yujunjun/LLMs-from-scratch
1
0
Fork 0
mirror of https://github.com/rasbt/LLMs-from-scratch.git synced 2025-06-26 23:50:03 +00:00
Code Issues Packages Projects Releases Wiki Activity
LLMs-from-scratch/ch05/01_main-chapter-code/exercise-solutions.ipynb
Henry Shi 02779f5e35
updated exercise 5.3 (#615) 
* updated exercise 5.3

temperature can be set to 0 to regardless of top_k setting to force deterministic behavior

* fix notebook json

---------

Co-authored-by: Sebastian Raschka <mail@sebastianraschka.com>
2025-04-13 13:06:57 -05:00

1017 lines
32 KiB
Plaintext
Raw Blame History

{
"cells": [
{
"cell_type": "markdown",
"id": "ba450fb1-8a26-4894-ab7a-5d7bfefe90ce",
"metadata": {},
"source": [
"<table style=\"width:100%\">\n",
"<tr>\n",
"<td style=\"vertical-align:middle; text-align:left;\">\n",
"<font size=\"2\">\n",
"Supplementary code for the <a href=\"http://mng.bz/orYv\">Build a Large Language Model From Scratch</a> book by <a href=\"https://sebastianraschka.com\">Sebastian Raschka</a><br>\n",
"<br>Code repository: <a href=\"https://github.com/rasbt/LLMs-from-scratch\">https://github.com/rasbt/LLMs-from-scratch</a>\n",
"</font>\n",
"</td>\n",
"<td style=\"vertical-align:middle; text-align:left;\">\n",
"<a href=\"http://mng.bz/orYv\"><img src=\"https://sebastianraschka.com/images/LLMs-from-scratch-images/cover-small.webp\" width=\"100px\"></a>\n",
"</td>\n",
"</tr>\n",
"</table>"
]
},
{
"cell_type": "markdown",
"id": "51c9672d-8d0c-470d-ac2d-1271f8ec3f14",
"metadata": {},
"source": [
"# Chapter 5 Exercise solutions"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "37aa4692-2357-4d88-b072-6d2d988d7f4f",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"numpy version: 1.26.4\n",
"tiktoken version: 0.7.0\n",
"torch version: 2.4.0\n",
"tensorflow version: 2.16.1\n"
]
}
],
"source": [
"from importlib.metadata import version\n",
"\n",
"pkgs = [\"numpy\", \n",
" \"tiktoken\", \n",
" \"torch\",\n",
" \"tensorflow\" # For OpenAI's pretrained weights\n",
" ]\n",
"for p in pkgs:\n",
" print(f\"{p} version: {version(p)}\")"
]
},
{
"cell_type": "markdown",
"id": "5fea8be3-30a1-4623-a6d7-b095c6c1092e",
"metadata": {},
"source": [
"# Exercise 5.1: Temperature-scaled softmax scores and sampling probabilities"
]
},
{
"cell_type": "markdown",
"id": "5860ba9f-2db3-4480-b96b-4be1c68981eb",
"metadata": {},
"source": [
"- We can print the number of times the word \"pizza\" is sampled using the `print_sampled_tokens` function we defined in this section\n",
"- Let's start with the code we defined in section 5.3.1\n",
"\n",
"- It is sampled 0x if the temperature is 0 or 0.1, and it is sampled 32x if the temperature is scaled up to 5. The estimated probability is 32/1000 * 100% = 3.2%\n",
"\n",
"- The actual probability is 4.3% and contained in the rescaled softmax probability tensor (`scaled_probas[2][6]`)"
]
},
{
"cell_type": "markdown",
"id": "9cba59c2-a8a3-4af3-add4-70230795225e",
"metadata": {},
"source": [
"- Below is a self-contained example using code from chapter 5:"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "42dda298-3014-4c36-8d63-97c210bcf4e8",
"metadata": {},
"outputs": [],
"source": [
"import torch\n",
"\n",
"vocab = { \n",
" \"closer\": 0,\n",
" \"every\": 1, \n",
" \"effort\": 2, \n",
" \"forward\": 3,\n",
" \"inches\": 4,\n",
" \"moves\": 5, \n",
" \"pizza\": 6,\n",
" \"toward\": 7,\n",
" \"you\": 8,\n",
"} \n",
"inverse_vocab = {v: k for k, v in vocab.items()}\n",
"\n",
"next_token_logits = torch.tensor(\n",
" [4.51, 0.89, -1.90, 6.75, 1.63, -1.62, -1.89, 6.28, 1.79]\n",
")\n",
"\n",
"def print_sampled_tokens(probas):\n",
" torch.manual_seed(123)\n",
" sample = [torch.multinomial(probas, num_samples=1).item() for i in range(1_000)]\n",
" sampled_ids = torch.bincount(torch.tensor(sample))\n",
