Merge pull request #6 from Md-Emon-Hasan/master

fix ui

Author: Md-Emon-Hasan

app.png

@@ -26,7 +26,6 @@
 model.eval()
 print("Model & Tokenizer loaded.")
 
-# Routes
 @app.route("/")
 def index():
     return render_template("index.html")

app.py

@@ -169,37 +169,25 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "### ***Importing required modules for model training and evaluation***"
+        "### ***Loading the dataset for English to Spanish translation task***"
       ]
     },
     {
       "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "id": "AWfPzkluGKuk"
-      },
-      "outputs": [],
-      "source": [
-        "from transformers import AutoTokenizer\n",
-        "from transformers import AutoModelForSeq2SeqLM\n",
-        "from transformers import Seq2SeqTrainer\n",
-        "from transformers import Seq2SeqTrainingArguments\n",
-        "from transformers import DataCollatorForSeq2Seq\n",
-        "from datasets import load_dataset\n",
-        "from peft import get_peft_model\n",
-        "from peft import LoraConfig\n",
-        "from peft import TaskType\n",
-        "from peft import prepare_model_for_kbit_training\n",
-        "import evaluate\n",
-        "import numpy as np\n",
-        "import torch"
-      ]
-    },
-    {
-      "cell_type": "markdown",
+      "execution_count": 1,
       "metadata": {},
+      "outputs": [
+        {
+          "name": "stderr",
+          "output_type": "stream",
+          "text": [
+            "c:\\Users\\emon1\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+            "  from .autonotebook import tqdm as notebook_tqdm\n"
+          ]
+        }
+      ],
       "source": [
-        "### ***Loading the dataset for English to Spanish translation task***"
+        "from datasets import load_dataset"
       ]
     },
     {
@@ -316,6 +304,15 @@
         "### ***Loading pre-trained model and tokenizer from Hugging Face Model Hub***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from transformers import AutoTokenizer"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -560,6 +557,15 @@
         "### ***Load Pretrained Model***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from transformers import AutoModelForSeq2SeqLM"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -636,6 +642,15 @@
         "### ***Apply PEFT + LoRA Configuration***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from peft import LoraConfig, TaskType"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -647,13 +662,22 @@
         "lora_config = LoraConfig(\n",
         "    r=8,\n",
         "    lora_alpha=32,\n",
-        "    target_modules=[\"q_proj\", \"v_proj\"],  # Adjust for model type\n",
+        "    target_modules=[\"q_proj\", \"v_proj\"],\n",
         "    lora_dropout=0.1,\n",
         "    bias=\"none\",\n",
         "    task_type=TaskType.SEQ_2_SEQ_LM\n",
         ")"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from peft import get_peft_model"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": 8,
@@ -685,6 +709,16 @@
         "### ***Training Arguments***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from transformers import Seq2SeqTrainingArguments\n",
+        "import torch"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -705,7 +739,7 @@
         "    logging_steps=100,\n",
         "    save_strategy=\"epoch\",\n",
         "    predict_with_generate=True,\n",
-        "    fp16=torch.cuda.is_available(),  # Optional: if CUDA available\n",
+        "    fp16=torch.cuda.is_available(),\n",
         ")"
       ]
     },
@@ -716,6 +750,15 @@
         "### ***Data Collator***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from transformers import DataCollatorForSeq2Seq"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -734,6 +777,16 @@
         "### ***Evaluation Metric: BLEU***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "import evaluate\n",
+        "import numpy as np"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -753,7 +806,7 @@
         "    decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)\n",
         "    decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)\n",
         "\n",
-        "    # Important: BLEU expects list of strings for preds and list of list of strings for refs\n",
+        "    # BLEU expects list of strings for preds and list of list of strings for refs\n",
         "    # So wrap references inside list\n",
         "    references = [[ref] for ref in decoded_labels]\n",
         "\n",
@@ -769,6 +822,15 @@
         "### ***Trainer Setup***"
       ]
     },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "from transformers import Seq2SeqTrainer"
+      ]
+    },
     {
       "cell_type": "code",
       "execution_count": null,
@@ -1406,7 +1468,16 @@
       "name": "python3"
     },
     "language_info": {
-      "name": "python"
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
+      "name": "python",
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.11.9"
     },
     "widgets": {
       "application/vnd.jupyter.widget-state+json": {

