modified files to test remote POST

app.py

@@ -12,7 +12,7 @@ os.environ["https_proxy"] = "http://127.0.0.1:7890"
 
 app = Flask(__name__)
 
-path_for_model = "./output/checkpoint-101500"
+path_for_model = "./output/gpt2_openprompt/checkpoint-4500"
 
 args = {
     "model_type": "gpt2",
@@ -20,6 +20,7 @@ args = {
     "length": 80,
     "stop_token": None,
     "temperature": 1.0,
+    "length_penalty": 1.2,
     "repetition_penalty": 1.2,
     "k": 3,
     "p": 0.9,
@@ -94,4 +95,4 @@ def chat():
 
 if __name__ == '__main__':
     load_model_and_components()
-    app.run(host='0.0.0.0', port=10006, debug=False)
+    app.run(host='0.0.0.0', port=10008, debug=False)

app_test.py

@@ -1,18 +1,14 @@
 import requests
 import json
 
-url = 'http://localhost:10006/chat'  # 根据实际的主机和端口进行修改
+url = 'http://localhost:10008/chat'
 
-# 构造请求数据
 data = {
-    'phrase': 'a spiece 和一只狼'  # 根据需要进行修改
+    'phrase': 'a spiece 和一只狼'
 }
 
-# 发送 POST 请求
 response = requests.post(url, json=data)
 
-# 解析响应
 response_data = response.json()
 
-# 打印响应结果
 print(json.dumps(response_data, indent=4))

utils.py

@@ -1,25 +1,7 @@
 import os
-import json
 
-from typing import Dict
-from torch.utils.data import Dataset
-from datasets import Dataset as AdvancedDataset
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
-
-DEFAULT_TRAIN_DATA_NAME = "test_openprompt.json"
-DEFAULT_TEST_DATA_NAME = "train_openprompt.json"
-DEFAULT_DICT_DATA_NAME = "dataset_openprompt.json"
-
-def get_open_prompt_data(path_for_data):
-    with open(os.path.join(path_for_data, DEFAULT_TRAIN_DATA_NAME)) as f:
-        train_data = json.load(f)
-    
-    with open(os.path.join(path_for_data, DEFAULT_TEST_DATA_NAME)) as f:
-        test_data = json.load(f)
-
-    return train_data, test_data
-
 def get_tok_and_model(path_for_model):
     if not os.path.exists(path_for_model):
         raise RuntimeError("no cached model.")
@@ -27,33 +9,4 @@ def get_tok_and_model(path_for_model):
     tok.pad_token_id = 50256
     # default for open-ended generation
     model = AutoModelForCausalLM.from_pretrained(path_for_model)
-    return tok, model
-
-
-class OpenPromptDataset(Dataset):
-    def __init__(self, data) -> None:
-        super().__init__()
-        self.data = data
-        
-    def __len__(self):
-        return len(self.data)
-
-    def __getitem__(self, index):
-        return self.data[index]
-
-def get_dataset(train_data, test_data):
-    train_dataset = OpenPromptDataset(train_data)
-    test_dataset = OpenPromptDataset(test_data)
-    return train_dataset, test_dataset
-
-def get_dict_dataset(path_for_data):
-    with open(os.path.join(path_for_data, DEFAULT_DICT_DATA_NAME)) as f:
-        dict_data = json.load(f)
-    return dict_data
-    
-def get_advance_dataset(dict_data):
-    if not isinstance(dict_data, Dict):
-        raise RuntimeError("dict_data is not a dict.")
-    dataset = AdvancedDataset.from_dict(dict_data)
-    
-    return dataset
+    return tok, model