LeCun likes: Running LLaMA on Apple M1/M2 chip! The 13 billion parameter model requires only 4GB of memory

Not long ago, Meta released the open source large language model LLaMA, but then netizens released a no-threshold download link, which was "miserably" open.

As soon as the news came out, the circle immediately became lively, and everyone began to download and test it.

But those friends who don’t have top-level graphics cards can only look at the model and sigh.

However, it’s not a big problem. Georgi Gerganov recently made a project called "llama.cpp" - LLaMA can be run without a GPU.

## Project address: https://github.com/ggerganov/llama.cpp

Yes, this also includes Macs equipped with Apple chips. And also received forwarding support from LeCun.

Running LLaMA on M1/M2 Mac

For now, this is a relatively comprehensive tutorial. There are two, based on Apple's M1 and M2 processors.

##First article: https://dev.l1x.be/posts/2023/03/ 12/using-llama-with-m1-mac/

##Second article: https ://til.simonwillison.net/llms/llama-7b-m2

Without further ado, let’s look at the effect first.

For example, on the smallest 7B model, ask: "Who was the first person to land on the moon?"

Very quickly, the results came out after a few seconds.

-p 'The first man to land on the moon was'

the first man to land on the moon The person is 38-year-old astronaut Neil A. Armstrong.

On July 20, 1969, Apollo 11 landed on the moon.

Apollo 11 was the first manned mission to land on the moon and was the climax of the space race. In July 1969, Neil Armstrong and Edwin "Buzz" Aldrin became the first humans to land on the moon.

Apollo 11 launched from the Kennedy Space Center on July 16.

There are no obvious factual errors in Armstrong’s age, middle name and date of landing on the moon. This is not easy for such a small model.

If you use the following tips, you can generate some practical Python code.

-p 'def open_and_return_content(filename):'

def open_and_return_content(filename):
"""Opens file (returning the content) and performs basic sanity checks"""if os.path.isfile(filename):
with open(filename) as f:
content = f.read()
return contentelse:
print('WARNING: file "{}" does not exist'.format(filename), file=sys.stderr)
return ''def get_file_info(filename, fullpath):
"""Get file information (i.e., permission, owner, group, size)"""

接下来，我们就来看看具体是如何实现的。

Step 1: Download the model

The first thing to do is to download the LLaMA model.

You can submit an application to Meta through the official form, or obtain it directly from the link shared by netizens.

In short, when you are done, you will see the following pile of things:

As you can see, different models are in different folders. Each model has a params.json containing details about the model. For example:

Step 2: Install dependencies

首先，你需要安装Xcode来编译C++项目。

xcode-select --install

接下来，是构建C++项目的依赖项（pkgconfig和cmake）。

brew install pkgconfig cmake

在环境的配置上，假如你用的是Python 3.11，则可以创建一个虚拟环境：

/opt/homebrew/bin/python3.11 -m venv venv

然后激活venv。（如果是fish以外的shell，只要去掉.fish后缀即可）

. venv/bin/activate.fish

最后，安装Torch。

pip3 install --pre torch torchvision --extra-index-url https://download.pytorch.org/whl/nightly/cpu

如果你对利用新的Metal性能着色器（MPS）后端进行GPU训练加速感兴趣，可以通过运行以下程序来进行验证。但这不是在M1上运行LLaMA的必要条件。

python
Python 3.11.2 (main, Feb 16 2023, 02:55:59) [Clang 14.0.0 (clang-1400.0.29.202)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch; torch.backends.mps.is_available()True

第三步：编译LLaMA CPP

git clone git@github.com:ggerganov/llama.cpp.git

在安装完所有的依赖项后，你可以运行make：

make
I llama.cpp build info:
I UNAME_S:Darwin
I UNAME_P:arm
I UNAME_M:arm64
I CFLAGS: -I.-O3 -DNDEBUG -std=c11 -fPIC -pthread -DGGML_USE_ACCELERATE
I CXXFLAGS: -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -pthread
I LDFLAGS: -framework Accelerate
I CC: Apple clang version 14.0.0 (clang-1400.0.29.202)I CXX:Apple clang version 14.0.0 (clang-1400.0.29.202)
cc-I.-O3 -DNDEBUG -std=c11 -fPIC -pthread -DGGML_USE_ACCELERATE -c ggml.c -o ggml.o
c++ -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -pthread -c utils.cpp -o utils.o
c++ -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -pthread main.cpp ggml.o utils.o -o main-framework Accelerate
./main -h
usage: ./main [options]
options:
-h, --helpshow this help message and exit
-s SEED, --seed SEEDRNG seed (default: -1)
-t N, --threads N number of threads to use during computation (default: 4)
-p PROMPT, --prompt PROMPT
prompt to start generation with (default: random)
-n N, --n_predict N number of tokens to predict (default: 128)
--top_k N top-k sampling (default: 40)
--top_p N top-p sampling (default: 0.9)
--temp Ntemperature (default: 0.8)
-b N, --batch_size Nbatch size for prompt processing (default: 8)
-m FNAME, --model FNAME
model path (default: models/llama-7B/ggml-model.bin)
c++ -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -pthread quantize.cpp ggml.o utils.o -o quantize-framework Accelerate

