update the README and reorganize the docker guides structure. (#11016)
* update the README and reorganize the docker guides structure. * modified docker install guide into overview
This commit is contained in:
Shengsheng Huang
GitHub
parent
893197434d
commit
586a151f9c
8 changed files with 896 additions and 827 deletions
|
|
@ -1,664 +1,47 @@
|
|||
# Getting started with IPEX-LLM in Docker
|
||||
|
||||
This guide provides step-by-step instructions for installing and using IPEX-LLM in a Docker environment. It covers setups for both CPU and XPU (accelerated processing units) on different operating systems.
|
||||
|
||||
### Index
|
||||
- [Docker Installation](#docker-installation-instructions)
|
||||
- [IPEX-LLM Inference](#ipex-llm-inference-on-cpu)
|
||||
- [On CPU](#ipex-llm-inference-on-cpu)
|
||||
- [On XPU](#ipex-llm-inference-on-xpu)
|
||||
- [IPEX-LLM Serving](#ipex-llm-serving-on-cpu)
|
||||
- [On CPU](#ipex-llm-serving-on-cpu)
|
||||
- [On XPU](#ipex-llm-serving-on-xpu)
|
||||
- [IPEX-LLM Fine Tuning](#ipex-llm-fine-tuning-on-cpu)
|
||||
- [On CPU](#ipex-llm-fine-tuning-on-cpu)
|
||||
- [On XPU](#ipex-llm-fine-tuning-on-xpu)
|
||||
|
||||
|
||||
## Docker Installation Instructions
|
||||
|
||||
1. **For New Users:**
|
||||
- Begin by visiting the [official Docker Get Started page](https://www.docker.com/get-started/) for a comprehensive introduction and installation guide.
|
||||
|
||||
2. **Additional Steps for Windows Users:**
|
||||
- For Windows Users, follow the step-by-step guide: [Docker Installation Instructions for Windows](https://github.com/intel-analytics/ipex-llm/blob/main/docs/readthedocs/source/doc/LLM/Quickstart/docker_windows_gpu.md).
|
||||
|
||||
|
||||
## IPEX-LLM Inference on CPU
|
||||
|
||||
### 1. Prepare ipex-llm-cpu Docker Image
|
||||
|
||||
Run the following command to pull image:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start bigdl-llm-cpu Docker Container
|
||||
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--privileged \
|
||||
-p 12345:12345 \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Start Inference and Tutorials
|
||||
**3.1 Chat Interface**: Use `chat.py` for conversational AI.
|
||||
|
||||
For example, if your model is Llama-2-7b-chat-hf and mounted on /llm/models, you can excute the following command to initiate a conversation:
|
||||
```bash
|
||||
cd /llm/portable-zip
|
||||
python chat.py --model-path /llm/models/Llama-2-7b-chat-hf
|
||||
```
|
||||
Here is a demostration:
|
||||
|
||||
<a align="left" href="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-chatpy-demo.gif">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-chatpy-demo.gif" width='60%' />
|
||||
|
||||
</a><br>
|
||||
|
||||
**3.2 Jupyter Lab Tutorials**: Start a Jupyter Lab session for IPEX-LLM tutorials.
|
||||
|
||||
Run the following command to start notebook:
|
||||
```bash
|
||||
cd /llm
|
||||
./start-notebook.sh [--port EXPECTED_PORT]
|
||||
```
|
||||
The default port is 12345, you could assign a different port by specifying the --port parameter.
|
||||
|
||||
If you're using the host network mode when booting the container, once the service is running successfully, you can access the tutorial at http://127.0.0.1:12345/lab. Alternatively, you need to ensure the correct ports are bound between the container and the host.
|
||||
|
||||
Here's a demonstration of how to navigate the tutorial in the explorer:
|
||||
|
||||
<a align="left" href="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-tutorial-demo.gif">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-tutorial-demo.gif" width='60%' />
|
||||
|
||||
</a> <br>
|
||||
|
||||
**3.3 Performance Benchmark**: We provide a benchmark tool help users to test all the benchmarks and record them in a result CSV.
|
||||
|
||||
```bash
|
||||
cd /llm/benchmark/all-in-one
|
||||
```
|
||||
|
||||
Users can provide models and related information in config.yaml.
|
||||
```bash
|
||||
repo_id:
|
||||
# - 'THUDM/chatglm-6b'
|
||||
# - 'THUDM/chatglm2-6b'
|
||||
- 'meta-llama/Llama-2-7b-chat-hf'
|
||||
# - 'liuhaotian/llava-v1.5-7b' # requires a LLAVA_REPO_DIR env variables pointing to the llava dir; added only for gpu win related test_api now
|
||||
local_model_hub: 'path to your local model hub'
|
||||
warm_up: 1
|
||||
num_trials: 3
|
||||
num_beams: 1 # default to greedy search
|
||||
low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
|
||||
batch_size: 1 # default to 1
|
||||
in_out_pairs:
|
||||
- '32-32'
|
||||
- '1024-128'
|
||||
test_api:
|
||||
# - "transformer_int4"
|
||||
# - "native_int4"
|
||||
# - "optimize_model"
|
||||
# - "pytorch_autocast_bf16"
|
||||
# - "transformer_autocast_bf16"
|
||||
# - "bigdl_ipex_bf16"
|
||||
# - "bigdl_ipex_int4"
|
||||
# - "bigdl_ipex_int8"
|
||||
# - "ipex_fp16_gpu" # on Intel GPU
|
||||
# - "bigdl_fp16_gpu" # on Intel GPU
|
||||
# - "transformer_int4_gpu" # on Intel GPU
|
||||
# - "optimize_model_gpu" # on Intel GPU
|
||||
# - "deepspeed_transformer_int4_cpu" # on Intel SPR Server
|
||||
# - "transformer_int4_gpu_win" # on Intel GPU for Windows
|
||||
# - "transformer_int4_fp16_gpu_win" # on Intel GPU for Windows, use fp16 for non-linear layer
|
||||
# - "transformer_int4_loadlowbit_gpu_win" # on Intel GPU for Windows using load_low_bit API. Please make sure you have used the save.py to save the converted low bit model
|
||||
# - "deepspeed_optimize_model_gpu" # deepspeed autotp on Intel GPU
|
||||
- "speculative_cpu"
|
||||
# - "speculative_gpu"
|
||||
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)
|
||||
streaming: False # whether output in streaming way (only avaiable now for gpu win related test_api)
|
||||
```
|
||||
|
||||
This benchmark tool offers various test APIs, including `transformer_int4`, `speculative_cpu`, and more.
|
||||
|
||||
For instance, if you wish to benchmark **inference with speculative decoding**, utilize the `speculative_cpu` test API in the `config.yml` file.
|
||||
|
||||
Then, execute `bash run-spr.sh`, which will generate output results in `results.csv`.
|
||||
```bash
|
||||
bash run-spr.sh
|
||||
```
|
||||
|
||||
For further details and comprehensive functionality of the benchmark tool, please refer to the [all-in-one benchmark tool](https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/dev/benchmark/all-in-one).
|
||||
|
||||
Additionally, for examples related to Inference with Speculative Decoding, you can explore [Speculative-Decoding Examples](https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/example/CPU/Speculative-Decoding).
|
||||
|
||||
|
||||
|
||||
## IPEX-LLM Inference on XPU
|
||||
|
||||
### 1. Prepare ipex-llm-xpu Docker Image
|
||||
|
||||
Run the following command to pull image from dockerhub:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start Chat Inference
|
||||
|
||||
We provide `chat.py` for conversational AI. If your model is Llama-2-7b-chat-hf and mounted on /llm/models, you can execute the following command to initiate a conversation:
|
||||
|
||||
To map the xpu into the container, you need to specify --device=/dev/dri when booting the container.
|
||||
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE bash -c "python chat.py --model-path /llm/models/Llama-2-7b-chat-hf"
|
||||
```
|
||||
|
||||
|
||||
### 3. Quick Performance Benchmark
|
||||
|
||||
Execute a quick performance benchmark by starting the ipex-llm-xpu container, specifying the model, test API, and device, then running the benchmark.sh script.
|
||||
|
||||
To map the XPU into the container, specify `--device=/dev/dri` when booting the container.
