Spaces:
Runtime error
Runtime error
DSXiangLi
commited on
Commit
•
15fc122
1
Parent(s):
b581f0e
- app.py +99 -99
- requirements.txt +1 -1
app.py
CHANGED
|
@@ -8,84 +8,84 @@ from self.generate import init_instance, generate_instruction
|
|
| 8 |
from self.prompt import self_prompt
|
| 9 |
|
| 10 |
with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as demo:
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
|
| 90 |
gr.Markdown('\n\n')
|
| 91 |
gr.Markdown('--------')
|
|
@@ -123,27 +123,27 @@ with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as de
|
|
| 123 |
"""
|
| 124 |
APE Callback
|
| 125 |
"""
|
| 126 |
-
#
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
#
|
| 130 |
-
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
#
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
#
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
#
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
#
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
#
|
| 146 |
-
|
| 147 |
|
| 148 |
"""
|
| 149 |
SELF Callback
|
|
|
|
| 8 |
from self.prompt import self_prompt
|
| 9 |
|
| 10 |
with gr.Blocks(title="Automatic Prompt Engineer", theme=gr.themes.Glass()) as demo:
|
| 11 |
+
gr.Markdown("# Automatic Prompt Engineer")
|
| 12 |
+
with gr.Row():
|
| 13 |
+
with gr.Column(scale=2):
|
| 14 |
+
gr.Markdown("## 第一步：输入参数")
|
| 15 |
+
with gr.Row():
|
| 16 |
+
openai_key = gr.Textbox(type='password', label='输入 API key')
|
| 17 |
+
with gr.Row():
|
| 18 |
+
n_train = gr.Slider(label="训练样本数", minimum=1, maximum=20, step=1, value=5)
|
| 19 |
+
n_few_shot = gr.Slider(label="每组几个样例", minimum=1, maximum=20, step=1, value=5)
|
| 20 |
+
|
| 21 |
+
with gr.Row():
|
| 22 |
+
n_eval = gr.Slider(label="评估样本数", minimum=1, maximum=30, step=5, value=20)
|
| 23 |
+
|
| 24 |
+
with gr.Column(scale=3):
|
| 25 |
+
gr.Markdown("## 第二步：加载数据(选任务或上传数据)")
|
| 26 |
+
with gr.Tab("选择数据"):
|
| 27 |
+
with gr.Row():
|
| 28 |
+
file = gr.File(label='上传txt文件，input[空格]output[换行]')
|
| 29 |
+
with gr.Row():
|
| 30 |
+
task = gr.Dropdown(label="Chosse Existing Task", choices=list(LoadFactory.keys()), value=None)
|
| 31 |
+
with gr.Row():
|
| 32 |
+
instance = gr.State()
|
| 33 |
+
load_button = gr.Button("Load Task")
|
| 34 |
+
load_flag = gr.Textbox()
|
| 35 |
+
sample_button = gr.Button('sample Data')
|
| 36 |
+
sample_flag = gr.Textbox()
|
| 37 |
+
|
| 38 |
+
with gr.Tab("展示数据"):
|
| 39 |
+
with gr.Row():
|
| 40 |
+
train_str = gr.Textbox(max_lines=100, lines=10, label="Data for prompt generation")
|
| 41 |
+
eval_str = gr.Textbox(max_lines=100, lines=10, label="Data for scoring")
|
| 42 |
+
|
| 43 |
+
with gr.Row():
|
| 44 |
+
with gr.Column(scale=2):
|
| 45 |
+
gr.Markdown("## 第三步: Run APE(可替换默认指令)")
|
| 46 |
+
gen_prompt = gr.Textbox(max_lines=100, lines=3, interative=True,
|
| 47 |
+
placeholder=MyTemplate['gen_user_prompt'],
|
| 48 |
+
value='', label="Prompt for generation")
|
| 49 |
+
eval_prompt = gr.Textbox(max_lines=100, lines=3, interative=True,
|
| 50 |
+
placeholder=MyTemplate['eval_prompt'],
|
| 51 |
+
value='', label="Prompt for Evaluation")
|
| 52 |
+
test_prompt = gr.Textbox(max_lines=100, lines=3, interative=True,
|
