File size: 14,472 Bytes
5446331 76e2397 5446331 7a4858f 5446331 76e2397 5446331 76e2397 5446331 76e2397 5446331 76e2397 5446331 76e2397 5446331 7a4858f 76e2397 5446331 76e2397 5446331 76e2397 5446331 76e2397 686461a 76e2397 686461a 76e2397 686461a 76e2397 5446331 76e2397 5446331 76e2397 5446331 |
1 2 3 4 5 6 7 8 9 10 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 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 |
import gradio
import LlamaManager
import os
import huggingface_hub
import random
import ast
HF_API = huggingface_hub.HfApi()
LLAMAMANAGER = LlamaManager.LlamaManager(os.environ.get("HF_KEY_2"), True)
def store_generated_data(data):
token = os.environ.get("HF_BOT")
data = f"{data}"
HF_API.comment_discussion("xqt/SyntheticMBPP2", 1, data, repo_type = "dataset", token = token)
def authenticate(secret_textbox):
global LLAMAMANAGER
password_list = os.environ.get("PASSWORD_LIST")
password_list = password_list.split(":")
api_key = ""
if secret_textbox in password_list:
api_key = os.environ.get("HF_KEY")
else:
api_key = secret_textbox
LLAMAMANAGER = LlamaManager.LlamaManager(api_key, True)
def generate_categories(categories_count, seed, temperature, top_p, frequency_penalty):
categories = LLAMAMANAGER.auto_generate_questions_categories(
count = categories_count,
seed = seed,
temperature = temperature,
top_p = top_p,
frequency_penalty = frequency_penalty
)
data = {
"type": "generate_categories",
"categories": categories,
"count": categories_count,
"seed": seed,
"temperature": temperature,
"top_p": top_p,
"frequency_penalty": frequency_penalty
}
store_generated_data(data)
return gradio.Dropdown(choices = categories, value = random.choice(categories), label = "Select Category", interactive = True)
def generate_shots(category, shots_count, seed, temperature, top_p, frequency_penalty):
shots = LLAMAMANAGER.auto_generate_shots_for_category(category, shots_count, seed, temperature, top_p, frequency_penalty)
shots = [[shot] for shot in shots]
data = {
"type": "generate_shots",
"category": category,
"shots": shots,
"count": shots_count,
"seed": seed,
"temperature": temperature,
"top_p": top_p,
"frequency_penalty": frequency_penalty
}
store_generated_data(data)
return gradio.DataFrame(value = shots, type = "array", label = "Generated Shots", interactive = False, headers = ["Shots"])
def generate_questions(questions_count, category, shots, seed, temperature, top_p, frequency_penalty):
questions = LLAMAMANAGER.auto_generate_questions_from_shots(questions_count, category, shots, seed, temperature, top_p, frequency_penalty)
questions_for_dataframe = [[question] for question in questions]
data = {
"type": "generate_questions",
"questions": questions_for_dataframe,
"count": questions_count,
"category": category,
"shots": shots,
"seed": seed,
"temperature": temperature,
"top_p": top_p,
"frequency_penalty": frequency_penalty
}
store_generated_data(data)
return gradio.DataFrame(value = questions_for_dataframe, type = "array", label = "Generated Shots", interactive = False, headers = ["Questions"]), \
gradio.Dropdown(choices = questions, value = random.choice(questions), label = "Select a Question", interactive = True)
def generate_function(question, temperature, top_p, frequency_penalty, seed):
function_name, function_parameters, function_return = LLAMAMANAGER.auto_generate_function_signature_from_question(
question, seed, temperature, top_p, frequency_penalty
)
data = {
"type": "generate_function",
"question": question,
"function_name": function_name,
"function_parameters": function_parameters,
"function_return": function_return,
"temperature": temperature,
"top_p": top_p,
"frequency_penalty": frequency_penalty,
"seed": seed
}
store_generated_data(data)
return function_name, function_parameters, function_return
def generate_answers_and_tests(question, function_name, function_parameters, function_return, temperature, top_p, frequency_penalty, seed):
function_parameters = ast.literal_eval(function_parameters)
code, tests = LLAMAMANAGER.auto_generate_answers_and_tests(
question, function_name, function_parameters, function_return, seed, temperature, top_p, frequency_penalty
)
data = {
"type": "generate_answers_and_test",
"question": question,
"function_name": function_name,
"function_parameters": function_parameters,
"function_return": function_return,
"code": code,
"tests": tests,
"temperature": temperature,
"top_p": top_p,
"frequency_penalty": frequency_penalty,
"seed": seed
}
store_generated_data(data)
for test in tests:
code += f"\n{test}"
return gradio.Markdown(f"\n```python\n{code}\n```", show_copy_button = True)
with gradio.Blocks(fill_height=True) as base_app:
gradio.Markdown("# Synthetic Python Programming Data Generation ⚙️")
gradio.Markdown("# ❗️ Note: The data generated here by Llama3 and the settings used to generate it will be stored in the repository [here](https://huggingface.co/datasets/xqt/SyntheticMBPP2) for future use.")