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models [change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
-e REPO_IDS="meta-llama/Llama-2-7b-chat-hf" \
|
||||
-e TEST_APIS="transformer_int4_gpu" \
|
||||
-e DEVICE=Arc \
|
||||
$DOCKER_IMAGE /llm/benchmark.sh
|
||||
```
|
||||
|
||||
Customize environment variables to specify:
|
||||
|
||||
- **REPO_IDS:** Model's name and organization, separated by commas if multiple values exist.
|
||||
- **TEST_APIS:** Different test functions based on the machine, separated by commas if multiple values exist.
|
||||
- **DEVICE:** Type of device - Max, Flex, Arc.
|
||||
|
||||
**Result**
|
||||
|
||||
Upon completion, you can obtain a CSV result file, the content of CSV results will be printed out. You can mainly look at the results of columns `1st token avg latency (ms)` and `2+ avg latency (ms/token)` for the benchmark results.
|
||||
|
||||
## IPEX-LLM Serving on CPU
|
||||
FastChat is an open platform for training, serving, and evaluating large language model based chatbots. You can find the detailed information at their [homepage](https://github.com/lm-sys/FastChat).
|
||||
|
||||
IPEX-LLM is integrated into FastChat so that user can use IPEX-LLM as a serving backend in the deployment.
|
||||
|
||||
### 1. Prepare ipex-llm-serving-cpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start ipex-llm-serving-cpu Docker Container
|
||||
|
||||
Please be noted that the CPU config is specified for Xeon CPUs, change it accordingly if you are not using a Xeon CPU.
|
||||
|
||||
```bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Serving with FastChat
|
||||
|
||||
To serve using the Web UI, you need three main components: web servers that interface with users, model workers that host one or more models, and a controller to coordinate the web server and model workers.
|
||||
|
||||
- #### **Step 1: Launch the Controller**
|
||||
```bash
|
||||
python3 -m fastchat.serve.controller &
|
||||
```
|
||||
|
||||
This controller manages the distributed workers.
|
||||
|
||||
- #### **Step 2: Launch the model worker(s)**
|
||||
|
||||
Using IPEX-LLM in FastChat does not impose any new limitations on model usage. Therefore, all Hugging Face Transformer models can be utilized in FastChat.
|
||||
```bash
|
||||
source ipex-llm-init -t
|
||||
|
||||
# Available low_bit format including sym_int4, sym_int8, bf16 etc.
|
||||
python3 -m ipex_llm.serving.fastchat.ipex_llm_worker --model-path path/to/vicuna-7b-v1.5 --low-bit "sym_int4" --trust-remote-code --device "cpu" &
|
||||
```
|
||||
Wait until the process finishes loading the model and you see "Uvicorn running on ...". The model worker will register itself to the controller.
|
||||
|
||||
- #### **Step 3: Launch Gradio web server or RESTful API server**
|
||||
You can launch Gradio web server to serve your models using the web UI or launch RESTful API server to serve with HTTP.
|
||||
|
||||
- **Option 1: Serving with Web UI**
|
||||
```bash
|
||||
python3 -m fastchat.serve.gradio_web_server &
|
||||
```
|
||||
This is the user interface that users will interact with.
|
||||
|
||||
By following these steps, you will be able to serve your models using the web UI with `IPEX-LLM` as the backend. You can open your browser and chat with a model now.
|
||||
|
||||
- **Option 2: Serving with OpenAI-Compatible RESTful APIs**
|
||||
|
||||
Launch the RESTful API server
|
||||
|
||||
```bash
|
||||
python3 -m fastchat.serve.openai_api_server --host 0.0.0.0 --port 8000 &
|
||||
```
|
||||
|
||||
Use curl for testing, an example could be:
|
||||
|
||||
```bash
|
||||
curl -X POST -H "Content-Type: application/json" -d '{
|
||||
"model": "Llama-2-7b-chat-hf",
|
||||
"prompt": "Once upon a time, there existed a little girl who liked to have adventures. She wanted to go to places and meet new people, and have fun",
|
||||
"n": 1,
|
||||
"best_of": 1,
|
||||
"use_beam_search": false,
|
||||
"stream": false
|
||||
}' http://YOUR_HTTP_HOST:8000/v1/completions
|
||||
```
|
||||
You can find more details here [Serving using IPEX-LLM and FastChat](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/src/ipex_llm/serving/fastchat/README.md)
|
||||
|
||||
### 4. Serving with vLLM Continuous Batching
|
||||
To fully utilize the continuous batching feature of the vLLM, you can send requests to the service using curl or other similar methods. The requests sent to the engine will be batched at token level. Queries will be executed in the same forward step of the LLM and be removed when they are finished instead of waiting for all sequences to be finished.
|
||||
|
||||
- #### **Step 1: Launch the api_server**
|
||||
```bash
|
||||
#!/bin/bash
|
||||
# You may also want to adjust the `--max-num-batched-tokens` argument, it indicates the hard limit
|
||||
# of batched prompt length the server will accept
|
||||
numactl -C 0-47 -m 0 python -m ipex_llm.vllm.entrypoints.openai.api_server \
|
||||
--model /llm/models/Llama-2-7b-chat-hf/ \
|
||||
--host 0.0.0.0 --port 8000 \
|
||||
--load-format 'auto' --device cpu --dtype bfloat16 \
|
||||
--max-num-batched-tokens 4096 &
|
||||
```
|
||||
|
||||
- #### **Step 2: Use curl for testing, access the api server as follows:**
|
||||
|
||||
```bash
|
||||
curl http://YOUR_HTTP_HOST:8000/v1/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"model": "/llm/models/Llama-2-7b-chat-hf/",
|
||||
"prompt": "San Francisco is a",
|
||||
"max_tokens": 128,
|
||||
"temperature": 0
|
||||
}' &
|
||||
```
|
||||
|
||||
You can find more details here: [Serving with vLLM Continuous Batching](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/example/CPU/vLLM-Serving/README.md)
|
||||
|
||||
|
||||
## IPEX-LLM Serving on XPU
|
||||
|
||||
FastChat is an open platform for training, serving, and evaluating large language model based chatbots. You can find the detailed information at their [homepage](https://github.com/lm-sys/FastChat).
|
||||
|
||||
IPEX-LLM is integrated into FastChat so that user can use IPEX-LLM as a serving backend in the deployment.
|
||||
|
||||
### 1. Prepare ipex-llm-serving-xpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start ipex-llm-serving-xpu Docker Container
|
||||
|
||||
To map the `xpu` into the container, you need to specify `--device=/dev/dri` when booting the container.
