[Question]: 跑AltDiffusion-m18模型推理时报错

[smallhackeryifa]

Description

在AltDiffusion-m18目录下执行generate.py文件报如下错误

**** text2img altdiffusion-m18 LatentDiffusion: Running in eps-prediction mode DiffusionWrapper has 865.91 M params. No module 'xformers'. Proceeding without it. making attention of type 'vanilla' with 512 in_channels Working with z of shape (1, 4, 32, 32) = 4096 dimensions. making attention of type 'vanilla' with 512 in_channels Traceback (most recent call last): File "/root/FlagAI/examples/AltDiffusion-m18/generate.py", line 11, in loader = AutoLoader(task_name="text2img", #contrastive learning File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/auto_model/auto_loader.py", line 216, in init self.model = getattr(LazyImport(self.model_name[0]), File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/base_model.py", line 178, in from_pretrain return load_diffusion_local(yaml_path, only_download_config=only_download_config, kwargs) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/base_model.py", line 166, in load_diffusion_local model = cls(kwargs) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/mm/AltDiffusionM18.py", line 569, in init self.instantiate_cond_stage(cond_stage_config) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/mm/AltDiffusionM18.py", line 645, in instantiate_cond_stage model = instantiate_from_config(config) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/mm/utils.py", line 112, in instantiate_from_config return get_obj_from_str(config["target"])(**config.get("params", dict())) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/mm/modeling_berts_m18.py", line 61, in init self.roberta = AltRobertaModel(config, add_pooling_layer=False) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/flagai/model/mm/modeling_altclip.py", line 909, in init self.post_init() File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/transformers/modeling_utils.py", line 1034, in post_init self.init_weights() File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/transformers/modeling_utils.py", line 1486, in init_weights self.apply(self._init_weights) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/torch/nn/modules/module.py", line 884, in apply module.apply(fn) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/torch/nn/modules/module.py", line 884, in apply module.apply(fn) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/torch/nn/modules/module.py", line 885, in apply fn(self) File "/root/miniconda3/envs/FlagAI/lib/python3.9/site-packages/transformers/modeling_utils.py", line 1149, in _init_weights raise NotImplementedError(f"Make sure _init_weights is implemented for {self.class}") NotImplementedError: Make sure _init_weights is implemented for <class 'flagai.model.mm.modeling_altclip.AltRobertaModel'>

Alternatives

No response

[BAAI-OpenPlatform]

没能复现出这个错，能确认用的最新的代码吗