Delete app.py

app.py

@@ -1,106 +0,0 @@
-import spaces
-import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
-import torch
-
-class ModelProcessor:
-    def __init__(self, repo_id="HuggingFaceTB/cosmo-1b"):
-        self.device = "cuda:0" if torch.cuda.is_available() else "cpu"
-        self.tokenizer = AutoTokenizer.from_pretrained(repo_id, use_fast=True)
-        self.model = AutoModelForCausalLM.from_pretrained(
-            repo_id, torch_dtype=torch.float16, device_map={"": self.device}, trust_remote_code=True
-        )
-        self.model.eval()
-        self.tokenizer.pad_token = self.tokenizer.eos_token
-
-    @torch.inference_mode()
-    def process_data_and_compute_statistics(self, prompt):
-        tokens = self.tokenizer(
-            prompt, return_tensors="pt", truncation=True, max_length=512
-        ).to(self.model.device)
-        outputs = self.model(tokens["input_ids"])
-        logits = outputs.logits
-        shifted_labels = tokens["input_ids"][..., 1:].contiguous()
-        shifted_logits = logits[..., :-1, :].contiguous()
-        shifted_probs = torch.softmax(shifted_logits, dim=-1)
-        shifted_log_probs = torch.log_softmax(shifted_logits, dim=-1)
-        entropy = -torch.sum(shifted_probs * shifted_log_probs, dim=-1).squeeze()
-        logits_flat = shifted_logits.view(-1, shifted_logits.size(-1))
-        labels_flat = shifted_labels.view(-1)
-        probabilities_flat = torch.softmax(logits_flat, dim=-1)
-        true_class_probabilities = probabilities_flat.gather(
-            1, labels_flat.unsqueeze(1)
-        ).squeeze(1)
-        nll = -torch.log(
-            true_class_probabilities.clamp(min=1e-9)
-        )
-        ranks = (
-            shifted_logits.argsort(dim=-1, descending=True)
-            == shifted_labels.unsqueeze(-1)
-        ).nonzero()[:, -1]
-        if entropy.clamp(max=4).median() < 2.0:
-            return 1
-        return 1 if (ranks.clamp(max=4) * nll.clamp(max=4)).mean() < 5.2 else 0
-
-processor = ModelProcessor()
-
-@spaces.GPU(duration=180)
-def detect(prompt):
-    prediction = processor.process_data_and_compute_statistics(prompt)
-    if prediction == 1:
-        return "<div class='output-text'>The text is likely <b>generated</b> by a language model.</div>"
-    else:
-        return "<div class='output-text'>The text is likely <b>not generated</b> by a language model.</div>"
-
-with gr.Blocks(
-    css="""
-    .gradio-container {
-        max-width: 800px;
-        margin: 0 auto;
-    }
-    .gr-box {
-        box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
-        padding: 20px;
-        border-radius: 4px;
-    }
-    .gr-button {
-        background-color: #007bff;
-        color: white;        
-        padding: 10px 20px;
-        border-radius: 4px;
-    }
-    .gr-button:hover {
-        background-color: #0056b3;
-    }
-    .hyperlinks a {
-        margin-right: 10px;
-    }
-    .output-text {
-        text-align: center;
-        font-size: 24px;
-        font-weight: bold;
-    }
-"""
-) as demo:
-    with gr.Row():
-        with gr.Column(scale=3):
-            gr.Markdown("# ENTELL Model Detection - ChatGPTBots.net")
-        with gr.Column(scale=1):
-            gr.HTML(
-                """
-            """,
-                elem_classes="hyperlinks",
-            )
-    with gr.Row():
-        with gr.Column():
-            prompt = gr.Textbox(
-                lines=8,
-                placeholder="Type your prompt here...",
-                label="Prompt",
-            )
-            submit_btn = gr.Button("Submit", variant="primary")
-            output = gr.HTML()  # Changed to gr.HTML() to support custom HTML
-
-    submit_btn.click(fn=detect, inputs=prompt, outputs=output)
-
-demo.launch()