|
||||
|
||||
```bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
To verify the device is successfully mapped into the container, run `sycl-ls` to check the result. In a machine with Arc A770, the sampled output is:
|
||||
|
||||
```bash
|
||||
root@arda-arc12:/# sycl-ls
|
||||
[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device 1.2 [2023.16.7.0.21_160000]
|
||||
[opencl:cpu:1] Intel(R) OpenCL, 13th Gen Intel(R) Core(TM) i9-13900K 3.0 [2023.16.7.0.21_160000]
|
||||
[opencl:gpu:2] Intel(R) OpenCL Graphics, Intel(R) Arc(TM) A770 Graphics 3.0 [23.17.26241.33]
|
||||
[ext_oneapi_level_zero:gpu:0] Intel(R) Level-Zero, Intel(R) Arc(TM) A770 Graphics 1.3 [1.3.26241]
|
||||
```
|
||||
|
||||
### 3. Serving with FastChat
|
||||
|
||||
To serve using the Web UI, you need three main components: web servers that interface with users, model workers that host one or more models, and a controller to coordinate the web server and model workers.
|
||||
|
||||
- #### **Step 1: Launch the Controller**
|
||||
```bash
|
||||
python3 -m fastchat.serve.controller &
|
||||
```
|
||||
|
||||
This controller manages the distributed workers.
|
||||
|
||||
- #### **Step 2: Launch the model worker(s)**
|
||||
|
||||
Using IPEX-LLM in FastChat does not impose any new limitations on model usage. Therefore, all Hugging Face Transformer models can be utilized in FastChat.
|
||||
```bash
|
||||
# Available low_bit format including sym_int4, sym_int8, fp16 etc.
|
||||
python3 -m ipex_llm.serving.fastchat.ipex_llm_worker --model-path /llm/models/Llama-2-7b-chat-hf/ --low-bit "sym_int4" --trust-remote-code --device "xpu" &
|
||||
```
|
||||
Wait until the process finishes loading the model and you see "Uvicorn running on ...". The model worker will register itself to the controller.
|
||||
|
||||
- #### **Step 3: Launch Gradio web server or RESTful API server**
|
||||
You can launch Gradio web server to serve your models using the web UI or launch RESTful API server to serve with HTTP.
|
||||
|
||||
- **Option 1: Serving with Web UI**
|
||||
```bash
|
||||
python3 -m fastchat.serve.gradio_web_server &
|
||||
```
|
||||
This is the user interface that users will interact with.
|
||||
|
||||
By following these steps, you will be able to serve your models using the web UI with `IPEX-LLM` as the backend. You can open your browser and chat with a model now.
|
||||
|
||||
- **Option 2: Serving with OpenAI-Compatible RESTful APIs**
|
||||
|
||||
Launch the RESTful API server
|
||||
|
||||
```bash
|
||||
python3 -m fastchat.serve.openai_api_server --host 0.0.0.0 --port 8000 &
|
||||
```
|
||||
|
||||
Use curl for testing, an example could be:
|
||||
|
||||
```bash
|
||||
curl -X POST -H "Content-Type: application/json" -d '{
|
||||
"model": "Llama-2-7b-chat-hf",
|
||||
"prompt": "Once upon a time, there existed a little girl who liked to have adventures. She wanted to go to places and meet new people, and have fun",
|
||||
"n": 1,
|
||||
"best_of": 1,
|
||||
"use_beam_search": false,
|
||||
"stream": false
|
||||
}' http://YOUR_HTTP_HOST:8000/v1/completions
|
||||
```
|
||||
You can find more details here [Serving using IPEX-LLM and FastChat](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/src/ipex_llm/serving/fastchat/README.md)
|
||||
|
||||
### 4. Serving with vLLM Continuous Batching
|
||||
To fully utilize the continuous batching feature of the vLLM, you can send requests to the service using curl or other similar methods. The requests sent to the engine will be batched at token level. Queries will be executed in the same forward step of the LLM and be removed when they are finished instead of waiting for all sequences to be finished.
|
||||
|
||||
- #### **Step 1: Launch the api_server**
|
||||
```bash
|
||||
#!/bin/bash
|
||||
# You may also want to adjust the `--max-num-batched-tokens` argument, it indicates the hard limit
|
||||
# of batched prompt length the server will accept
|
||||
python -m ipex_llm.vllm.entrypoints.openai.api_server \
|
||||
--model /llm/models/Llama-2-7b-chat-hf/ \
|
||||
--host 0.0.0.0 --port 8000 \
|
||||
--load-format 'auto' --device xpu --dtype bfloat16 \
|
||||
--max-num-batched-tokens 4096 &
|
||||
```
|
||||
|
||||
- #### **Step 2: Use curl for testing, access the api server as follows:**
|
||||
|
||||
```bash
|
||||
curl http://YOUR_HTTP_HOST:8000/v1/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"model": "/llm/models/Llama-2-7b-chat-hf/",
|
||||
"prompt": "San Francisco is a",
|
||||
"max_tokens": 128,
|
||||
"temperature": 0
|
||||
}' &
|
||||
```
|
||||
You can find more details here [Serving with vLLM Continuous Batching](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/example/GPU/vLLM-Serving/README.md)
|
||||
|
||||
## IPEX-LLM Fine Tuning on CPU
|
||||
|
||||
### 1. Prepare Docker Image
|
||||
|
||||
You can download directly from Dockerhub like:
|
||||
|
||||
```bash
|
||||
# For standalone
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
|
||||
# For k8s
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
Or build the image from source:
|
||||
|
||||
```bash
|
||||
# For standalone
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker build \
|
||||
--build-arg http_proxy=${HTTP_PROXY} \
|
||||
--build-arg https_proxy=${HTTPS_PROXY} \
|
||||
-t intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT \
|
||||
-f ./Dockerfile .
|
||||
|
||||
# For k8s
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker build \
|
||||
--build-arg http_proxy=${HTTP_PROXY} \
|
||||
--build-arg https_proxy=${HTTPS_PROXY} \
|
||||
-t intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT \
|
||||
-f ./Dockerfile.k8s .
|
||||
```
|
||||
|
||||
### 2. Prepare Base Model, Data and Container
|
||||
|
||||
Here, we try to fine-tune a [Llama2-7b](https://huggingface.co/meta-llama/Llama-2-7b) with [yahma/alpaca-cleaned](https://huggingface.co/datasets/yahma/alpaca-cleaned) dataset, and please download them and start a docker container with files mounted like below:
|
||||
|
||||
```bash
|
||||
export BASE_MODE_PATH=your_downloaded_base_model_path
|
||||
export DATA_PATH=your_downloaded_data_path
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--name=ipex-llm-fintune-qlora-cpu \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
-v $BASE_MODE_PATH:/ipex_llm/model \
|
||||
-v $DATA_PATH:/ipex_llm/data/alpaca-cleaned \
|
||||
intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
The download and mount of base model and data to a docker container demonstrates a standard fine-tuning process. You can skip this step for a quick start, and in this way, the fine-tuning codes will automatically download the needed files:
|
||||
|
||||
```bash
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--name=ipex-llm-fintune-qlora-cpu \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
However, we do recommend you to handle them manually, because the automatical download can be blocked by Internet access and Huggingface authentication etc. according to different environment, and the manual method allows you to fine-tune in a custom way (with different base model and dataset).
