nandigama/Hands-On-Large-Language-Models
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix loading Phi-3 (#48)

Browse Source
This commit is contained in:
Maarten Grootendorst
2025-04-04 08:34:35 +02:00
committed by GitHub
parent e55a0738aa
commit aabef6b341
6 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
chapter01/Chapter 1 - Introduction to Language Models.ipynb
View File

@@ -69,7 +69,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 5, "execution_count": null,
"metadata": { "metadata": {
"id": "RSNalRXZyTTk" "id": "RSNalRXZyTTk"
}, },
@@ -82,7 +82,7 @@
" \"microsoft/Phi-3-mini-4k-instruct\",\n", " \"microsoft/Phi-3-mini-4k-instruct\",\n",
" device_map=\"cuda\",\n", " device_map=\"cuda\",\n",
" torch_dtype=\"auto\",\n", " torch_dtype=\"auto\",\n",
" trust_remote_code=True,\n", " trust_remote_code=False,\n",
")\n", ")\n",
"tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")" "tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")"
] ]

4
chapter02/Chapter 2 - Tokens and Token Embeddings.ipynb
View File

@@ -64,7 +64,7 @@
}, },
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": 1, "execution_count": null,
"metadata": { "metadata": {
"colab": { "colab": {
"base_uri": "https://localhost:8080/", "base_uri": "https://localhost:8080/",
@@ -278,7 +278,7 @@
" \"microsoft/Phi-3-mini-4k-instruct\",\n", " \"microsoft/Phi-3-mini-4k-instruct\",\n",
" device_map=\"cuda\",\n", " device_map=\"cuda\",\n",
" torch_dtype=\"auto\",\n", " torch_dtype=\"auto\",\n",
" trust_remote_code=True,\n", " trust_remote_code=False,\n",
")\n", ")\n",
"tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")" "tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")"
] ]

2
chapter03/Chapter 3 - Looking Inside LLMs.ipynb
View File

@@ -483,7 +483,7 @@
" \"microsoft/Phi-3-mini-4k-instruct\",\n", " \"microsoft/Phi-3-mini-4k-instruct\",\n",
" device_map=\"cuda\",\n", " device_map=\"cuda\",\n",
" torch_dtype=\"auto\",\n", " torch_dtype=\"auto\",\n",
" trust_remote_code=True,\n", " trust_remote_code=False,\n",
")\n", ")\n",
"\n", "\n",
"# Create a pipeline\n", "# Create a pipeline\n",

2
chapter05/Chapter 5 - Text Clustering and Topic Modeling.ipynb
View File

@@ -100,7 +100,7 @@
"\n", "\n",
"# Extract metadata\n", "# Extract metadata\n",
"abstracts = dataset[\"Abstracts\"]\n", "abstracts = dataset[\"Abstracts\"]\n",
"titles = dataset[\"Titles\"]" "titles = dataset[\"Titles\"] "
] ]
}, },
{ {

2
chapter06/Chapter 6 - Prompt Engineering.ipynb
View File

@@ -142,7 +142,7 @@
" \"microsoft/Phi-3-mini-4k-instruct\",\n", " \"microsoft/Phi-3-mini-4k-instruct\",\n",
" device_map=\"cuda\",\n", " device_map=\"cuda\",\n",
" torch_dtype=\"auto\",\n", " torch_dtype=\"auto\",\n",
" trust_remote_code=True,\n", " trust_remote_code=False,\n",
")\n", ")\n",
"tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")\n", "tokenizer = AutoTokenizer.from_pretrained(\"microsoft/Phi-3-mini-4k-instruct\")\n",
"\n", "\n",

4
chapter12/Chapter 12 - Fine-tuning Generation Models.ipynb
View File

@@ -859,7 +859,7 @@
"model.config.pretraining_tp = 1\n", "model.config.pretraining_tp = 1\n",
"\n", "\n",
"# Load LLaMA tokenizer\n", "# Load LLaMA tokenizer\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)\n", "tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=False)\n",
"tokenizer.pad_token = \"<PAD>\"\n", "tokenizer.pad_token = \"<PAD>\"\n",
"tokenizer.padding_side = \"left\"" "tokenizer.padding_side = \"left\""
] ]
@@ -1598,7 +1598,7 @@
"\n", "\n",
"# Load LLaMA tokenizer\n", "# Load LLaMA tokenizer\n",
"model_name = \"TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T\"\n", "model_name = \"TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T\"\n",
"tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)\n", "tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=False)\n",
"tokenizer.pad_token = \"<PAD>\"\n", "tokenizer.pad_token = \"<PAD>\"\n",
"tokenizer.padding_side = \"left\"" "tokenizer.padding_side = \"left\""
] ]