模型名

server.py

@@ -225,6 +225,27 @@ def main_dev(opt):
         # 遍历 data 返回dict
         for item in data:
             print(item) # {'model': '768-v-ema', 'prompt': '一只猫', 'watermark': '0'}
+            '''
+            task: {
+              model: 'SD2',
+              ckpt: 'latest',
+              prompt: '猫猫',
+              number: 1,
+              tid: '06ruxroiuo3u',
+              uid: 1234567890,
+              status: 'waiting',
+              createdAt: 1676660844133,
+              remove: '狗狗',
+              w: 512,
+              h: 512,
+              seed: 0,
+              sampler: 'pndm',
+              prompt_guidance: 0.5,
+              quality_details: 25,
+              image: '',
+              data: null
+            }
+            '''
             # 设置参数
             if 'prompt'    in item: opt.prompt    = item['prompt']    # 描述
             if 'n_samples' in item: opt.n_samples = item['n_samples'] # 列数
@@ -232,9 +253,9 @@ def main_dev(opt):
             if 'scale'     in item: opt.scale     = item['scale']     # 比例
 
             # 如果模型不同，重新加载模型(注意释放内存)
-            if item['model'] != model_name:
+            if item['ckpt'] != model_name:
                 # 获取环境配置
-                model_name = item['model']
+                model_name = item['ckpt']
                 opt.config = f'/data/{model_name}.yaml'
                 opt.ckpt   = f'/data/{model_name}.ckpt'
                 opt.device = 'cuda'
@@ -242,10 +263,9 @@ def main_dev(opt):
                 config = OmegaConf.load(f"{opt.config}")
                 device = torch.device("cuda") if opt.device == "cuda" else torch.device("cpu")
                 # 加载模型(到显存)
-                print(f"load model: {item['model']}..")
-                model_name = item['model']
+                print(f"加载模型到显存: {model_name}..")
                 model = load_model_from_config(config, f"{opt.ckpt}", device)
-                print(f"model_name: {model_name}")
+                print(f"加载到显存完成: {model_name}")
             # 使用指定的模型和配置文件进行推理一组参数
             if opt.plms:
                 sampler = PLMSSampler(model, device=device)