" for i, freq in enumerate(sampled_ids):\n",
" print(f\"{freq} x {inverse_vocab[i]}\")\n",
"\n",
"\n",
"def softmax_with_temperature(logits, temperature):\n",
" scaled_logits = logits / temperature\n",
" return torch.softmax(scaled_logits, dim=0)\n",
"\n",
"\n",
"temperatures = [1, 0.1, 5] # Original, higher, and lower temperature\n",
"scaled_probas = [softmax_with_temperature(next_token_logits, T) for T in temperatures]"
]
},
{
"cell_type": "markdown",
"id": "1ee0f9f3-4132-42c7-8324-252fd8f59145",
"metadata": {},
"source": [
"- Now, we can iterate over the `scaled_probas` and print the sampling frequencies in each case:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "b5605236-e300-4844-aea7-509d868efbdd",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"\n",
"Temperature: 1\n",
"73 x closer\n",
"0 x every\n",
"0 x effort\n",
"582 x forward\n",
"2 x inches\n",
"0 x moves\n",
"0 x pizza\n",
"343 x toward\n",
"\n",
"\n",
"Temperature: 0.1\n",
"0 x closer\n",
"0 x every\n",
"0 x effort\n",
"985 x forward\n",
"0 x inches\n",
"0 x moves\n",
"0 x pizza\n",
"15 x toward\n",
"\n",
"\n",
"Temperature: 5\n",
"165 x closer\n",
"75 x every\n",
"42 x effort\n",
"239 x forward\n",
"71 x inches\n",
"46 x moves\n",
"32 x pizza\n",
"227 x toward\n",
"103 x you\n"
]
}
],
"source": [
"for i, probas in enumerate(scaled_probas):\n",
" print(\"\\n\\nTemperature:\", temperatures[i])\n",
" print_sampled_tokens(probas)"
]
},
{
"cell_type": "markdown",
"id": "fbf88c97-19c4-462c-924a-411c8c765d2c",
"metadata": {},
"source": [
"- Note that sampling offers an approximation of the actual probabilities when the word \"pizza\" is sampled\n",
"- E.g., if it is sampled 32/1000 times, the estimated probability is 3.2%\n",
"- To obtain the actual probability, we can check the probabilities directly by accessing the corresponding entry in `scaled_probas`\n",
"\n",
"- Since \"pizza\" is the 7th entry in the vocabulary, for the temperature of 5, we obtain it as follows:"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "1d4163c0-22ad-4f5b-8e20-b7420e9dbfc6",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"tensor(0.0430)"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"temp5_idx = 2\n",
"pizza_idx = 6\n",
"\n",
"scaled_probas[temp5_idx][pizza_idx]"
]
},
{
"cell_type": "markdown",
"id": "d3dcb438-5f18-4332-9627-66009f30a1a4",
"metadata": {},
"source": [
"There is a 4.3% probability that the word \"pizza\" is sampled if the temperature is set to 5"
]
},
{
"cell_type": "markdown",
"id": "b510ffb0-adca-4d64-8a12-38c4646fd736",
"metadata": {},
"source": [
"# Exercise 5.2: Different temperature and top-k settings"
]
},
{
"cell_type": "markdown",
"id": "884990db-d1a6-4c4e-8e36-2c1e4c1e67c7",
"metadata": {},
"source": [
"- Both temperature and top-k settings have to be adjusted based on the individual LLM (a kind of trial and error process until it generates desirable outputs)\n",
"- The desirable outcomes are also application-specific, though\n",
" - Lower top-k and temperatures result in less random outcomes, which is desired when creating educational content, technical writing or question answering, data analyses, code generation, and so forth\n",
" - Higher top-k and temperatures result in more diverse and random outputs, which is more desirable for brainstorming tasks, creative writing, and so forth"
]
},
{
"cell_type": "markdown",
"id": "3f35425d-529d-4179-a1c4-63cb8b25b156",
"metadata": {},
"source": [
"# Exercise 5.3: Deterministic behavior in the decoding functions"
]
},
{
"cell_type": "markdown",
"id": "d12229a2-1d52-46ff-b1e8-198f2e58a7d2",
"metadata": {},
"source": [
"There are multiple ways to force deterministic behavior with the `generate` function:\n",
"\n",
"1. Setting to `temperature=0.0`;\n",
"2. Setting `top_k=1`."
