Fix E402 Module level import not at top of file

.ruff.toml

@@ -0,0 +1,21 @@
+# Assume Python 3.10.
+target-version = "py310"
+# Decrease the maximum line length to 79 characters.
+line-length = 300
+
+# pyflakes, pycodestyle, isort
+# flake8 YTT, pydocstyle D, pylint PLC
+select = ["F", "E", "W", "I001", "YTT", "D", "PLC"]
+# select = ["ALL"]
+
+# E501 Line too long
+# D102 Missing docstring in public method
+# D100 Missing docstring in public module
+# E501 Line too long
+# D103 Missing docstring in public function
+# D101 Missing docstring in public class
+# `multi-line-summary-first-line` (D212)
+# `one-blank-line-before-class` (D203)
+extend-ignore = ["E501", "D100", "D101", "D102", "D103", "D212", "D203"]
+
+exclude = [".venv", "ultrachat-13B-test.py"]

app.py

@@ -1,12 +1,16 @@
-import os 
+import os
+import time
 
+# ruff: noqa: E402
 # os.system("pip install --upgrade torch transformers sentencepiece scipy cpm_kernels accelerate bitsandbytes loguru")
+
 os.system("pip install torch transformers sentencepiece loguru")
 
-logger.debug("load")
-import gradio as gr
+from pathlib import Path
+
 import torch
-from transformers import AutoTokenizer, AutoModel, AutoModelForCausalLM
+from logru import logger
+from transformers import AutoModel, AutoTokenizer
 
 # fix timezone in Linux
 os.environ["TZ"] = "Asia/Shanghai"
@@ -18,7 +22,9 @@ except Exception:
 
 model_name = "THUDM/chatglm2-6b-int4"  # 3.9G
 
-tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b-int4", trust_remote_code=True)
+tokenizer = AutoTokenizer.from_pretrained(
+    "THUDM/chatglm2-6b-int4", trust_remote_code=True
+)
 
 has_cuda = torch.cuda.is_available()
 # has_cuda = False  # force cpu
@@ -42,7 +48,7 @@ logger.debug("done load")
 # tokenizer = AutoTokenizer.from_pretrained("openchat/openchat_v2_w")
 # model = AutoModelForCausalLM.from_pretrained("openchat/openchat_v2_w", load_in_8bit_fp32_cpu_offload=True, load_in_8bit=True)
 
-model_path = model.config._dict['model_name_or_path']
+model_path = model.config._dict["model_name_or_path"]
 logger.debug(f"{model_path=}")
 
 model_size_gb = Path(model_path).stat().st_size / 2**30

ultrachat-13B-test.py

@@ -1,8 +1,9 @@
 import os
+
 os.system("pip install llama-cpp-python")
 
-from gradio import Interface, Textbox, Slider
 import requests
+from gradio import Interface, Slider, Textbox
 from llama_cpp import Llama
 
 url = "https://huggingface.co/TheBloke/UltraLM-13B-GGML/resolve/main/ultralm-13b.ggmlv3.q4_0.bin"
@@ -24,4 +25,4 @@ ASSISTANT:'''
     return output
 
 
-Interface(fn=generate_text, inputs=[Textbox(type="text", lines=10), Slider(minimum=0, maximum=2, step=0.1, value=0.7), Slider(minimum=1, maximum=2048, step=2, value=256)], outputs=Textbox(type="text", lines=20), title="UltraChat 13B Text Generation", description="Enter a prompt to generate text.").launch()
+Interface(fn=generate_text, inputs=[Textbox(type="text", lines=10), Slider(minimum=0, maximum=2, step=0.1, value=0.7), Slider(minimum=1, maximum=2048, step=2, value=256)], outputs=Textbox(type="text", lines=20), title="UltraChat 13B Text Generation", description="Enter a prompt to generate text.").launch()