finish char-gpt annotation

ch11_llm/char_gpt.ipynb

@@ -786,7 +786,7 @@
     }
    ],
    "source": [
-    "# 使用模型来生成文本\n",
+    "# Test text generation\n",
     "begin_text = torch.tensor(tok.encode('def'), device=device).unsqueeze(0)\n",
     "print(''.join(tok.decode(generate_batch(model, begin_text))))"
    ]
@@ -816,25 +816,25 @@
    "source": [
     "def process(data, sequence_len=sequence_len):\n",
     "    '''\n",
-    "    根据文本生成训练数据\n",
+    "    Transform text to train data.\n",
     "    '''\n",
-    "    # text是字符串列表\n",
+    "    # text is list[str]\n",
     "    text = data['whole_func_string']\n",
     "    inputs, labels = [], []\n",
     "    for i in text:\n",
     "        enc = tok.encode(i)\n",
-    "        # 0对应着文本结束\n",
+    "        # 0 means the end of text\n",
     "        enc += [0]\n",
-    "        # 将文本转换为多个训练数据\n",
+    "        # Transform one text to multiple train data\n",
     "        for i in range(len(enc) - sequence_len):\n",
     "            inputs.append(enc[i: i + sequence_len])\n",
-    "            # 预测标签是下一个字母，因此只需要挪动一个位置即可\n",
+    "            # The label is the next char. It is sufficient to move one position.\n",
     "            labels.append(enc[i + 1: i + 1 + sequence_len])\n",
     "    return {'inputs': inputs, 'labels': labels}\n",
     "\n",
-    "# 将数据分为训练集和测试集\n",
+    "# Split data into train set and test set\n",
     "tokenized = datasets.train_test_split(test_size=0.1, seed=1024, shuffle=True)\n",
-    "# 将文本转换为训练数据，里面包含inputs和labels\n",
+    "# The train data merely contains inputs and labels.\n",
     "tokenized = tokenized.map(process, batched=True, remove_columns=datasets.column_names)\n",
     "tokenized.set_format(type='torch', device=device)\n",
     "\n",
@@ -877,10 +877,10 @@
     }
    ],
    "source": [
-    "# 构建数据读取器\n",
+    "# Build data loader\n",
     "train_loader = DataLoader(tokenized['train'], batch_size=batch_size, shuffle=True)\n",
     "test_loader = DataLoader(tokenized['test'], batch_size=batch_size, shuffle=True)\n",
-    "# 获取一个批量的数据\n",
+    "# Get one batch data\n",
     "next(iter(test_loader))"
    ]
   },
@@ -909,22 +909,22 @@
    "source": [
     "def estimate_loss(model):\n",
     "    re = {}\n",
-    "    # 将模型切换至评估模式\n",
+    "    # Put the model on evaluation mode\n",
     "    model.eval()\n",
     "    re['train'] = _loss(model, train_loader)\n",
     "    re['test'] = _loss(model, test_loader)\n",
-    "    # 将模型切换至训练模式\n",
+    "    # Put the model on train mode\n",
     "    model.train()\n",
     "    return re\n",
     "\n",
     "@torch.no_grad()\n",
     "def _loss(model, data_loader):\n",
     "    '''\n",
-    "    计算模型在不同数据集下面的评估指标\n",
+    "    Measure the performance of model based on different data sets.\n",
     "    '''\n",
     "    loss = []\n",
     "    data_iter= iter(data_loader)\n",
-    "    # 随机使用多个批量数据来预估模型效果\n",
+    "    # Use eval_iters batch data to measure the performance\n",
     "    for k in range(eval_iters):\n",
     "        data = next(data_iter, None)\n",
     "        if data is None:\n",
@@ -932,8 +932,9 @@
     "            data = next(data_iter, None)\n",
     "        inputs, labels = data['inputs'], data['labels']\n",
     "        logits = model(inputs)\n",
-    "        # 根据cross_entropy的定义，需要对logits进行转置运算\n",
-    "        # 具体细节请参考cross_entropy的官方文档\n",
+    "        # According to the definition of cross_entropy in PyTorch,\n",
+    "        # we need to transpose the logits.\n",
+    "        # More details can be found in official document.\n",
     "        logits = logits.transpose(-2, -1)\n",
     "        loss.append(F.cross_entropy(logits, labels).item())\n",
     "    return torch.tensor(loss).mean().item()\n",
@@ -956,14 +957,15 @@
     "            inputs, labels = data['inputs'], data['labels']\n",
     "            optimizer.zero_grad()\n",
     "            logits = model(inputs)\n",
-    "            # 根据cross_entropy的定义，需要对logits进行转置运算\n",
-    "            # 具体细节请参考cross_entropy的官方文档\n",
+    "            # According to the definition of cross_entropy in PyTorch,\n",
+    "            # we need to transpose the logits.\n",
+    "            # More details can be found in official document.\n",
     "            logits = logits.transpose(-2, -1)\n",
     "            loss = F.cross_entropy(logits, labels)\n",
     "            lossi.append(loss.item())\n",
     "            loss.backward()\n",
     "            optimizer.step()\n",
-    "        # 评估模型，并输出结果\n",
+    "        # Evaluate the performance\n",
     "        stats = estimate_loss(model)\n",
     "        train_loss = f'train loss {stats['train']:.4f}'\n",
     "        test_loss = f'test loss {stats['test']:.4f}'\n",
@@ -1066,7 +1068,7 @@
     }
    ],
    "source": [
-    "# 使用模型来生成文本\n",
+    "# Text generation\n",
     "begin_text = torch.tensor(tok.encode('def '), device=device).unsqueeze(0)\n",
     "print(''.join(tok.decode(generate_batch(model, begin_text))))"
    ]