|
||||
|
||||
### 3. Start Fine-Tuning (Local Mode)
|
||||
|
||||
Enter the running container:
|
||||
|
||||
```bash
|
||||
docker exec -it ipex-llm-fintune-qlora-cpu bash
|
||||
```
|
||||
|
||||
Then, start QLoRA fine-tuning:
|
||||
If the machine memory is not enough, you can try to set `use_gradient_checkpointing=True`.
|
||||
|
||||
```bash
|
||||
cd /ipex_llm
|
||||
bash start-qlora-finetuning-on-cpu.sh
|
||||
```
|
||||
|
||||
After minutes, it is expected to get results like:
|
||||
|
||||
```bash
|
||||
{'loss': 2.0251, 'learning_rate': 0.0002, 'epoch': 0.02}
|
||||
{'loss': 1.2389, 'learning_rate': 0.00017777777777777779, 'epoch': 0.03}
|
||||
{'loss': 1.032, 'learning_rate': 0.00015555555555555556, 'epoch': 0.05}
|
||||
{'loss': 0.9141, 'learning_rate': 0.00013333333333333334, 'epoch': 0.06}
|
||||
{'loss': 0.8505, 'learning_rate': 0.00011111111111111112, 'epoch': 0.08}
|
||||
{'loss': 0.8713, 'learning_rate': 8.888888888888889e-05, 'epoch': 0.09}
|
||||
{'loss': 0.8635, 'learning_rate': 6.666666666666667e-05, 'epoch': 0.11}
|
||||
{'loss': 0.8853, 'learning_rate': 4.4444444444444447e-05, 'epoch': 0.12}
|
||||
{'loss': 0.859, 'learning_rate': 2.2222222222222223e-05, 'epoch': 0.14}
|
||||
{'loss': 0.8608, 'learning_rate': 0.0, 'epoch': 0.15}
|
||||
{'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15}
|
||||
100%|███████████████████████████████████████████████████████████████████████████████████| 200/200 [07:16<00:00, 2.18s/it]
|
||||
TrainOutput(global_step=200, training_loss=1.0400420665740966, metrics={'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15})
|
||||
```
|
||||
|
||||
### 4. Merge the adapter into the original model
|
||||
|
||||
Using the [export_merged_model.py](../../../../../../python/llm/example/GPU/LLM-Finetuning/QLoRA/export_merged_model.py) to merge.
|
||||
|
||||
```
|
||||
python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged
|
||||
```
|
||||
|
||||
Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference.
|
||||
|
||||
|
||||
## IPEX-LLM Fine Tuning on XPU
|
||||
|
||||
The following shows how to fine-tune LLM with Quantization (QLoRA built on IPEX-LLM 4bit optimizations) in a docker environment, which is accelerated by Intel XPU.
|
||||
|
||||
### 1. Prepare ipex-llm-finetune-qlora-xpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Prepare Base Model, Data and Start Docker Container
|
||||
|
||||
Here, we try to fine-tune a [Llama2-7b](https://huggingface.co/meta-llama/Llama-2-7b) with [yahma/alpaca-cleaned](https://huggingface.co/datasets/yahma/alpaca-cleaned) dataset, and please download them and start a docker container with files mounted like below:
|
||||
|
||||
```bash
|
||||
export BASE_MODE_PATH=your_downloaded_base_model_path
|
||||
export DATA_PATH=your_downloaded_data_path
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
export CONTAINER_NAME=my_container
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
-v $BASE_MODE_PATH:/model \
|
||||
-v $DATA_PATH:/data/alpaca-cleaned \
|
||||
--shm-size="16g" \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
|
||||
After the container is booted, you could get into the container through docker exec.
|
||||
|
||||
```bash
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Start Fine-Tuning (Local Mode)
|
||||
|
||||
Then, start QLoRA fine-tuning:
|
||||
|
||||
```bash
|
||||
bash start-qlora-finetuning-on-xpu.sh
|
||||
```
|
||||
|
||||
After minutes, it is expected to get results like:
|
||||
|
||||
```bash
|
||||
{'loss': 2.0251, 'learning_rate': 0.0002, 'epoch': 0.02}
|
||||
{'loss': 1.2389, 'learning_rate': 0.00017777777777777779, 'epoch': 0.03}
|
||||
{'loss': 1.032, 'learning_rate': 0.00015555555555555556, 'epoch': 0.05}
|
||||
{'loss': 0.9141, 'learning_rate': 0.00013333333333333334, 'epoch': 0.06}
|
||||
{'loss': 0.8505, 'learning_rate': 0.00011111111111111112, 'epoch': 0.08}
|
||||
{'loss': 0.8713, 'learning_rate': 8.888888888888889e-05, 'epoch': 0.09}
|
||||
{'loss': 0.8635, 'learning_rate': 6.666666666666667e-05, 'epoch': 0.11}
|
||||
{'loss': 0.8853, 'learning_rate': 4.4444444444444447e-05, 'epoch': 0.12}
|
||||
{'loss': 0.859, 'learning_rate': 2.2222222222222223e-05, 'epoch': 0.14}
|
||||
{'loss': 0.8608, 'learning_rate': 0.0, 'epoch': 0.15}
|
||||
{'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15}
|
||||
100%|███████████████████████████████████████████████████████████████████████████████████| 200/200 [07:16<00:00, 2.18s/it]
|
||||
TrainOutput(global_step=200, training_loss=1.0400420665740966, metrics={'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15})
|
||||
```
|
||||
|
||||
### 4. Merge the adapter into the original model
|
||||
|
||||
Using the [export_merged_model.py](../../python/llm/example/GPU/LLM-Finetuning/QLoRA/alpaca-qlora/export_merged_model.py) to merge.
|
||||
|
||||
```
|
||||
python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged
|
||||
```
|
||||
|
||||
Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference.
|
||||
# IPEX-LLM Docker Containers
|
||||
|
||||
You can run IPEX-LLM containers (via docker or k8s) for inference, serving and fine-tuning on Intel CPU and GPU. Details on how to use these containers are available at [IPEX-LLM Docker Container Guides](https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Docker/index.html).
|
||||
|
||||
### Prerequisites
|
||||
|
||||
- Docker on Windows or Linux
|
||||
- Windows Subsystem for Linux (WSL) is required if using Windows.
|
||||
|
||||
### Quick Start
|
||||
|
||||
|
||||
#### Pull a IPEX-LLM Docker Image
|
||||
To pull IPEX-LLM Docker images from [Intel Analytics Docker Hub](https://hub.docker.com/u/intelanalytics), use the `docker pull` command. For instance, to pull the CPU inference image:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
Available images in hub are:
|
||||
|
||||
| Image Name | Description |
|
||||
| --- | --- |
|
||||
| intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT | CPU Inference |
|
||||
| intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT | GPU Inference |
|
||||
| intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT | CPU Serving|
|
||||
| intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT | GPU Serving|
|
||||
| intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT | CPU Finetuning via Docker|
|
||||
|intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT|CPU Finetuning via Kubernetes|
|
||||
| intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT| GPU Finetuning|
|
||||
|
||||
#### Run a Container
|
||||
Use `docker run` command to run an IPEX-LLM docker container. For detailed instructions, refer to the [IPEX-LLM Docker Container Guides](https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Docker/index.html).
|
||||
|
||||
|
||||
#### Build Docker Image
|
||||
|
||||
To build a Docker image from source, first clone the IPEX-LLM repository and navigate to the Dockerfile directory. For example, to build the CPU inference image, navigate to `docker/llm/inference/cpu/docker`.