]
},
{
"cell_type": "markdown",
"id": "391c5dc8-8dd7-4a0a-90bd-519b72f528c7",
"metadata": {},
"source": [
"Below is a self-contained example using code from chapter 5:"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "a61a4034-797a-4635-bf42-ddfff1b07125",
"metadata": {},
"outputs": [],
"source": [
"import tiktoken\n",
"import torch\n",
"from previous_chapters import GPTModel\n",
"\n",
"\n",
"GPT_CONFIG_124M = {\n",
" \"vocab_size\": 50257, # Vocabulary size\n",
" \"context_length\": 256, # Shortened context length (orig: 1024)\n",
" \"emb_dim\": 768, # Embedding dimension\n",
" \"n_heads\": 12, # Number of attention heads\n",
" \"n_layers\": 12, # Number of layers\n",
" \"drop_rate\": 0.1, # Dropout rate\n",
" \"qkv_bias\": False # Query-key-value bias\n",
"}\n",
"\n",
"\n",
"torch.manual_seed(123)\n",
"\n",
"tokenizer = tiktoken.get_encoding(\"gpt2\")\n",
"model = GPTModel(GPT_CONFIG_124M)\n",
"model.load_state_dict(torch.load(\"model.pth\", weights_only=True))\n",
"model.eval();"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "ee95a272-b852-43b4-9827-ea7e1dbd5724",
"metadata": {},
"outputs": [],
"source": [
"from gpt_generate import generate, text_to_token_ids, token_ids_to_text\n",
"from previous_chapters import generate_text_simple"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "4ab43658-3240-484a-9072-a40a0ed85be6",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Output text:\n",
" Every effort moves you know,\" was one of the axioms he laid down across the Sevres and silver of an exquisitely appointed lun\n"
]
}
],
"source": [
"# Deterministic function that used torch.argmax\n",
"\n",
"start_context = \"Every effort moves you\"\n",
"\n",
"token_ids = generate_text_simple(\n",
" model=model,\n",
" idx=text_to_token_ids(start_context, tokenizer),\n",
" max_new_tokens=25,\n",
" context_size=GPT_CONFIG_124M[\"context_length\"]\n",
")\n",
"\n",
"print(\"Output text:\\n\", token_ids_to_text(token_ids, tokenizer))"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "ebb22d06-393a-42d3-ab64-66646d33b39b",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Output text:\n",
" Every effort moves you know,\" was one of the axioms he laid down across the Sevres and silver of an exquisitely appointed lun\n"
]
}
],
"source": [
"# Deterministic behavior: No top_k, no temperature scaling\n",
"\n",
"token_ids = generate(\n",
" model=model,\n",
" idx=text_to_token_ids(\"Every effort moves you\", tokenizer),\n",
" max_new_tokens=25,\n",
" context_size=GPT_CONFIG_124M[\"context_length\"],\n",
" top_k=None,\n",
" temperature=0.0\n",
")\n",
"\n",
"print(\"Output text:\\n\", token_ids_to_text(token_ids, tokenizer))"
]
},
{
"cell_type": "markdown",
"id": "c85b1f11-37a5-477d-9c2d-170a6865e669",
"metadata": {},
"source": [
"- Note that re-executing the previous code cell will produce the exact same generated text:"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "75469f24-47cc-458d-a200-fe64c648131d",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Output text:\n",
" Every effort moves you know,\" was one of the axioms he laid down across the Sevres and silver of an exquisitely appointed lun\n"
]
}
],
"source": [
"# Deterministic behavior: No top_k, no temperature scaling\n",
"\n",
"token_ids = generate(\n",
" model=model,\n",
