update app.py

app.py

@@ -1,64 +1,129 @@
+import csv
+import os
+from datetime import datetime
+from typing import Optional
+
 import gradio as gr
-from huggingface_hub import InferenceClient
+from huggingface_hub import HfApi, Repository
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+from optimum_neuron_export import convert
+from gradio_huggingfacehub_search import HuggingfaceHubSearch
+from apscheduler.schedulers.background import BackgroundScheduler
+
+DATASET_REPO_URL = "https://huggingface.co/datasets/optimum/neuron-exports"
+DATA_FILENAME = "exports.csv"
+DATA_FILE = os.path.join("data", DATA_FILENAME)
+
+HF_TOKEN = os.environ.get("HF_WRITE_TOKEN")
 
+DATADIR = "neuron_exports_data"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+repo: Optional[Repository] = None
+# Uncomment if you want to push to dataset repo with token
+# if HF_TOKEN:
+#     repo = Repository(local_dir=DATADIR, clone_from=DATASET_REPO_URL, token=HF_TOKEN)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
+def neuron_export(model_id: str, task: str) -> str:
+    if not model_id:
+        return f"### Invalid input 🐞 Please specify a model name, got {model_id}"
 
-    response = ""
+    try:
+        api = HfApi(token=HF_TOKEN)  # Use HF_TOKEN if available, else anonymous
+        token = HF_TOKEN  # Pass token to convert only if available
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        error, commit_info = convert(api=api, model_id=model_id, task=task, token=token)
+        if error != "0":
+            return error
 
-        response += token
-        yield response
+        print("[commit_info]", commit_info)
 
+        # Save in a private dataset if repo initialized
+        if repo is not None:
+            repo.git_pull(rebase=True)
+            with open(os.path.join(DATADIR, DATA_FILE), "a") as csvfile:
+                writer = csv.DictWriter(
+                    csvfile, fieldnames=["model_id", "pr_url", "time"]
+                )
+                writer.writerow(
+                    {
+                        "model_id": model_id,
+                        "pr_url": commit_info.pr_url,
+                        "time": str(datetime.now()),
+                    }
+                )
+            commit_url = repo.push_to_hub()
+            print("[dataset]", commit_url)
 
+        pr_revision = commit_info.pr_revision.replace("/", "%2F")
+        return f"#### Success 🔥 This model was successfully exported and a PR was opened: [{commit_info.pr_url}]({commit_info.pr_url}). To use the model before the PR is approved, go to https://huggingface.co/{model_id}/tree/{pr_revision}"
+
+    except Exception as e:
+        return f"#### Error: {e}"
+
+
+TITLE_IMAGE = """
+<div style="display: block; margin-left: auto; margin-right: auto; width: 50%;">
+<img src="https://huggingface.co/spaces/optimum/neuron-export/resolve/main/huggingfaceXneuron.png"/>
+</div>
 """
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
+
+TITLE = """
+<div style="display: inline-flex; align-items: center; text-align: center; max-width: 1400px; gap: 0.8rem; font-size: 2.2rem;">
+<h1 style="font-weight: 900; margin-bottom: 10px; margin-top: 10px;">
+    🤗 Optimum Neuron Model Exporter
+</h1>
+</div>
 """
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+
+DESCRIPTION = """
+Export 🤗 Transformers models hosted on the Hugging Face Hub to AWS Neuron-optimized format for Inferentia/Trainium acceleration.
+
+*Features:*
+- Automatically opens PR with Neuron-optimized model
+- Preserves original model weights
+- Adds proper tags to model card
+
+*Note:*
+- PR creation requires the Space owner to have a valid write token set via HF_WRITE_TOKEN
+"""
+
+with gr.Blocks() as demo:
+    gr.HTML(TITLE_IMAGE)
+    gr.HTML(TITLE)
+
+    with gr.Row():
+        with gr.Column(scale=50):
+            gr.Markdown(DESCRIPTION)
+
+        with gr.Column(scale=50):
+            input_model = HuggingfaceHubSearch(
+                label="Hub model ID",
+                placeholder="Search for model ID on the hub",
+                search_type="model",
+            )
+            input_task = gr.Textbox(
+                value="auto",
+                max_lines=1,
+                label='Task (can be left to "auto", will be automatically inferred)',
+            )
+            btn = gr.Button("Export to Neuron")
+            output = gr.Markdown(label="Output")
+
+    btn.click(
+        fn=neuron_export,
+        inputs=[input_model, input_task],
+        outputs=output,
+    )
 
 
 if __name__ == "__main__":
+    def restart_space():
+        if HF_TOKEN:
+            HfApi().restart_space(repo_id="optimum/neuron-export", token=HF_TOKEN, factory_reboot=True)
+
+    scheduler = BackgroundScheduler()
+    scheduler.add_job(restart_space, "interval", seconds=21600)
+    scheduler.start()
+
     demo.launch()