[BUG] LTX-Video 推理中, FP8Linear.forward() 参数不匹配

Traceback (most recent call last) Cell In[1], line 8 5 WIDTH = 1280 6 NUM_FRAMES = 81 --> 8 infer( 9 InferenceConfig( 10 pipeline_config="configs/ltxv-13b-0.9.8-distilled-fp8.yaml", 11 prompt=PROMPT, 12 height=HEIGHT, 13 width=WIDTH, 14 num_frames=NUM_FRAMES, 15 output_path="/workspace/output.mp4", 16 ) 17 )

File /workspace/LTX-Video/ltx_video/inference.py:569, in infer(config) 560 sample = { 561 "prompt": config.prompt, 562 "prompt_attention_mask": None, 563 "negative_prompt": config.negative_prompt, 564 "negative_prompt_attention_mask": None, 565 } 567 generator = torch.Generator(device=device).manual_seed(config.seed) --> 569 images = pipeline( 570 **pipeline_config, 571 skip_layer_strategy=skip_layer_strategy, 572 generator=generator, 573 output_type="pt", 574 callback_on_step_end=None, 575 height=height_padded, 576 width=width_padded, 577 num_frames=num_frames_padded, 578 frame_rate=config.frame_rate, 579 **sample, 580 media_items=media_item, 581 conditioning_items=conditioning_items, 582 is_video=True, 583 vae_per_channel_normalize=True, 584 image_cond_noise_scale=config.image_cond_noise_scale, 585 mixed_precision=(precision == "mixed_precision"), 586 offload_to_cpu=offload_to_cpu, 587 device=device, 588 enhance_prompt=enhance_prompt, 589 ).images