" idx=text_to_token_ids(\"Every effort moves you\", tokenizer),\n",
" max_new_tokens=25,\n",
" context_size=GPT_CONFIG_124M[\"context_length\"],\n",
" top_k=None,\n",
" temperature=0.0\n",
")\n",
"\n",
"print(\"Output text:\\n\", token_ids_to_text(token_ids, tokenizer))"
]
},
{
"cell_type": "markdown",
"id": "6d0480e5-fb4e-41f8-a161-7ac980d71d47",
"metadata": {},
"source": [
"# Exercise 5.4: Continued pretraining"
]
},
{
"cell_type": "markdown",
"id": "f40044e8-a0f5-476c-99fd-489b999fd80a",
"metadata": {},
"source": [
"- If we are still in the Python session where you first trained the model in chapter 5, to continue the pretraining for one more epoch, we just have to load the model and optimizer that we saved in the main chapter and call the `train_model_simple` function again\n",
"\n",
"- It takes a couple more steps to make this reproducible in this new code environment\n",
"- First, we load the tokenizer, model, and optimizer:"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "94eae6ba-d9fd-417a-8e31-fc39e9299870",
"metadata": {},
"outputs": [],
"source": [
"import tiktoken\n",
"import torch\n",
"from previous_chapters import GPTModel\n",
"\n",
"\n",
"GPT_CONFIG_124M = {\n",
" \"vocab_size\": 50257, # Vocabulary size\n",
" \"context_length\": 256, # Shortened context length (orig: 1024)\n",
" \"emb_dim\": 768, # Embedding dimension\n",
" \"n_heads\": 12, # Number of attention heads\n",
" \"n_layers\": 12, # Number of layers\n",
" \"drop_rate\": 0.1, # Dropout rate\n",
" \"qkv_bias\": False # Query-key-value bias\n",
"}\n",
"\n",
"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
"\n",
"tokenizer = tiktoken.get_encoding(\"gpt2\")\n",
"\n",
"checkpoint = torch.load(\"model_and_optimizer.pth\", weights_only=True)\n",
"model = GPTModel(GPT_CONFIG_124M)\n",
"model.load_state_dict(checkpoint[\"model_state_dict\"])\n",
"model.to(device)\n",
"\n",
"optimizer = torch.optim.AdamW(model.parameters(), lr=0.0004, weight_decay=0.1)\n",
"optimizer.load_state_dict(checkpoint[\"optimizer_state_dict\"])\n",
"model.train();"
]
},
{
"cell_type": "markdown",
"id": "688fce4a-9ab2-4d97-a95c-fef02c32b4f3",
"metadata": {},
"source": [
"- Next, we initialize the data loader:"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "b5a78470-0652-4abd-875a-664e23c07c36",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import urllib.request\n",
"from previous_chapters import create_dataloader_v1\n",
"\n",
"\n",
"file_path = \"the-verdict.txt\"\n",
"url = \"https://raw.githubusercontent.com/rasbt/LLMs-from-scratch/main/ch02/01_main-chapter-code/the-verdict.txt\"\n",
"\n",
"if not os.path.exists(file_path):\n",
" with urllib.request.urlopen(url) as response:\n",
" text_data = response.read().decode('utf-8')\n",
" with open(file_path, \"w\", encoding=\"utf-8\") as file:\n",
" file.write(text_data)\n",
"else:\n",
" with open(file_path, \"r\", encoding=\"utf-8\") as file:\n",
" text_data = file.read()\n",
"\n",
"\n",
"# Train/validation ratio\n",
"train_ratio = 0.90\n",
"split_idx = int(train_ratio * len(text_data))\n",
"train_data = text_data[:split_idx]\n",
"val_data = text_data[split_idx:]\n",
"\n",
"\n",
"torch.manual_seed(123)\n",
"\n",
"train_loader = create_dataloader_v1(\n",
" train_data,\n",
" batch_size=2,\n",
" max_length=GPT_CONFIG_124M[\"context_length\"],\n",
" stride=GPT_CONFIG_124M[\"context_length\"],\n",