|
||||
|
||||
Then, use the following command to build the image (replace `your_image_name` with your desired image name):
|
||||
|
||||
```bash
|
||||
docker build \
|
||||
--build-arg no_proxy=localhost,127.0.0.1 \
|
||||
--rm --no-cache -t your_image_name .
|
||||
```
|
||||
|
||||
> Note: If you're working behind a proxy, also add args `--build-arg http_proxy=http://your_proxy_uri:port` and `--build-arg https_proxy=https://your_proxy_url:port`
|
||||
|
|
|
|||
664
docker/llm/README_backup.md
Normal file
664
docker/llm/README_backup.md
Normal file
|
|
@ -0,0 +1,664 @@
|
|||
# Getting started with IPEX-LLM in Docker
|
||||
|
||||
This guide provides step-by-step instructions for installing and using IPEX-LLM in a Docker environment. It covers setups for both CPU and XPU (accelerated processing units) on different operating systems.
|
||||
|
||||
### Index
|
||||
- [Docker Installation](#docker-installation-instructions)
|
||||
- [IPEX-LLM Inference](#ipex-llm-inference-on-cpu)
|
||||
- [On CPU](#ipex-llm-inference-on-cpu)
|
||||
- [On XPU](#ipex-llm-inference-on-xpu)
|
||||
- [IPEX-LLM Serving](#ipex-llm-serving-on-cpu)
|
||||
- [On CPU](#ipex-llm-serving-on-cpu)
|
||||
- [On XPU](#ipex-llm-serving-on-xpu)
|
||||
- [IPEX-LLM Fine Tuning](#ipex-llm-fine-tuning-on-cpu)
|
||||
- [On CPU](#ipex-llm-fine-tuning-on-cpu)
|
||||
- [On XPU](#ipex-llm-fine-tuning-on-xpu)
|
||||
|
||||
|
||||
## Docker Installation Instructions
|
||||
|
||||
1. **For New Users:**
|
||||
- Begin by visiting the [official Docker Get Started page](https://www.docker.com/get-started/) for a comprehensive introduction and installation guide.
|
||||
|
||||
2. **Additional Steps for Windows Users:**
|
||||
- For Windows Users, follow the step-by-step guide: [Docker Installation Instructions for Windows](https://github.com/intel-analytics/ipex-llm/blob/main/docs/readthedocs/source/doc/LLM/Quickstart/docker_windows_gpu.md).
|
||||
|
||||
|
||||
## IPEX-LLM Inference on CPU
|
||||
|
||||
### 1. Prepare ipex-llm-cpu Docker Image
|
||||
|
||||
Run the following command to pull image:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start bigdl-llm-cpu Docker Container
|
||||
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--privileged \
|
||||
-p 12345:12345 \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Start Inference and Tutorials
|
||||
**3.1 Chat Interface**: Use `chat.py` for conversational AI.
|
||||
|
||||
For example, if your model is Llama-2-7b-chat-hf and mounted on /llm/models, you can excute the following command to initiate a conversation:
|
||||
```bash
|
||||
cd /llm/portable-zip
|
||||
python chat.py --model-path /llm/models/Llama-2-7b-chat-hf
|
||||
```
|
||||
Here is a demostration:
|
||||
|
||||
<a align="left" href="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-chatpy-demo.gif">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-chatpy-demo.gif" width='60%' />
|
||||
|
||||
</a><br>
|
||||
|
||||
**3.2 Jupyter Lab Tutorials**: Start a Jupyter Lab session for IPEX-LLM tutorials.
|
||||
|
||||
Run the following command to start notebook:
|
||||
```bash
|
||||
cd /llm
|
||||
./start-notebook.sh [--port EXPECTED_PORT]
|
||||
```
|
||||
The default port is 12345, you could assign a different port by specifying the --port parameter.
|
||||
|
||||
If you're using the host network mode when booting the container, once the service is running successfully, you can access the tutorial at http://127.0.0.1:12345/lab. Alternatively, you need to ensure the correct ports are bound between the container and the host.
|
||||
|
||||
Here's a demonstration of how to navigate the tutorial in the explorer:
|
||||
|
||||
<a align="left" href="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-tutorial-demo.gif">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/llm-inference-cpu-docker-tutorial-demo.gif" width='60%' />
|
||||
|
||||
</a> <br>
|
||||
|
||||
**3.3 Performance Benchmark**: We provide a benchmark tool help users to test all the benchmarks and record them in a result CSV.
|
||||
|
||||
```bash
|
||||
cd /llm/benchmark/all-in-one
|
||||
```
|
||||
|
||||
Users can provide models and related information in config.yaml.
|
||||
```bash
|
||||
repo_id:
|
||||
# - 'THUDM/chatglm-6b'
|
||||
# - 'THUDM/chatglm2-6b'
|
||||
- 'meta-llama/Llama-2-7b-chat-hf'
|
||||
# - 'liuhaotian/llava-v1.5-7b' # requires a LLAVA_REPO_DIR env variables pointing to the llava dir; added only for gpu win related test_api now
|
||||
local_model_hub: 'path to your local model hub'
|
||||
warm_up: 1
|
||||
num_trials: 3
|
||||
num_beams: 1 # default to greedy search
|
||||
low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
|
||||
batch_size: 1 # default to 1
|
||||
in_out_pairs:
|
||||
- '32-32'
|
||||
- '1024-128'
|
||||
test_api:
|
||||
# - "transformer_int4"
|
||||
# - "native_int4"
|
||||
# - "optimize_model"
|
||||
# - "pytorch_autocast_bf16"
|
||||
# - "transformer_autocast_bf16"
|
||||
# - "bigdl_ipex_bf16"
|
||||
# - "bigdl_ipex_int4"
|
||||
# - "bigdl_ipex_int8"
|
||||
# - "ipex_fp16_gpu" # on Intel GPU
|
||||
# - "bigdl_fp16_gpu" # on Intel GPU
|
||||
# - "transformer_int4_gpu" # on Intel GPU
|
||||
# - "optimize_model_gpu" # on Intel GPU
|
||||
# - "deepspeed_transformer_int4_cpu" # on Intel SPR Server
|
||||
# - "transformer_int4_gpu_win" # on Intel GPU for Windows
|
||||
# - "transformer_int4_fp16_gpu_win" # on Intel GPU for Windows, use fp16 for non-linear layer
|
||||
# - "transformer_int4_loadlowbit_gpu_win" # on Intel GPU for Windows using load_low_bit API. Please make sure you have used the save.py to save the converted low bit model
|
||||
# - "deepspeed_optimize_model_gpu" # deepspeed autotp on Intel GPU
|
||||
- "speculative_cpu"
|
||||
# - "speculative_gpu"
|
||||
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)
|
||||
streaming: False # whether output in streaming way (only avaiable now for gpu win related test_api)
|
||||
```
|
||||
|
||||
This benchmark tool offers various test APIs, including `transformer_int4`, `speculative_cpu`, and more.
|
||||
|
||||
For instance, if you wish to benchmark **inference with speculative decoding**, utilize the `speculative_cpu` test API in the `config.yml` file.
|
||||
|
||||
Then, execute `bash run-spr.sh`, which will generate output results in `results.csv`.
|
||||
```bash
|
||||
bash run-spr.sh
|
||||
```
|
||||
|
||||
For further details and comprehensive functionality of the benchmark tool, please refer to the [all-in-one benchmark tool](https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/dev/benchmark/all-in-one).