gradio.Markdown("# ❗️ Each successful interaction is saved [here](https://huggingface.co/datasets/xqt/SyntheticMBPP2/discussions/1).")
gradio.Markdown("# ❗️ Feel free to use your own API key if the key here is rate limited. API Key is never stored in the repository.")
gradio.Markdown("# ❗️ If you want to use a passcode, please text me.")
gradio.Markdown("# Step 0: Use your own API Key/Passcode")
with gradio.Row():
with gradio.Column():
__secret_textbox = gradio.Textbox(label = "API Key/Passcode", placeholder = "Enter your API Key/Passcode here", type = "password", interactive = True)
with gradio.Column():
__passcode_authenticate = gradio.Button("Authenticate", scale = 2)
gradio.Markdown("# Step 1: How many categories do you want to generate?")
with gradio.Row(equal_height = True):
with gradio.Column(scale = 2):
__categories_count = gradio.Slider(minimum = 1, maximum = 20, step = 1, value = 10, label = "Number of Categories", interactive = True)
with gradio.Column():
__categories_generate = gradio.Button("Generate Categories", scale = 2)
with gradio.Accordion("Advanced Settings", open = False):
with gradio.Row():
with gradio.Column():
__categories_temperature = gradio.Slider(minimum = 0.1, maximum = 2.0, step = 0.01, value = 1.0, label = "Temperature", interactive = True)
__categories_top_p = gradio.Slider(minimum = 0.1, maximum = 0.99, step = 0.01, value = 0.9, label = "Top P", interactive = True)
with gradio.Column():
__categories_frequency_penalty = gradio.Slider(minimum = -2.0, maximum = 2.0, step = 0.01, value = 0.0, label = "Frequency Penalty", interactive = True)
__categories_seed = gradio.Slider(minimum = 0, maximum = 1000, step = 1, value = 123, label = "Seed", interactive = True)
gradio.Markdown("# Step 2: Select a category to generate shots for and select the number of shots to generate")
with gradio.Row():
with gradio.Column(scale = 2):
__shots_category = gradio.Dropdown(choices = [], label = "Select Category", interactive = True)
__shots_count = gradio.Slider(minimum = 2, maximum = 5, step = 1, value = 2, label = "Number of Shots", interactive = True)
with gradio.Column():
__shots_generate = gradio.Button("Generate Shots", scale = 2)
with gradio.Accordion("Advanced Settings", open = False):
with gradio.Row():
with gradio.Column():
__shots_temperature = gradio.Slider(minimum = 0.1, maximum = 2.0, step = 0.01, value = 1.0, label = "Temperature", interactive = True)
__shots_top_p = gradio.Slider(minimum = 0.1, maximum = 0.99, step = 0.01, value = 0.9, label = "Top P", interactive = True)
with gradio.Column():
__shots_frequency_penalty = gradio.Slider(minimum = -2.0, maximum = 2.0, step = 0.01, value = 0.0, label = "Frequency Penalty", interactive = True)
__shots_seed = gradio.Slider(minimum = 0, maximum = 1000, step = 1, value = 123, label = "Seed", interactive = True)
__generated_shots = gradio.DataFrame(value = [], col_count = 1, type = "array", label = "Generated Shots", interactive = False, headers = ["Shots"])
gradio.Markdown("# Step 3: Generate Python Programming Questions for the generated shots")
with gradio.Row():
with gradio.Column(scale = 2):
__questions_count = gradio.Slider(minimum = 1, maximum = 30, step = 1, value = 10, label = "Number of Questions", interactive = True)
with gradio.Column():
__questions_generate = gradio.Button("Generate Questions", scale = 2)
with gradio.Accordion("Advanced Settings", open = False):
with gradio.Row():
with gradio.Column():
__questions_temperature = gradio.Slider(minimum = 0.1, maximum = 2.0, step = 0.01, value = 1.0, label = "Temperature", interactive = True)
__questions_top_p = gradio.Slider(minimum = 0.1, maximum = 0.99, step = 0.01, value = 0.9, label = "Top P", interactive = True)
with gradio.Column():