" drop_last=True,\n",
" shuffle=True,\n",
" num_workers=0\n",
")\n",
"\n",
"val_loader = create_dataloader_v1(\n",
" val_data,\n",
" batch_size=2,\n",
" max_length=GPT_CONFIG_124M[\"context_length\"],\n",
" stride=GPT_CONFIG_124M[\"context_length\"],\n",
" drop_last=False,\n",
" shuffle=False,\n",
" num_workers=0\n",
")"
]
},
{
"cell_type": "markdown",
"id": "76598ef8-165c-4bcc-af5e-b6fe72398365",
"metadata": {},
"source": [
"- Lastly, we use the `train_model_simple` function to train the model:"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "ab4693dc-1359-47a7-8110-1e90f514a49e",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Ep 1 (Step 000000): Train loss 0.271, Val loss 6.545\n",
"Ep 1 (Step 000005): Train loss 0.244, Val loss 6.614\n",
"Every effort moves you?\" \"Yes--quite insensible to the irony. She wanted him vindicated--and by me!\" He laughed again, and threw back his head to look up at the sketch of the donkey. \"There were days when I\n"
]
}
],
"source": [
"from gpt_train import train_model_simple\n",
"\n",
"num_epochs = 1\n",
"train_losses, val_losses, tokens_seen = train_model_simple(\n",
" model, train_loader, val_loader, optimizer, device,\n",
" num_epochs=num_epochs, eval_freq=5, eval_iter=5,\n",
" start_context=\"Every effort moves you\", tokenizer=tokenizer\n",
")"
]
},
{
"cell_type": "markdown",
"id": "3384e788-f5a1-407c-8dd1-87959b75026d",
"metadata": {},
"source": [
"# Exercise 5.5: Training and validation set losses of the pretrained model"
]
},
{
"cell_type": "markdown",
"id": "7cb1140b-2027-4156-8d19-600ac849edbe",
"metadata": {},
"source": [
"- We can use the following code to calculate the training and validation set losses of the GPT model:\n",
"\n",
"```python\n",
"train_loss = calc_loss_loader(train_loader, gpt, device)\n",
"val_loss = calc_loss_loader(val_loader, gpt, device)\n",
"```\n",
"\n",
"- The resulting losses for the 124M parameter are as follows:\n",
"\n",
"```\n",
"Training loss: 3.754748503367106\n",
"Validation loss: 3.559617757797241\n",
"```\n",
"\n",
"- The main observation is that the training and validation set performances are in the same ballpark\n",
"- This can have multiple explanations:\n",
"\n",
"1. The Verdict was not part of the pretraining dataset when OpenAI trained GPT-2. Hence, the model is not explicitly overfitting to the training set and performs similarly well on The Verdict's training and validation set portions. (The validation set loss is slightly lower than the training set loss, which is unusual in deep learning. However, it's likely due to random noise since the dataset is relatively small. In practice, if there is no overfitting, the training and validation set performances are expected to be roughly identical).\n",
"\n",
"2. The Verdict was part of GPT -2's training dataset. In this case, we can't tell whether the model is overfitting the training data because the validation set would have been used for training as well. To evaluate the degree of overfitting, we'd need a new dataset generated after OpenAI finished training GPT-2 to make sure that it couldn't have been part of the pretraining."
]
},
{
"cell_type": "markdown",
"id": "66bb4316-a57c-437f-9a01-fe99b1678524",
"metadata": {},
"source": [
"The code below is a reproducible standalone example for this new notebook."