|
||||
|
||||
Additionally, for examples related to Inference with Speculative Decoding, you can explore [Speculative-Decoding Examples](https://github.com/intel-analytics/ipex-llm/tree/main/python/llm/example/CPU/Speculative-Decoding).
|
||||
|
||||
|
||||
|
||||
## IPEX-LLM Inference on XPU
|
||||
|
||||
### 1. Prepare ipex-llm-xpu Docker Image
|
||||
|
||||
Run the following command to pull image from dockerhub:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start Chat Inference
|
||||
|
||||
We provide `chat.py` for conversational AI. If your model is Llama-2-7b-chat-hf and mounted on /llm/models, you can execute the following command to initiate a conversation:
|
||||
|
||||
To map the xpu into the container, you need to specify --device=/dev/dri when booting the container.
|
||||
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE bash -c "python chat.py --model-path /llm/models/Llama-2-7b-chat-hf"
|
||||
```
|
||||
|
||||
|
||||
### 3. Quick Performance Benchmark
|
||||
|
||||
Execute a quick performance benchmark by starting the ipex-llm-xpu container, specifying the model, test API, and device, then running the benchmark.sh script.
|
||||
|
||||
To map the XPU into the container, specify `--device=/dev/dri` when booting the container.
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models [change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
-e REPO_IDS="meta-llama/Llama-2-7b-chat-hf" \
|
||||
-e TEST_APIS="transformer_int4_gpu" \
|
||||
-e DEVICE=Arc \
|
||||
$DOCKER_IMAGE /llm/benchmark.sh
|
||||
```
|
||||
|
||||
Customize environment variables to specify:
|
||||
|
||||
- **REPO_IDS:** Model's name and organization, separated by commas if multiple values exist.
|
||||
- **TEST_APIS:** Different test functions based on the machine, separated by commas if multiple values exist.
|
||||
- **DEVICE:** Type of device - Max, Flex, Arc.
|
||||
|
||||
**Result**
|
||||
|
||||
Upon completion, you can obtain a CSV result file, the content of CSV results will be printed out. You can mainly look at the results of columns `1st token avg latency (ms)` and `2+ avg latency (ms/token)` for the benchmark results.
|
||||
|
||||
## IPEX-LLM Serving on CPU
|
||||
FastChat is an open platform for training, serving, and evaluating large language model based chatbots. You can find the detailed information at their [homepage](https://github.com/lm-sys/FastChat).
|
||||
|
||||
IPEX-LLM is integrated into FastChat so that user can use IPEX-LLM as a serving backend in the deployment.
|
||||
|
||||
### 1. Prepare ipex-llm-serving-cpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start ipex-llm-serving-cpu Docker Container
|
||||
|
||||
Please be noted that the CPU config is specified for Xeon CPUs, change it accordingly if you are not using a Xeon CPU.
|
||||
|
||||
```bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Serving with FastChat
|
||||
|
||||
To serve using the Web UI, you need three main components: web servers that interface with users, model workers that host one or more models, and a controller to coordinate the web server and model workers.
|
||||
|
||||
- #### **Step 1: Launch the Controller**
|
||||
```bash
|
||||
python3 -m fastchat.serve.controller &
|
||||
```
|
||||
|
||||
This controller manages the distributed workers.
|
||||
|
||||
- #### **Step 2: Launch the model worker(s)**
|
||||
|
||||
Using IPEX-LLM in FastChat does not impose any new limitations on model usage. Therefore, all Hugging Face Transformer models can be utilized in FastChat.
|
||||
```bash
|
||||
source ipex-llm-init -t
|
||||
|
||||
# Available low_bit format including sym_int4, sym_int8, bf16 etc.
|
||||
python3 -m ipex_llm.serving.fastchat.ipex_llm_worker --model-path path/to/vicuna-7b-v1.5 --low-bit "sym_int4" --trust-remote-code --device "cpu" &
|
||||
```
|
||||
Wait until the process finishes loading the model and you see "Uvicorn running on ...". The model worker will register itself to the controller.
|
||||
|
||||
- #### **Step 3: Launch Gradio web server or RESTful API server**
|
||||
You can launch Gradio web server to serve your models using the web UI or launch RESTful API server to serve with HTTP.
|
||||
|
||||
- **Option 1: Serving with Web UI**
|
||||
```bash
|
||||
python3 -m fastchat.serve.gradio_web_server &
|
||||
```
|
||||
This is the user interface that users will interact with.
|
||||
|
||||
By following these steps, you will be able to serve your models using the web UI with `IPEX-LLM` as the backend. You can open your browser and chat with a model now.
|
||||
|
||||
- **Option 2: Serving with OpenAI-Compatible RESTful APIs**
|
||||
|
||||
Launch the RESTful API server
|
||||
|
||||
```bash
|
||||
python3 -m fastchat.serve.openai_api_server --host 0.0.0.0 --port 8000 &
|
||||
```
|
||||
|
||||
Use curl for testing, an example could be:
|
||||
|
||||
```bash
|
||||
curl -X POST -H "Content-Type: application/json" -d '{
|
||||
"model": "Llama-2-7b-chat-hf",
|
||||
"prompt": "Once upon a time, there existed a little girl who liked to have adventures. She wanted to go to places and meet new people, and have fun",
|
||||
"n": 1,
|
||||
"best_of": 1,
|
||||
"use_beam_search": false,
|
||||
"stream": false
|
||||
}' http://YOUR_HTTP_HOST:8000/v1/completions
|
||||
```
|
||||
You can find more details here [Serving using IPEX-LLM and FastChat](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/src/ipex_llm/serving/fastchat/README.md)
|
||||
|
||||
### 4. Serving with vLLM Continuous Batching
|
||||
To fully utilize the continuous batching feature of the vLLM, you can send requests to the service using curl or other similar methods. The requests sent to the engine will be batched at token level. Queries will be executed in the same forward step of the LLM and be removed when they are finished instead of waiting for all sequences to be finished.
|
||||
|
||||
- #### **Step 1: Launch the api_server**
|
||||
```bash
|
||||
#!/bin/bash
|
||||
# You may also want to adjust the `--max-num-batched-tokens` argument, it indicates the hard limit
|
||||
# of batched prompt length the server will accept
|
||||
numactl -C 0-47 -m 0 python -m ipex_llm.vllm.entrypoints.openai.api_server \
|
||||
--model /llm/models/Llama-2-7b-chat-hf/ \
|
||||
--host 0.0.0.0 --port 8000 \
|
||||
--load-format 'auto' --device cpu --dtype bfloat16 \
|
||||
--max-num-batched-tokens 4096 &
|
||||
```
|
||||
|
||||
- #### **Step 2: Use curl for testing, access the api server as follows:**
|
||||
|
||||
```bash
|
||||
curl http://YOUR_HTTP_HOST:8000/v1/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"model": "/llm/models/Llama-2-7b-chat-hf/",
|
||||
"prompt": "San Francisco is a",
|
||||
"max_tokens": 128,
|
||||
"temperature": 0
|
||||
}' &
|
||||
```
|
||||
|
||||
You can find more details here: [Serving with vLLM Continuous Batching](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/example/CPU/vLLM-Serving/README.md)
|
||||
|
||||
|
||||
## IPEX-LLM Serving on XPU
|
||||
|
||||
FastChat is an open platform for training, serving, and evaluating large language model based chatbots. You can find the detailed information at their [homepage](https://github.com/lm-sys/FastChat).
