add logger

app.py

@@ -9,17 +9,19 @@ import gradio as gr
 import queue
 import time
 import shutil
+from gradio_log import Log
+import logging
 
 
 # 全局日志
-log_queue = queue.Queue()
-current_logs = []
+log = logging.getLogger("space_convert")
+log.setLevel(logging.INFO)
+log.addHandler(logging.StreamHandler())
+log.addHandler(logging.FileHandler("convert.log"))
 
 def log(msg):
     """追加并打印日志信息"""
-    print(msg)
-    current_logs.append(msg)
-    return "\n".join(current_logs)
+    log.info(msg)
 
 def timeit(func):
     def wrapper(*args, **kwargs):
@@ -216,20 +218,22 @@ tags:
 - text-generation
 - peft{quantization}
 library_name: transformers
-base_model: {base_model_name} {lora_model_name}
+base_model: {base_model_name}
 widget:
 - messages:
     - role: user
       content: What is your favorite condiment?
 license: other
+datasets:
+- {lora_model_name}
 ---
 # Model
 
 {repo_name}
 
 ## Details:
-- base_model: {base_model_name}
-- lora_model: {lora_model_name}
+- base_model: [{base_model_name}](https://huggingface.co/{base_model_name})
+- lora_model: [{lora_model_name}](https://huggingface.co/{lora_model_name})
 - quant_methods: {quant_methods}
 - created_at: {created_at}
 - created_by: [Steven10429/apply_lora_and_quantize](https://github.com/Steven10429/apply_lora_and_quantize)
@@ -258,8 +262,9 @@ def process_model(base_model_name, lora_model_name, repo_name, quant_methods, hf
       6. 最后统一等待所有 Future 完成，再返回日志。
     """
     try:
-        current_logs.clear()
-        if hf_token.startswith("hf_"):
+        if hf_token.strip().lower() == "auto":
+            hf_token = os.getenv("HF_TOKEN")
+        elif hf_token.startswith("hf_"):
             os.environ["HF_TOKEN"] = hf_token
         login(hf_token)
         api = HfApi(token=hf_token)
@@ -281,7 +286,7 @@ def process_model(base_model_name, lora_model_name, repo_name, quant_methods, hf
         
         
         # 量化模型
-        for quant_method in quant_methods[::-1]:
+        for quant_method in quant_methods:
             quantize(output_dir, repo_name, quant_method=quant_method)
             
         create_readme(repo_name, base_model_name, lora_model_name, quant_methods)
@@ -300,12 +305,11 @@ def process_model(base_model_name, lora_model_name, repo_name, quant_methods, hf
         shutil.rmtree(model_path)
         log("Removed model from local")
         
-        return "\n".join(current_logs)
     except Exception as e:
         error_message = f"Error during processing: {e}"
         log(error_message)
         raise e
-        return "\n".join(current_logs)
+
 
 @timeit
 def create_ui():
@@ -331,32 +335,25 @@ def create_ui():
                 repo_name = gr.Textbox(
                     label="Hugging Face Repository Name",
                     placeholder="Enter the repository name to create",
-                    value="Auto"
+                    value="Auto",
+                    type="password"
                 )
                 quant_method = gr.CheckboxGroup(
                     choices=["Q2_K", "Q4_K", "IQ4_NL", "Q5_K_M", "Q6_K", "Q8_0"],
-                    value=["Q2_K", "Q4_K", "IQ4_NL", "Q5_K_M", "Q6_K", "Q8_0"],
+                    value=["Q4_K", "Q8_0"],
                     label="Quantization Method"
                 )
                 hf_token = gr.Textbox(
                     label="Hugging Face Token",
                     placeholder="Enter your Hugging Face Token",
-                    value=os.getenv("HF_TOKEN"),
-                    type="password"
+                    value="Auto"
                 )
                 convert_btn = gr.Button("Start Conversion", variant="primary")
             with gr.Column():
-                output = gr.TextArea(
-                    label="Logs",
-                    placeholder="Processing logs will appear here...",
-                    interactive=False,
-                    autoscroll=True,
-                    lines=20
-                )
+                Log("convert.log", dark=True, xterm_font_size=12)
         convert_btn.click(
             fn=process_model,
             inputs=[base_model, lora_model, repo_name, quant_method, hf_token],
-            outputs=output
         )
     return app