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "68d162d6-bbb9-4d6d-82ee-1c410694f872",
"metadata": {},
"outputs": [],
"source": [
"import tiktoken\n",
"import torch\n",
"from previous_chapters import GPTModel\n",
"\n",
"\n",
"GPT_CONFIG_124M = {\n",
" \"vocab_size\": 50257, # Vocabulary size\n",
" \"context_length\": 256, # Shortened context length (orig: 1024)\n",
" \"emb_dim\": 768, # Embedding dimension\n",
" \"n_heads\": 12, # Number of attention heads\n",
" \"n_layers\": 12, # Number of layers\n",
" \"drop_rate\": 0.1, # Dropout rate\n",
" \"qkv_bias\": False # Query-key-value bias\n",
"}\n",
"\n",
"\n",
"torch.manual_seed(123)\n",
"\n",
"tokenizer = tiktoken.get_encoding(\"gpt2\")"
]
},
{
"cell_type": "code",
"execution_count": 14,
"id": "d8373461-7dad-47da-a489-3e23f0799b23",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"File already exists and is up-to-date: gpt2/124M/checkpoint\n",
"File already exists and is up-to-date: gpt2/124M/encoder.json\n",
"File already exists and is up-to-date: gpt2/124M/hparams.json\n",
"File already exists and is up-to-date: gpt2/124M/model.ckpt.data-00000-of-00001\n",
"File already exists and is up-to-date: gpt2/124M/model.ckpt.index\n",
"File already exists and is up-to-date: gpt2/124M/model.ckpt.meta\n",
"File already exists and is up-to-date: gpt2/124M/vocab.bpe\n"
]
}
],
"source": [
"from gpt_download import download_and_load_gpt2\n",
"\n",
"settings, params = download_and_load_gpt2(model_size=\"124M\", models_dir=\"gpt2\")"
]
},
{
"cell_type": "code",
"execution_count": 15,
"id": "cdd44873-d6c2-4471-a20f-f639b09fdcd3",
"metadata": {},
"outputs": [],
"source": [
"# Define model configurations in a dictionary for compactness\n",
"model_configs = {\n",
" \"gpt2-small (124M)\": {\"emb_dim\": 768, \"n_layers\": 12, \"n_heads\": 12},\n",
" \"gpt2-medium (355M)\": {\"emb_dim\": 1024, \"n_layers\": 24, \"n_heads\": 16},\n",
" \"gpt2-large (774M)\": {\"emb_dim\": 1280, \"n_layers\": 36, \"n_heads\": 20},\n",
" \"gpt2-xl (1558M)\": {\"emb_dim\": 1600, \"n_layers\": 48, \"n_heads\": 25},\n",
"}\n",
"\n",
"# Copy the base configuration and update with specific model settings\n",
"model_name = \"gpt2-small (124M)\" # Example model name\n",
"NEW_CONFIG = GPT_CONFIG_124M.copy()\n",
"NEW_CONFIG.update(model_configs[model_name])\n",
"NEW_CONFIG.update({\"context_length\": 1024, \"qkv_bias\": True})\n",
"\n",
"gpt = GPTModel(NEW_CONFIG)\n",
"gpt.eval();"
]
},
{
"cell_type": "code",
"execution_count": 16,
"id": "c7d562e4-33f6-4611-9b75-6ad1cb441d3b",
"metadata": {},
"outputs": [],
"source": [
"from gpt_generate import load_weights_into_gpt\n",
"\n",
"\n",
"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
"load_weights_into_gpt(gpt, params)\n",
"gpt.to(device);"
]
},
{
"cell_type": "code",
"execution_count": 17,
"id": "46eda9ea-ccb0-46ee-931b-3c07502b2544",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import urllib.request\n",
"from previous_chapters import create_dataloader_v1\n",
"\n",
"\n",
"file_path = \"the-verdict.txt\"\n",
"url = \"https://raw.githubusercontent.com/rasbt/LLMs-from-scratch/main/ch02/01_main-chapter-code/the-verdict.txt\"\n",
"\n",
"if not os.path.exists(file_path):\n",
" with urllib.request.urlopen(url) as response:\n",
" text_data = response.read().decode('utf-8')\n",
" with open(file_path, \"w\", encoding=\"utf-8\") as file:\n",
" file.write(text_data)\n",
"else:\n",
" with open(file_path, \"r\", encoding=\"utf-8\") as file:\n",
" text_data = file.read()\n",
"\n",
"\n",
"# Train/validation ratio\n",
"train_ratio = 0.90\n",
"split_idx = int(train_ratio * len(text_data))\n",