|
||||
|
||||
IPEX-LLM is integrated into FastChat so that user can use IPEX-LLM as a serving backend in the deployment.
|
||||
|
||||
### 1. Prepare ipex-llm-serving-xpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Start ipex-llm-serving-xpu Docker Container
|
||||
|
||||
To map the `xpu` into the container, you need to specify `--device=/dev/dri` when booting the container.
|
||||
|
||||
```bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--cpuset-cpus="0-47" \
|
||||
--cpuset-mems="0" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
To verify the device is successfully mapped into the container, run `sycl-ls` to check the result. In a machine with Arc A770, the sampled output is:
|
||||
|
||||
```bash
|
||||
root@arda-arc12:/# sycl-ls
|
||||
[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device 1.2 [2023.16.7.0.21_160000]
|
||||
[opencl:cpu:1] Intel(R) OpenCL, 13th Gen Intel(R) Core(TM) i9-13900K 3.0 [2023.16.7.0.21_160000]
|
||||
[opencl:gpu:2] Intel(R) OpenCL Graphics, Intel(R) Arc(TM) A770 Graphics 3.0 [23.17.26241.33]
|
||||
[ext_oneapi_level_zero:gpu:0] Intel(R) Level-Zero, Intel(R) Arc(TM) A770 Graphics 1.3 [1.3.26241]
|
||||
```
|
||||
|
||||
### 3. Serving with FastChat
|
||||
|
||||
To serve using the Web UI, you need three main components: web servers that interface with users, model workers that host one or more models, and a controller to coordinate the web server and model workers.
|
||||
|
||||
- #### **Step 1: Launch the Controller**
|
||||
```bash
|
||||
python3 -m fastchat.serve.controller &
|
||||
```
|
||||
|
||||
This controller manages the distributed workers.
|
||||
|
||||
- #### **Step 2: Launch the model worker(s)**
|
||||
|
||||
Using IPEX-LLM in FastChat does not impose any new limitations on model usage. Therefore, all Hugging Face Transformer models can be utilized in FastChat.
|
||||
```bash
|
||||
# Available low_bit format including sym_int4, sym_int8, fp16 etc.
|
||||
python3 -m ipex_llm.serving.fastchat.ipex_llm_worker --model-path /llm/models/Llama-2-7b-chat-hf/ --low-bit "sym_int4" --trust-remote-code --device "xpu" &
|
||||
```
|
||||
Wait until the process finishes loading the model and you see "Uvicorn running on ...". The model worker will register itself to the controller.
|
||||
|
||||
- #### **Step 3: Launch Gradio web server or RESTful API server**
|
||||
You can launch Gradio web server to serve your models using the web UI or launch RESTful API server to serve with HTTP.
|
||||
|
||||
- **Option 1: Serving with Web UI**
|
||||
```bash
|
||||
python3 -m fastchat.serve.gradio_web_server &
|
||||
```
|
||||
This is the user interface that users will interact with.
|
||||
|
||||
By following these steps, you will be able to serve your models using the web UI with `IPEX-LLM` as the backend. You can open your browser and chat with a model now.
|
||||
|
||||
- **Option 2: Serving with OpenAI-Compatible RESTful APIs**
|
||||
|
||||
Launch the RESTful API server
|
||||
|
||||
```bash
|
||||
python3 -m fastchat.serve.openai_api_server --host 0.0.0.0 --port 8000 &
|
||||
```
|
||||
|
||||
Use curl for testing, an example could be:
|
||||
|
||||
```bash
|
||||
curl -X POST -H "Content-Type: application/json" -d '{
|
||||
"model": "Llama-2-7b-chat-hf",
|
||||
"prompt": "Once upon a time, there existed a little girl who liked to have adventures. She wanted to go to places and meet new people, and have fun",
|
||||
"n": 1,
|
||||
"best_of": 1,
|
||||
"use_beam_search": false,
|
||||
"stream": false
|
||||
}' http://YOUR_HTTP_HOST:8000/v1/completions
|
||||
```
|
||||
You can find more details here [Serving using IPEX-LLM and FastChat](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/src/ipex_llm/serving/fastchat/README.md)
|
||||
|
||||
### 4. Serving with vLLM Continuous Batching
|
||||
To fully utilize the continuous batching feature of the vLLM, you can send requests to the service using curl or other similar methods. The requests sent to the engine will be batched at token level. Queries will be executed in the same forward step of the LLM and be removed when they are finished instead of waiting for all sequences to be finished.
|
||||
|
||||
- #### **Step 1: Launch the api_server**
|
||||
```bash
|
||||
#!/bin/bash
|
||||
# You may also want to adjust the `--max-num-batched-tokens` argument, it indicates the hard limit
|
||||
# of batched prompt length the server will accept
|
||||
python -m ipex_llm.vllm.entrypoints.openai.api_server \
|
||||
--model /llm/models/Llama-2-7b-chat-hf/ \
|
||||
--host 0.0.0.0 --port 8000 \
|
||||
--load-format 'auto' --device xpu --dtype bfloat16 \
|
||||
--max-num-batched-tokens 4096 &
|
||||
```
|
||||
|
||||
- #### **Step 2: Use curl for testing, access the api server as follows:**
|
||||
|
||||
```bash
|
||||
curl http://YOUR_HTTP_HOST:8000/v1/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"model": "/llm/models/Llama-2-7b-chat-hf/",
|
||||
"prompt": "San Francisco is a",
|
||||
"max_tokens": 128,
|
||||
"temperature": 0
|
||||
}' &
|
||||
```
|
||||
You can find more details here [Serving with vLLM Continuous Batching](https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/example/GPU/vLLM-Serving/README.md)
|
||||
|
||||
## IPEX-LLM Fine Tuning on CPU
|
||||
|
||||
### 1. Prepare Docker Image
|
||||
|
||||
You can download directly from Dockerhub like:
|
||||
|
||||
```bash
|
||||
# For standalone
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
|
||||
# For k8s
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
Or build the image from source:
|
||||
|
||||
```bash
|
||||
# For standalone
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker build \
|
||||
--build-arg http_proxy=${HTTP_PROXY} \
|
||||
--build-arg https_proxy=${HTTPS_PROXY} \
|
||||
-t intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT \
|
||||
-f ./Dockerfile .
|
||||
|
||||
# For k8s
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker build \
|
||||
--build-arg http_proxy=${HTTP_PROXY} \
|
||||
--build-arg https_proxy=${HTTPS_PROXY} \
|
||||
-t intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT \
|
||||
-f ./Dockerfile.k8s .
|
||||
```
|
||||
|
||||
### 2. Prepare Base Model, Data and Container
|
||||
|
||||
Here, we try to fine-tune a [Llama2-7b](https://huggingface.co/meta-llama/Llama-2-7b) with [yahma/alpaca-cleaned](https://huggingface.co/datasets/yahma/alpaca-cleaned) dataset, and please download them and start a docker container with files mounted like below:
|
||||
|
||||
```bash
|
||||
export BASE_MODE_PATH=your_downloaded_base_model_path
|
||||
export DATA_PATH=your_downloaded_data_path
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--name=ipex-llm-fintune-qlora-cpu \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
-v $BASE_MODE_PATH:/ipex_llm/model \
|
||||
-v $DATA_PATH:/ipex_llm/data/alpaca-cleaned \
|
||||
intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
The download and mount of base model and data to a docker container demonstrates a standard fine-tuning process. You can skip this step for a quick start, and in this way, the fine-tuning codes will automatically download the needed files:
|
||||
|
||||
```bash
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--name=ipex-llm-fintune-qlora-cpu \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
However, we do recommend you to handle them manually, because the automatical download can be blocked by Internet access and Huggingface authentication etc. according to different environment, and the manual method allows you to fine-tune in a custom way (with different base model and dataset).
