
|
||||
</p>
|
||||
|
||||
## 🔥🔥🔥 最新动态
|
||||
* 🚀 Dec 05, 2025: **新模型发布**：我们现已发布 [480p I2V 步数蒸馏模型](https://huggingface.co/tencent/HunyuanVideo-1.5/tree/main/transformer/480p_i2v_step_distilled)，建议使用 8 或 12 步生成视频！在 RTX 4090 上，端到端生成耗时减少 75%，单卡 RTX 4090 可在 75 秒内生成视频。步数蒸馏模型在保持与原模型相当质量的同时实现了显著的加速。详细的质量对比请参见[步数蒸馏对比文档](./assets/step_distillation_comparison.md)。如需更快的生成速度，您也可以尝试使用4步推理（速度更快，质量略有下降）。**启用步数蒸馏模型，请运行 `generate.py` 并使用 `--enable_step_distill` 参数。** 详细的使用说明请参见[使用方法](#-使用方法)。 🔥🔥🔥🆕
|
||||
* 🚀 Dec 09, 2025: LoRA 微调脚本已发布，欢迎使用！ 🔥🔥🔥🆕
|
||||
* 🚀 Dec 05, 2025: **新模型发布**：我们现已发布 [480p I2V 步数蒸馏模型](https://huggingface.co/tencent/HunyuanVideo-1.5/tree/main/transformer/480p_i2v_step_distilled)，建议使用 8 或 12 步生成视频！在 RTX 4090 上，端到端生成耗时减少 75%，单卡 RTX 4090 可在 **75 秒**内生成视频。步数蒸馏模型在保持与原模型相当质量的同时实现了显著的加速。详细的质量对比请参见[步数蒸馏对比文档](./assets/step_distillation_comparison.md)。如需更快的生成速度，您也可以尝试使用4步推理（速度更快，质量略有下降）。**启用步数蒸馏模型，请运行 `generate.py` 并使用 `--enable_step_distill` 参数。** 详细的使用说明请参见[使用方法](#-使用方法)。 🔥🔥🔥🆕
|
||||
* 📚 Dec 05, 2025: **训练代码已发布**：我们现已开源 HunyuanVideo-1.5 的完整训练代码！训练脚本（`train.py`）提供了完整的训练流程，支持分布式训练、FSDP、context parallel、梯度检查点等功能。HunyuanVideo-1.5 使用 Muon 优化器进行训练，我们在[训练](#-训练)部分已开源。**如果您希望继续训练我们的模型，或使用 LoRA 进行微调，请使用 Muon 优化器。** 详细使用说明请参见[训练](#-训练)部分。 🔥🔥🔥🆕
|
||||
* 🎉 **Diffusers 支持**：HunyuanVideo-1.5 现已支持 Hugging Face Diffusers！查看我们的 [Diffusers 集合](https://huggingface.co/collections/hunyuanvideo-community/hunyuanvideo-15) 以便轻松集成。 🔥🔥🔥🆕
|
||||
* 🚀 Nov 27, 2025: 我们现已支持 cache 推理（deepcache, teacache, taylorcache），可极大加速推理！请 pull 最新代码体验。 🔥🔥🔥🆕
|
||||
@ -481,6 +482,11 @@ torchrun --nproc_per_node=8 train.py \
|
||||
| `--i2v_prob` | 视频数据使用 i2v 任务的概率 | 0.3 |
|
||||
| `--use_muon` | 使用 Muon 优化器 | true |
|
||||
| `--resume_from_checkpoint` | 从检查点目录恢复训练 | None |
|
||||
| `--use_lora` | 启用 LoRA 微调 | false |
|
||||
| `--lora_r` | LoRA rank | 8 |
|
||||
| `--lora_alpha` | LoRA alpha 缩放参数 | 16 |
|
||||
| `--lora_dropout` | LoRA dropout 率 | 0.0 |
|
||||
| `--pretrained_lora_path` | 预训练 LoRA 适配器路径 | None |
|
||||
|
||||
#### 4. 监控训练
|
||||
|
||||
@ -497,6 +503,28 @@ python train.py \
|
||||
--resume_from_checkpoint ./outputs/checkpoint-1000
|
||||
```
|
||||
|
||||
#### 6. LoRA 微调
|
||||
|
||||
启用 LoRA 微调，在训练命令中添加 `--use_lora`。LoRA 适配器将保存在检查点目录的 `lora/` 子目录下：
|
||||
|
||||
```bash
|
||||
torchrun --nproc_per_node=8 train.py \
|
||||
--pretrained_model_root ./ckpts \
|
||||
--use_lora \
|
||||
--lora_r 8 \
|
||||
--lora_alpha 16 \
|
||||
--learning_rate 1e-4 \
|
||||
--output_dir ./outputs
|
||||
```
|
||||
|
||||
加载预训练的 LoRA 适配器，使用 `--pretrained_lora_path`：
|
||||
```bash
|
||||
torchrun --nproc_per_node=8 train.py \
|
||||
--pretrained_model_root ./ckpts \
|
||||
--use_lora \
|
||||
--pretrained_lora_path ./outputs/checkpoint-1000/lora/default
|
||||
```
|
||||
|
||||
|
||||
## 📊 性能评估
|
||||
### 评分
|
||||
|
||||
Reference in New Issue
Block a user