| 53 |
+
placeholder=MyTemplate['test_prompt'],
|
| 54 |
+
value='', label="Prompt for Single Test")
|
| 55 |
+
|
| 56 |
+
with gr.Row():
|
| 57 |
+
cost = gr.Textbox(lines=1, value="", label="Estimated Cost ($)")
|
| 58 |
+
cost_button = gr.Button("Estimate Cost")
|
| 59 |
+
with gr.Row():
|
| 60 |
+
gen_button = gr.Button("Generate")
|
| 61 |
+
eval_button = gr.Button("Eval")
|
| 62 |
+
|
| 63 |
+
with gr.Column(scale=3):
|
| 64 |
+
gr.Markdown("## 第四步：APE 结果")
|
| 65 |
+
with gr.Tab("生成指令"):
|
| 66 |
+
all_prompt = gr.Textbox(label='Generated Prompt')
|
| 67 |
+
# Display all generated prompt with log probs
|
| 68 |
+
output_df = gr.DataFrame(type='pandas', headers=['Prompt', 'Likelihood'], wrap=True, interactive=False)
|
| 69 |
+
|
| 70 |
+
with gr.Tab("指令单测"):
|
| 71 |
+
# Test the output of LLM using prompt
|
| 72 |
+
with gr.Row():
|
| 73 |
+
with gr.Column(scale=1):
|
| 74 |
+
test_instruction = gr.Textbox(lines=4, value="", label="Prompt to test")
|
| 75 |
+
test_input = gr.Textbox(lines=4, value="", label="Inputs used to test prompt[多个输入以换行分割]")
|
| 76 |
+
test_button = gr.Button("Test")
|
| 77 |
+
with gr.Column(scale=1):
|
| 78 |
+
test_output = gr.Textbox(lines=9, value="", label="Model Output")
|
| 79 |
+
|
| 80 |
+
with gr.Tab("指令评估"):
|
| 81 |
+
# By Default use the Evaluation Set in APE
|
| 82 |
+
with gr.Row():
|
| 83 |
+
with gr.Column(scale=1):
|
| 84 |
+
score_instruction = gr.Textbox(lines=3, value="",
|
| 85 |
+
label="Prompt to Evaluate")
|
| 86 |
+
score_button = gr.Button("Evaluate")
|
| 87 |
+
with gr.Column(scale=1):
|
| 88 |
+
test_score = gr.Textbox(lines=1, value="", label="Log(p)", disabled=True)
|
| 89 |
|
| 90 |
gr.Markdown('\n\n')
|
| 91 |
gr.Markdown('--------')
|
|
|
|
| 123 |
"""
|
| 124 |
APE Callback
|
| 125 |
"""
|
| 126 |
+
# 1. 选择已有任务/上传文件，实例化Instance
|
| 127 |
+
load_button.click(load_task, [task, file], [instance, load_flag])
|
| 128 |
+
|
| 129 |
+
# 2. 按 Configuration Sample数据 得到训练样本和验证集, 并在前端展示。支持重采样
|
| 130 |
+
sample_button.click(sample_data, [instance, n_train, n_few_shot, n_eval],
|
| 131 |
+
[train_str, eval_str, instance, sample_flag])
|
| 132 |
+
|
| 133 |
+
# 3. Estimate Cost for train + Eval
|
| 134 |
+
cost_button.click(esttimate_cost, [instance], [cost])
|
| 135 |
+
|
| 136 |
+
# 4. Run APE -> 所有指令
|
| 137 |
+
gen_button.click(generate, [gen_prompt, instance, openai_key], [all_prompt])
|
| 138 |
+
|
| 139 |
+
# 5. Evaluate -> 得到所有指令的Log Prob
|
| 140 |
+
eval_button.click(evaluate, [eval_prompt, all_prompt, instance, openai_key], [output_df])
|
| 141 |
+
|
| 142 |
+
# 6. 输入指令单测
|
| 143 |
+
test_button.click(single_test, [test_prompt, test_instruction, test_input, openai_key], [test_output])
|
| 144 |
+
|
| 145 |
+
# 7. 输入指令打分
|
| 146 |
+
score_button.click(score_single, [eval_prompt, instance, score_instruction, openai_key], [test_score])
|
| 147 |
|
| 148 |
"""
|
| 149 |
SELF Callback
|
requirements.txt
CHANGED
|
@@ -3,4 +3,4 @@ gradio==3.21.0
|
|
| 3 |
langchain==0.0.113
|
| 4 |
tiktoken==0.1.2
|
| 5 |
sentence-transformers==2.2.2
|
| 6 |
-
|
|
|
|
| 3 |
langchain==0.0.113
|
| 4 |
tiktoken==0.1.2
|
| 5 |
sentence-transformers==2.2.2
|
| 6 |
+
rouge_score
|