第四步：转换模型

假设你已经把模型放在llama.cpp repo中的models/下。

python convert-pth-to-ggml.py models/7B 1

那么，应该会看到像这样的输出：

{'dim': 4096, 'multiple_of': 256, 'n_heads': 32, 'n_layers': 32, 'norm_eps': 1e-06, 'vocab_size': 32000}n_parts =1Processing part0Processing variable: tok_embeddings.weight with shape:torch.Size([32000, 4096])and type:torch.float16
Processing variable: norm.weight with shape:torch.Size([4096])and type:torch.float16
Converting to float32
Processing variable: output.weight with shape:torch.Size([32000, 4096])and type:torch.float16
Processing variable: layers.0.attention.wq.weight with shape:torch.Size([4096, 4096])and type:torch.f
loat16
Processing variable: layers.0.attention.wk.weight with shape:torch.Size([4096, 4096])and type:torch.f
loat16
Processing variable: layers.0.attention.wv.weight with shape:torch.Size([4096, 4096])and type:torch.f
loat16
Processing variable: layers.0.attention.wo.weight with shape:torch.Size([4096, 4096])and type:torch.f
loat16
Processing variable: layers.0.feed_forward.w1.weight with shape:torch.Size([11008, 4096])and type:tor
ch.float16
Processing variable: layers.0.feed_forward.w2.weight with shape:torch.Size([4096, 11008])and type:tor
ch.float16
Processing variable: layers.0.feed_forward.w3.weight with shape:torch.Size([11008, 4096])and type:tor
ch.float16
Processing variable: layers.0.attention_norm.weight with shape:torch.Size([4096])and type:torch.float
16...
Done. Output file: models/7B/ggml-model-f16.bin, (part0 )

下一步将是进行量化处理：

./quantize ./models/7B/ggml-model-f16.bin ./models/7B/ggml-model-q4_0.bin 2

输出如下：

llama_model_quantize: loading model from './models/7B/ggml-model-f16.bin'llama_model_quantize: n_vocab = 32000llama_model_quantize: n_ctx = 512llama_model_quantize: n_embd= 4096llama_model_quantize: n_mult= 256llama_model_quantize: n_head= 32llama_model_quantize: n_layer = 32llama_model_quantize: f16 = 1...
layers.31.attention_norm.weight - [ 4096, 1], type =f32 size =0.016 MB
layers.31.ffn_norm.weight - [ 4096, 1], type =f32 size =0.016 MB
llama_model_quantize: model size= 25705.02 MB
llama_model_quantize: quant size=4017.27 MB
llama_model_quantize: hist: 0.000 0.022 0.019 0.033 0.053 0.078 0.104 0.125 0.134 0.125 0.104 0.078 0.053 0.033 0.019 0.022


main: quantize time = 29389.45 ms
main:total time = 29389.45 ms

第五步：运行模型

./main -m ./models/7B/ggml-model-q4_0.bin 
-t 8 
-n 128 
-p 'The first president of the USA was '

main: seed = 1678615879llama_model_load: loading model from './models/7B/ggml-model-q4_0.bin' - please wait ...
llama_model_load: n_vocab = 32000llama_model_load: n_ctx = 512llama_model_load: n_embd= 4096llama_model_load: n_mult= 256llama_model_load: n_head= 32llama_model_load: n_layer = 32llama_model_load: n_rot = 128llama_model_load: f16 = 2llama_model_load: n_ff= 11008llama_model_load: n_parts = 1llama_model_load: ggml ctx size = 4529.34 MB
llama_model_load: memory_size = 512.00 MB, n_mem = 16384llama_model_load: loading model part 1/1 from './models/7B/ggml-model-q4_0.bin'llama_model_load: .................................... donellama_model_load: model size =4017.27 MB / num tensors = 291
main: prompt: 'The first president of the USA was 'main: number of tokens in prompt = 9 1 -> ''1576 -> 'The' 937 -> ' first'6673 -> ' president' 310 -> ' of' 278 -> ' the'8278 -> ' USA' 471 -> ' was' 29871 -> ' '
sampling parameters: temp = 0.800000, top_k = 40, top_p = 0.950000


The first president of the USA was 57 years old when he assumed office (George Washington). Nowadays, the US electorate expects the new president to be more young at heart. President Donald Trump was 70 years old when he was inaugurated. In contrast to his predecessors, he is physically fit, healthy and active. And his fitness has been a prominent theme of his presidency. During the presidential campaign, he famously said he
 would be the “most active president ever” — a statement Trump has not yet achieved, but one that fits his approach to the office. His tweets demonstrate his physical activity.


main: mem per token = 14434244 bytes
main: load time =1311.74 ms
main: sample time = 278.96 ms
main:predict time =7375.89 ms / 54.23 ms per token
main:total time =9216.61 ms

资源使用情况

第二位博主表示，在运行时，13B模型使用了大约4GB的内存，以及748%的CPU。（设定的就是让模型使用8个CPU核心）

没有指令微调

GPT-3和ChatGPT效果如此之好的关键原因之一是，它们都经过了指令微调，

这种额外的训练使它们有能力对人类的指令做出有效的反应。比如「总结一下这个」或「写一首关于水獭的诗」或「从这篇文章中提取要点」。

撰写教程的博主表示，据他观察，LLaMA并没有这样的能力。

也就是说，给LLaMA的提示需要采用经典的形式：「一些将由......完成的文本」。这也让提示工程变得更加困难。

举个例子，博主至今都还没有想出一个正确的提示，从而让LLaMA实现文本的总结。

The above is the detailed content of LeCun likes: Running LLaMA on Apple M1/M2 chip! The 13 billion parameter model requires only 4GB of memory. For more information, please follow other related articles on the PHP Chinese website!