"train_data = text_data[:split_idx]\n",
"val_data = text_data[split_idx:]\n",
"\n",
"\n",
"torch.manual_seed(123)\n",
"\n",
"train_loader = create_dataloader_v1(\n",
" train_data,\n",
" batch_size=2,\n",
" max_length=GPT_CONFIG_124M[\"context_length\"],\n",
" stride=GPT_CONFIG_124M[\"context_length\"],\n",
" drop_last=True,\n",
" shuffle=True,\n",
" num_workers=0\n",
")\n",
"\n",
"val_loader = create_dataloader_v1(\n",
" val_data,\n",
" batch_size=2,\n",
" max_length=GPT_CONFIG_124M[\"context_length\"],\n",
" stride=GPT_CONFIG_124M[\"context_length\"],\n",
" drop_last=False,\n",
" shuffle=False,\n",
" num_workers=0\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 18,
"id": "4e3574a2-687d-47a2-a2f6-457fe9d595f1",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Training loss: 3.7547486888037787\n",
"Validation loss: 3.5596182346343994\n"
]
}
],
"source": [
"from gpt_train import calc_loss_loader\n",
"\n",
"torch.manual_seed(123) # For reproducibility due to the shuffling in the data loader\n",
"train_loss = calc_loss_loader(train_loader, gpt, device)\n",
"val_loss = calc_loss_loader(val_loader, gpt, device)\n",
"\n",
"print(\"Training loss:\", train_loss)\n",
"print(\"Validation loss:\", val_loss)"
]
},
{
"cell_type": "markdown",
"id": "96485d6b-bf1f-4bc0-a53f-73b08d85726e",
"metadata": {},
"source": [
"We can also repeat this for the largest GPT-2 model, but don't forget to update the context length:"
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "1a79a4b6-fe8f-40c2-a018-e731dcf391b3",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"checkpoint: 100%|███████████████████████████| 77.0/77.0 [00:00<00:00, 43.5kiB/s]\n",
"encoder.json: 100%|███████████████████████| 1.04M/1.04M [00:00<00:00, 2.75MiB/s]\n",
"hparams.json: 100%|█████████████████████████| 91.0/91.0 [00:00<00:00, 60.2kiB/s]\n",
"model.ckpt.data-00000-of-00001: 100%|█████| 6.23G/6.23G [06:02<00:00, 17.2MiB/s]\n",
"model.ckpt.index: 100%|████████████████████| 20.7k/20.7k [00:00<00:00, 171kiB/s]\n",
"model.ckpt.meta: 100%|████████████████████| 1.84M/1.84M [00:00<00:00, 4.27MiB/s]\n",
"vocab.bpe: 100%|████████████████████████████| 456k/456k [00:00<00:00, 1.73MiB/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Training loss: 3.3046312861972384\n",
"Validation loss: 3.1195147037506104\n"
]
}
],
"source": [
"settings, params = download_and_load_gpt2(model_size=\"1558M\", models_dir=\"gpt2\")\n",
"\n",
"model_name = \"gpt2-xl (1558M)\"\n",
"NEW_CONFIG = GPT_CONFIG_124M.copy()\n",
"NEW_CONFIG.update(model_configs[model_name])\n",
"NEW_CONFIG.update({\"context_length\": 1024, \"qkv_bias\": True})\n",
"\n",
"gpt = GPTModel(NEW_CONFIG)\n",
"gpt.eval()\n",
"\n",
"load_weights_into_gpt(gpt, params)\n",
"gpt.to(device)\n",
"\n",
"torch.manual_seed(123)\n",
"train_loss = calc_loss_loader(train_loader, gpt, device)\n",
"val_loss = calc_loss_loader(val_loader, gpt, device)\n",
"\n",
"print(\"Training loss:\", train_loss)\n",
"print(\"Validation loss:\", val_loss)"
]
},
{
"cell_type": "markdown",
"id": "3a76a1e0-9635-480a-9391-3bda7aea402d",
"metadata": {},
"source": [
"# Exercise 5.6: Trying larger models"
]
},
{
"cell_type": "markdown",
"id": "b3d313f4-0038-4bc9-a340-84b3b55dc0e3",
"metadata": {},
"source": [
"- In the main chapter, we experimented with the smallest GPT-2 model, which has only 124M parameters\n",
"- The reason was to keep the resource requirements as low as possible\n",
"- However, you can easily experiment with larger models with minimal code changes\n",
"- For example, instead of loading the 1558M instead of 124M model in chapter 5, the only 2 lines of code that we have to change are\n",
