Killed进程被杀死

我是5090显卡，(/root/autodl-tmp/conda_envs/turbodiffusion) root@autodl-container-85ec459d18-73cd4757:~/autodl-tmp/TurboDiffusion# time python turbodiffusion/inference/wan2.2_i2v_infer.py
--model Wan2.2-A14B
--low_noise_model_path checkpoints/TurboWan2.2-I2V-A14B-low-720P-quant.pth
--high_noise_model_path checkpoints/TurboWan2.2-I2V-A14B-high-720P-quant.pth
--resolution 480p
--adaptive_resolution
--image_path /root/autodl-tmp/data/zheng/image-8.jpg
--prompt "镜头不变，有人出门拿起快递"
--num_samples 1
--num_steps 4
--num_frames 81
--quant_linear
--attention_type sagesla
--sla_topk 0.1
--save_path /root/autodl-tmp/stable-diffusion-webui-forge/outputs/1.mp4\
--ode
Megatron-core is not installed.
[01-04 11:38:12|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:85:] Computing embedding for prompt: 镜头不变，有人出门拿起快递
[01-04 11:38:12|INFO|turbodiffusion/rcm/utils/umt5.py:495:init] loading checkpoints/models_t5_umt5-xxl-enc-bf16.pth
'(ReadTimeoutError("HTTPSConnectionPool(host='huggingface.co', port=443): Read timed out. (read timeout=10)"), '(Request ID: 20555700-1487-465b-bf39-8b9b0875676b)')' thrown while requesting HEAD https://huggingface.co/google/umt5-xxl/resolve/main/tokenizer_config.json
Retrying in 1s [Retry 1/5].
'(ReadTimeoutError("HTTPSConnectionPool(host='huggingface.co', port=443): Read timed out. (read timeout=10)"), '(Request ID: 86841327-7734-43c1-b34d-a99d1b879d9d)')' thrown while requesting HEAD https://huggingface.co/google/umt5-xxl/resolve/main/tokenizer_config.json
Retrying in 2s [Retry 2/5].
'(ReadTimeoutError("HTTPSConnectionPool(host='huggingface.co', port=443): Read timed out. (read timeout=10)"), '(Request ID: 8daa41b4-ad74-4166-8334-a5c3b2a5873e)')' thrown while requesting HEAD https://huggingface.co/google/umt5-xxl/resolve/main/tokenizer_config.json
Retrying in 4s [Retry 3/5].
[01-04 11:39:07|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:90:] Loading DiT models.
[01-04 11:39:08|INFO|turbodiffusion/rcm/networks/wan2pt2.py:738:enable_selective_checkpoint] Enable selective checkpoint with mm_only, for every 1 blocks. Total blocks: 40
[01-04 11:39:28|INFO|turbodiffusion/rcm/networks/wan2pt2.py:738:enable_selective_checkpoint] Enable selective checkpoint with mm_only, for every 1 blocks. Total blocks: 40
[01-04 11:39:43|SUCCESS|turbodiffusion/inference/wan2.2_i2v_infer.py:95:] Successfully loaded DiT model.
[01-04 11:39:44|INFO|turbodiffusion/rcm/tokenizers/wan2pt1.py:590:_video_vae] loading checkpoints/Wan2.1_VAE.pth
[01-04 11:39:44|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:99:] Loading and preprocessing image from: /root/autodl-tmp/data/zheng/image-8.jpg
[01-04 11:39:44|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:102:] Adaptive resolution mode enabled.
[01-04 11:39:44|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:105:] Target area is based on 480p 16:9 (~399360 pixels).
[01-04 11:39:44|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:119:] Input image aspect ratio: 0.5625. Adaptive resolution set to: 848x480
[01-04 11:39:44|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:129:] Preprocessing image to 848x480...
[01-04 11:39:46|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:155:] Generating with prompt: 镜头不变，有人出门拿起快递
Sampling: 50%|█████████████████████████████████████ | 2/4 [00:06<00:06, 3.33s/it][01-04 11:40:05|INFO|turbodiffusion/inference/wan2.2_i2v_infer.py:197:] Switched to low noise model.
Sampling: 100%|██████████████████████████████████████████████████████████████████████████| 4/4 [00:22<00:00, 5.64s/it]
Killed

real 2m8.970s
user 0m32.614s
sys 0m54.770s
bash: --ode: command not found，在最后的时候一直被killed不知道是什么情况，环境是2.8的torch

Start with turbodiffusion/inference/wan2.2_i2v_infer.py, especially the model-loading and sampling paths referenced in the log. Reproduce the shown command under torch 2.8, checking the trailing command-line continuation and resource behavior after sampling completes. Done means identifying why the process is killed and documenting or validating a specific correction.