demo.mp4

@@ -2,11 +2,11 @@ function translateText() {
       const input = document.getElementById("inputText").value.trim();
       const output = document.getElementById("output");
       if (!input) {
-        output.innerHTML = "<span class='text-warning'>⚠️ Please enter some text!</span>";
+        output.innerHTML = "<span class='text-warning'>Please enter some text!</span>";
         return;
       }
 
-      output.innerHTML = "<em>⏳ Translating...</em>";
+      output.innerHTML = "<em>Translating...</em>";
 
       fetch("/translate", {
         method: "POST",
@@ -16,12 +16,12 @@ function translateText() {
         .then((res) => res.json())
         .then((data) => {
           if (data.translation) {
-            output.innerHTML = `<strong>✅ Translation:</strong><br>${data.translation}`;
+            output.innerHTML = `<strong>Translation:</strong><br>${data.translation}`;
           } else {
-            output.innerHTML = `<span class="text-danger">❌ Error: ${data.error}</span>`;
+            output.innerHTML = `<span class="text-danger">Error: ${data.error}</span>`;
           }
         })
         .catch((err) => {
-          output.innerHTML = "<span class='text-danger'>⚠️ Failed to translate.</span>";
+          output.innerHTML = "<span class='text-danger'>Failed to translate.</span>";
         });
     }

notebook/Experiment.ipynb

@@ -2,7 +2,7 @@
 <html lang="en">
 <head>
   <meta charset="UTF-8">
-  <title>🌐 Translatica</title>
+  <title>Translatica</title>
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <!-- Bootstrap CDN -->
   <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.2/dist/css/bootstrap.min.css" rel="stylesheet">
@@ -17,15 +17,15 @@
     <div class="overlay"></div>
 
     <div class="content">
-      <div class="title">🌐 Translatica</div>
+      <div class="title">Translatica</div>
       <div class="subtitle">
-        English ➡️ Spanish Translation | Fine-tuned using LoRA + PEFT
+        English Spanish Translation | Fine-tuned using LoRA + PEFT
       </div>
 
 
       <div class="form-container shadow-lg">
-        <textarea id="inputText" class="form-control mb-3" placeholder="✍️ Enter your text here..."></textarea>
-        <button class="btn" onclick="translateText()">🚀 Translate</button>
+        <textarea id="inputText" class="form-control mb-3" placeholder="Enter your text here..."></textarea>
+        <button class="btn" onclick="translateText()">Translate</button>
         <div id="output" class="mt-4"></div>
       </div>
     </div>

static/js/script.js

@@ -37,7 +37,7 @@
 lora_config = LoraConfig(
     r=8,
     lora_alpha=32,
-    target_modules=["q_proj", "v_proj"],  # Adjust for model type
+    target_modules=["q_proj", "v_proj"],
     lora_dropout=0.1,
     bias="none",
     task_type=TaskType.SEQ_2_SEQ_LM
@@ -59,7 +59,7 @@
     logging_steps=100,
     save_strategy="epoch",
     predict_with_generate=True,
-    fp16=torch.cuda.is_available(),  # Optional: if CUDA available
+    fp16=torch.cuda.is_available(),
 )
 
 # Data Collator
@@ -77,7 +77,7 @@ def compute_metrics(eval_preds):
     decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
     decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
 
-    # Important: BLEU expects list of strings for preds and list of list of strings for refs
+    # BLEU expects list of strings for preds and list of list of strings for refs
     # So wrap references inside list
     references = [[ref] for ref in decoded_labels]