__questions_frequency_penalty = gradio.Slider(minimum = -2.0, maximum = 2.0, step = 0.01, value = 0.0, label = "Frequency Penalty", interactive = True)
__questions_seed = gradio.Slider(minimum = 0, maximum = 1000, step = 1, value = 123, label = "Seed", interactive = True)
__generated_questions = gradio.DataFrame(value = [], col_count = 1, type = "array", label = "Generated Questions", interactive = False, headers = ["Questions"])
gradio.Markdown("# Step 4: Generate a function name, input parameters, and return type for the generated questions")
with gradio.Row():
with gradio.Column(scale = 2):
__function_question_dropdown = gradio.Dropdown(choices = [], label = "Select a Question", interactive = True, scale = 2)
with gradio.Column():
__function_generate = gradio.Button("Generate Function", scale = 2)
with gradio.Accordion("Advanced Settings", open = False):
with gradio.Row():
with gradio.Column():
__function_temperature = gradio.Slider(minimum = 0.1, maximum = 2.0, step = 0.01, value = 1.0, label = "Temperature", interactive = True)
__function_top_p = gradio.Slider(minimum = 0.1, maximum = 0.99, step = 0.01, value = 0.9, label = "Top P", interactive = True)
with gradio.Column():
__function_frequency_penalty = gradio.Slider(minimum = -2.0, maximum = 2.0, step = 0.01, value = 0.0, label = "Frequency Penalty", interactive = True)
__function_seed = gradio.Slider(minimum = 0, maximum = 1000, step = 1, value = 123, label = "Seed", interactive = True)
with gradio.Row():
with gradio.Column():
__function_name = gradio.Textbox(label = "Function Name", placeholder = "dummy_foo", interactive = False)
with gradio.Column():
__function_parameters = gradio.Textbox(label = "Input Parameters", placeholder = "['input_dict: dict, 'a': int]", interactive = False)
with gradio.Column():
__function_return = gradio.Textbox(label = "Return Type", placeholder = "str", interactive = False)
gradio.Markdown("# 🚀 Step 5: Generate a code.")
__code_generate = gradio.Button("Generate Code", scale = 2)
with gradio.Accordion("Advanced Settings", open = False):
with gradio.Row():
with gradio.Column():
__code_temperature = gradio.Slider(minimum = 0.1, maximum = 2.0, step = 0.01, value = 1.0, label = "Temperature", interactive = True)
__code_top_p = gradio.Slider(minimum = 0.1, maximum = 0.99, step = 0.01, value = 0.9, label = "Top P", interactive = True)
with gradio.Column():
__code_frequency_penalty = gradio.Slider(minimum = -2.0, maximum = 2.0, step = 0.01, value = 0.0, label = "Frequency Penalty", interactive = True)
__code_seed = gradio.Slider(minimum = 0, maximum = 1000, step = 1, value = 123, label = "Seed", interactive = True)
__code = gradio.Markdown("🚀 Code will be generated here...", show_copy_button = True)
__passcode_authenticate.click(authenticate,
inputs = [__secret_textbox],
outputs = []
)
__categories_generate.click(generate_categories,
inputs = [__categories_count, __categories_seed, __categories_temperature, __categories_top_p, __categories_frequency_penalty],
outputs = [__shots_category]
)
__shots_generate.click(generate_shots,
inputs = [__shots_category, __shots_count, __shots_seed, __shots_temperature, __shots_top_p, __shots_frequency_penalty],
outputs = [__generated_shots]
)
__questions_generate.click(generate_questions,
inputs = [__questions_count, __shots_category, __generated_shots, __questions_seed, __questions_temperature, __questions_top_p, __questions_frequency_penalty],
outputs = [__generated_questions, __function_question_dropdown]
)
__function_generate.click(generate_function,
inputs = [__function_question_dropdown, __function_temperature, __function_top_p, __function_frequency_penalty, __function_seed],
outputs = [__function_name, __function_parameters, __function_return]
)
__code_generate.click(generate_answers_and_tests,
inputs = [__function_question_dropdown, __function_name, __function_parameters, __function_return, __code_temperature, __code_top_p, __code_frequency_penalty, __code_seed],
outputs = [__code]
)
if __name__ == "__main__":
base_app.launch() |