lightx2v commited on
Commit
c61dd38
·
verified ·
1 Parent(s): db93455

Update README.md

Browse files

docs: update Wan2.2 distilled model inference instructions

Files changed (1) hide show
  1. README.md +14 -11
README.md CHANGED
@@ -170,23 +170,26 @@ Or refer to [Quick Start Documentation](https://lightx2v-zhcn.readthedocs.io/zh-
170
 
171
  4. Select and modify configuration file
172
 
173
- Choose appropriate configuration based on your GPU memory:
174
 
175
- **80GB+ GPUs (A100/H100)**
176
- - I2V: [wan_moe_i2v_distill.json](https://github.com/ModelTC/LightX2V/blob/main/configs/wan22/wan_moe_i2v_distill.json)
177
 
178
- **24GB+ GPUs (RTX 4090)**
179
- - I2V: [wan_moe_i2v_distill_4090.json](https://github.com/ModelTC/LightX2V/blob/main/configs/wan22/wan_moe_i2v_distill_4090.json)
180
 
 
181
 
182
- 5. Run inference (using [I2V]((https://github.com/ModelTC/LightX2V/blob/main/scripts/wan22/run_wan22_moe_i2v_distill.sh)) as example)
183
- ```bash
184
- cd scripts
185
- bash wan22/run_wan22_moe_i2v_distill.sh
186
- ```
187
 
188
- > 📝 **Note**: Update model paths in the script to point to your Wan2.2 model. Also refer to [LightX2V Model Structure Documentation](https://lightx2v-zhcn.readthedocs.io/zh-cn/latest/getting_started/model_structure.html)
189
 
 
 
 
 
 
190
 
191
  #### LightX2V Documentation
192
  - **Quick Start Guide**: [LightX2V Quick Start](https://lightx2v-zhcn.readthedocs.io/zh-cn/latest/getting_started/quickstart.html)
 
170
 
171
  4. Select and modify configuration file
172
 
173
+ For the FP8 I2V weights downloaded above, choose:
174
 
175
+ - [FP8 configuration](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan22/wan_moe_i2v_distill_quant.json)
176
+ - [RTX 4090 FP8 configuration with CPU offload](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan22/wan_moe_i2v_distill_4090.json)
177
 
178
+ Set `high_noise_quantized_ckpt`, `low_noise_quantized_ckpt`, and `t5_quantized_ckpt` in the selected JSON to your local checkpoint paths. Use an attention backend supported by your GPU and installed environment.
 
179
 
180
+ For BF16 distilled weights, use the [BF16 configuration](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan22/wan_moe_i2v_distill_model.json) instead, and update `high_noise_original_ckpt` and `low_noise_original_ckpt`.
181
 
182
+ 5. Run inference
183
+
184
+ Edit the [I2V launch script](https://github.com/ModelTC/LightX2V/blob/main/scripts/wan22/inference/distill/run_wan22_moe_i2v_distill.sh): set `lightx2v_path`, `model_path`, `CUDA_VISIBLE_DEVICES`, and the input/output paths. Set `--config_json` to the configuration selected above.
 
 
185
 
186
+ Keep `model_path` pointed at the complete Wan2.2-I2V-A14B model directory containing the required encoders, tokenizer, and VAE.
187
 
188
+ Run from the LightX2V repository root:
189
+
190
+ ```bash
191
+ bash scripts/wan22/inference/distill/run_wan22_moe_i2v_distill.sh
192
+ ```
193
 
194
  #### LightX2V Documentation
195
  - **Quick Start Guide**: [LightX2V Quick Start](https://lightx2v-zhcn.readthedocs.io/zh-cn/latest/getting_started/quickstart.html)