"\n",
"```python\n",
"settings, params = download_and_load_gpt2(model_size=\"124M\", models_dir=\"gpt2\")\n",
"model_name = \"gpt2-small (124M)\"\n",
"```\n",
"\n",
"- The updated code becomes\n",
"\n",
"\n",
"```python\n",
"settings, params = download_and_load_gpt2(model_size=\"1558M\", models_dir=\"gpt2\")\n",
"model_name = \"gpt2-xl (1558M)\"\n",
"```"
]
},
{
"cell_type": "code",
"execution_count": 20,
"id": "31e0972b-e85e-4904-a0f5-24c3eacd5fa2",
"metadata": {},
"outputs": [],
"source": [
"import tiktoken\n",
"import torch\n",
"from previous_chapters import GPTModel\n",
"\n",
"\n",
"GPT_CONFIG_124M = {\n",
" \"vocab_size\": 50257, # Vocabulary size\n",
" \"context_length\": 256, # Shortened context length (orig: 1024)\n",
" \"emb_dim\": 768, # Embedding dimension\n",
" \"n_heads\": 12, # Number of attention heads\n",
" \"n_layers\": 12, # Number of layers\n",
" \"drop_rate\": 0.1, # Dropout rate\n",
" \"qkv_bias\": False # Query-key-value bias\n",
"}\n",
"\n",
"\n",
"tokenizer = tiktoken.get_encoding(\"gpt2\")"
]
},
{
"cell_type": "code",
"execution_count": 21,
"id": "b641ee88-f9d4-43ec-a787-e34199eed356",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"File already exists and is up-to-date: gpt2/1558M/checkpoint\n",
"File already exists and is up-to-date: gpt2/1558M/encoder.json\n",
"File already exists and is up-to-date: gpt2/1558M/hparams.json\n",
"File already exists and is up-to-date: gpt2/1558M/model.ckpt.data-00000-of-00001\n",
"File already exists and is up-to-date: gpt2/1558M/model.ckpt.index\n",
"File already exists and is up-to-date: gpt2/1558M/model.ckpt.meta\n",
"File already exists and is up-to-date: gpt2/1558M/vocab.bpe\n"
]
}
],
"source": [
"from gpt_download import download_and_load_gpt2\n",
"from gpt_generate import load_weights_into_gpt\n",
"\n",
"\n",
"model_configs = {\n",
" \"gpt2-small (124M)\": {\"emb_dim\": 768, \"n_layers\": 12, \"n_heads\": 12},\n",
" \"gpt2-medium (355M)\": {\"emb_dim\": 1024, \"n_layers\": 24, \"n_heads\": 16},\n",
" \"gpt2-large (774M)\": {\"emb_dim\": 1280, \"n_layers\": 36, \"n_heads\": 20},\n",
" \"gpt2-xl (1558M)\": {\"emb_dim\": 1600, \"n_layers\": 48, \"n_heads\": 25},\n",
"}\n",
"\n",
"model_name = \"gpt2-xl (1558M)\"\n",
"NEW_CONFIG = GPT_CONFIG_124M.copy()\n",
"NEW_CONFIG.update(model_configs[model_name])\n",
"NEW_CONFIG.update({\"context_length\": 1024, \"qkv_bias\": True})\n",
"\n",
"gpt = GPTModel(NEW_CONFIG)\n",
"gpt.eval()\n",
"\n",
"settings, params = download_and_load_gpt2(model_size=\"1558M\", models_dir=\"gpt2\")\n",
"load_weights_into_gpt(gpt, params)"
]
},
{
"cell_type": "code",
"execution_count": 22,
"id": "c98f56f4-98fc-43b4-9ee5-726e9d17c73f",
"metadata": {},
"outputs": [],
"source": [
"from gpt_generate import generate, text_to_token_ids, token_ids_to_text"
]
},
{
"cell_type": "code",
"execution_count": 23,
"id": "b1f7853c-6e81-4f1f-a1d0-61e2c7d33a20",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Output text:\n",
" Every effort moves you toward finding an ideal life. You don't have to accept your current one at once, because if you do you'll never\n"
]
}
],
"source": [
"torch.manual_seed(123)\n",
"\n",
"token_ids = generate(\n",
" model=gpt,\n",
" idx=text_to_token_ids(\"Every effort moves you\", tokenizer),\n",
" max_new_tokens=25,\n",
" context_size=NEW_CONFIG[\"context_length\"],\n",
" top_k=50,\n",
" temperature=1.5\n",
")\n",
"\n",
"print(\"Output text:\\n\", token_ids_to_text(token_ids, tokenizer))"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.16"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Reference in New Issue View Git Blame Copy Permalink