Create app.py

app.py

@@ -0,0 +1,92 @@
+import gradio as gr
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# Load the models and tokenizers
+tokenizer1 = AutoTokenizer.from_pretrained("JasperV13/Fhamator-30000")
+model1 = AutoModelForCausalLM.from_pretrained("JasperV13/Fhamator-30000")
+
+tokenizer2 = AutoTokenizer.from_pretrained("JasperV13/Fhamator-SFT")
+model2 = AutoModelForCausalLM.from_pretrained("JasperV13/Fhamator-SFT")
+
+def generate_text_fhamator(input_text, max_length, num_return_sequences, no_repeat_ngram_size, top_k, top_p, temperature):
+    input_ids = tokenizer1.encode(input_text, return_tensors='pt')
+    output = model1.generate(
+        input_ids,
+        max_length=max_length,
+        num_return_sequences=num_return_sequences,
+        no_repeat_ngram_size=no_repeat_ngram_size,
+        top_k=top_k,
+        top_p=top_p,
+        temperature=temperature,
+        do_sample=True
+    )
+    generated_texts = [tokenizer1.decode(output[i], skip_special_tokens=True) for i in range(num_return_sequences)]
+    return "\n\n".join(generated_texts)
+
+def generate_text_sft(input_text, max_length, num_return_sequences, no_repeat_ngram_size, top_k, top_p, temperature):
+    inputs = tokenizer2(input_text, return_tensors="pt")
+    output = model2.generate(
+        inputs['input_ids'],
+        max_length=max_length,
+        num_return_sequences=num_return_sequences,
+        no_repeat_ngram_size=no_repeat_ngram_size,
+        top_k=top_k,
+        top_p=top_p,
+        temperature=temperature,
+        do_sample=True
+    )
+    generated_texts = [tokenizer2.decode(output[i], skip_special_tokens=True) for i in range(num_return_sequences)]
+    return "\n\n".join(generated_texts)
+
+with gr.Blocks() as demo:
+    with gr.Row():
+        gr.Markdown("## 🤖✨ Fhamator: The Darija-Speaking Model")
+    
+    with gr.Tab("📖 Explanation"):
+      gr.Markdown("""
+      # 📚 Explanation of the Work Done
+      Welcome to the **Fhamator** application! This app consists of three main tabs, each with a unique purpose:
+    
+      1. **🔍 Explanation**: Provides an overview of what this app does and how it works.
+      2. **🧠 Fhamator Model**: Test the base 'Fhamator-30000' language model and tweak its hyperparameters to see how it performs.
+        3. **🚀 Fhamator-SFT Model**: Experiment with the fine-tuned 'Fhamator-SFT' model, designed for more specific tasks, also with customizable hyperparameters.
+    """)
+    
+    with gr.Tab("Test Fhamator-30000 Model"):
+        with gr.Group():
+            input_text = gr.Textbox(label="Input Prompt", value="المغرب هو دولة معروفة ب")
+            max_length = gr.Slider(50, 200, value=100, step=1, label="Max Length")
+            num_return_sequences = gr.Slider(1, 5, value=3, step=1, label="Number of Sequences")
+            no_repeat_ngram_size = gr.Slider(1, 5, value=2, step=1, label="No Repeat N-Gram Size")
+            top_k = gr.Slider(1, 100, value=50, step=1, label="Top K")
+            top_p = gr.Slider(0.0, 1.0, value=0.95, step=0.01, label="Top P")
+            temperature = gr.Slider(0.1, 1.0, value=0.7, step=0.1, label="Temperature")
+            output_text = gr.Textbox(label="Generated Texts", lines=8)
+            generate_btn = gr.Button("Generate")
+        
+        generate_btn.click(
+            generate_text_fhamator,
+            inputs=[input_text, max_length, num_return_sequences, no_repeat_ngram_size, top_k, top_p, temperature],
+            outputs=output_text
+        )
+    
+    with gr.Tab("Test Fhamator-SFT Model"):
+        with gr.Group():
+            input_text_sft = gr.Textbox(label="Instruction", value="السؤال :س - التاريخ المغربي؟\n\n: الجواب\n")
+            max_length = gr.Slider(50, 200, value=60, step=1, label="Max Length")
+            num_return_sequences = gr.Slider(1, 5, value=5, step=1, label="Number of Sequences")
+            no_repeat_ngram_size = gr.Slider(1, 5, value=2, step=1, label="No Repeat N-Gram Size")
+            top_k_sft = gr.Slider(1, 100, value=50, step=1, label="Top K")
+            top_p_sft = gr.Slider(0.0, 1.0, value=0.95, step=0.01, label="Top P")
+            temperature_sft = gr.Slider(0.1, 1.0, value=0.7, step=0.1, label="Temperature")
+            output_text_sft = gr.Textbox(label="Generated Texts", lines=8)
+            generate_btn_sft = gr.Button("Generate")
+        
+        generate_btn_sft.click(
+            generate_text_sft,
+            inputs=[input_text_sft, max_length, num_return_sequences, no_repeat_ngram_size, top_k_sft, top_p_sft, temperature_sft],
+            outputs=output_text_sft
+        )
+
+demo.launch(debug = True)