|
||||
|
||||
### 3. Start Fine-Tuning (Local Mode)
|
||||
|
||||
Enter the running container:
|
||||
|
||||
```bash
|
||||
docker exec -it ipex-llm-fintune-qlora-cpu bash
|
||||
```
|
||||
|
||||
Then, start QLoRA fine-tuning:
|
||||
If the machine memory is not enough, you can try to set `use_gradient_checkpointing=True`.
|
||||
|
||||
```bash
|
||||
cd /ipex_llm
|
||||
bash start-qlora-finetuning-on-cpu.sh
|
||||
```
|
||||
|
||||
After minutes, it is expected to get results like:
|
||||
|
||||
```bash
|
||||
{'loss': 2.0251, 'learning_rate': 0.0002, 'epoch': 0.02}
|
||||
{'loss': 1.2389, 'learning_rate': 0.00017777777777777779, 'epoch': 0.03}
|
||||
{'loss': 1.032, 'learning_rate': 0.00015555555555555556, 'epoch': 0.05}
|
||||
{'loss': 0.9141, 'learning_rate': 0.00013333333333333334, 'epoch': 0.06}
|
||||
{'loss': 0.8505, 'learning_rate': 0.00011111111111111112, 'epoch': 0.08}
|
||||
{'loss': 0.8713, 'learning_rate': 8.888888888888889e-05, 'epoch': 0.09}
|
||||
{'loss': 0.8635, 'learning_rate': 6.666666666666667e-05, 'epoch': 0.11}
|
||||
{'loss': 0.8853, 'learning_rate': 4.4444444444444447e-05, 'epoch': 0.12}
|
||||
{'loss': 0.859, 'learning_rate': 2.2222222222222223e-05, 'epoch': 0.14}
|
||||
{'loss': 0.8608, 'learning_rate': 0.0, 'epoch': 0.15}
|
||||
{'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15}
|
||||
100%|███████████████████████████████████████████████████████████████████████████████████| 200/200 [07:16<00:00, 2.18s/it]
|
||||
TrainOutput(global_step=200, training_loss=1.0400420665740966, metrics={'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15})
|
||||
```
|
||||
|
||||
### 4. Merge the adapter into the original model
|
||||
|
||||
Using the [export_merged_model.py](../../../../../../python/llm/example/GPU/LLM-Finetuning/QLoRA/export_merged_model.py) to merge.
|
||||
|
||||
```
|
||||
python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged
|
||||
```
|
||||
|
||||
Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference.
|
||||
|
||||
|
||||
## IPEX-LLM Fine Tuning on XPU
|
||||
|
||||
The following shows how to fine-tune LLM with Quantization (QLoRA built on IPEX-LLM 4bit optimizations) in a docker environment, which is accelerated by Intel XPU.
|
||||
|
||||
### 1. Prepare ipex-llm-finetune-qlora-xpu Docker Image
|
||||
|
||||
Run the following command:
|
||||
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
|
||||
### 2. Prepare Base Model, Data and Start Docker Container
|
||||
|
||||
Here, we try to fine-tune a [Llama2-7b](https://huggingface.co/meta-llama/Llama-2-7b) with [yahma/alpaca-cleaned](https://huggingface.co/datasets/yahma/alpaca-cleaned) dataset, and please download them and start a docker container with files mounted like below:
|
||||
|
||||
```bash
|
||||
export BASE_MODE_PATH=your_downloaded_base_model_path
|
||||
export DATA_PATH=your_downloaded_data_path
|
||||
export HTTP_PROXY=your_http_proxy
|
||||
export HTTPS_PROXY=your_https_proxy
|
||||
export CONTAINER_NAME=my_container
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
-e http_proxy=${HTTP_PROXY} \
|
||||
-e https_proxy=${HTTPS_PROXY} \
|
||||
-v $BASE_MODE_PATH:/model \
|
||||
-v $DATA_PATH:/data/alpaca-cleaned \
|
||||
--shm-size="16g" \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
|
||||
After the container is booted, you could get into the container through docker exec.
|
||||
|
||||
```bash
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
### 3. Start Fine-Tuning (Local Mode)
|
||||
|
||||
Then, start QLoRA fine-tuning:
|
||||
|
||||
```bash
|
||||
bash start-qlora-finetuning-on-xpu.sh
|
||||
```
|
||||
|
||||
After minutes, it is expected to get results like:
|
||||
|
||||
```bash
|
||||
{'loss': 2.0251, 'learning_rate': 0.0002, 'epoch': 0.02}
|
||||
{'loss': 1.2389, 'learning_rate': 0.00017777777777777779, 'epoch': 0.03}
|
||||
{'loss': 1.032, 'learning_rate': 0.00015555555555555556, 'epoch': 0.05}
|
||||
{'loss': 0.9141, 'learning_rate': 0.00013333333333333334, 'epoch': 0.06}
|
||||
{'loss': 0.8505, 'learning_rate': 0.00011111111111111112, 'epoch': 0.08}
|
||||
{'loss': 0.8713, 'learning_rate': 8.888888888888889e-05, 'epoch': 0.09}
|
||||
{'loss': 0.8635, 'learning_rate': 6.666666666666667e-05, 'epoch': 0.11}
|
||||
{'loss': 0.8853, 'learning_rate': 4.4444444444444447e-05, 'epoch': 0.12}
|
||||
{'loss': 0.859, 'learning_rate': 2.2222222222222223e-05, 'epoch': 0.14}
|
||||
{'loss': 0.8608, 'learning_rate': 0.0, 'epoch': 0.15}
|
||||
{'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15}
|
||||
100%|███████████████████████████████████████████████████████████████████████████████████| 200/200 [07:16<00:00, 2.18s/it]
|
||||
TrainOutput(global_step=200, training_loss=1.0400420665740966, metrics={'train_runtime': xxxx, 'train_samples_per_second': xxxx, 'train_steps_per_second': xxxx, 'train_loss': 1.0400420665740966, 'epoch': 0.15})
|
||||
```
|
||||
|
||||
### 4. Merge the adapter into the original model
|
||||
|
||||
Using the [export_merged_model.py](../../python/llm/example/GPU/LLM-Finetuning/QLoRA/alpaca-qlora/export_merged_model.py) to merge.
|
||||
|
||||
```
|
||||
python ./export_merged_model.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --adapter_path ./outputs/checkpoint-200 --output_path ./outputs/checkpoint-200-merged
|
||||
```
|
||||
|
||||
Then you can use `./outputs/checkpoint-200-merged` as a normal huggingface transformer model to do inference.
|
||||
|
|
@ -25,12 +25,6 @@
|
|||
<li>
|
||||
<a href="doc/LLM/Quickstart/install_windows_gpu.html">Install IPEX-LLM on Windows with Intel GPU</a>
|
||||
</li>
|
||||
<li>
|
||||
<a href="doc/LLM/Quickstart/docker_windows_gpu.html">Install IPEX-LLM in Docker on Windows with Intel GPU</a>
|
||||
</li>
|
||||
<li>
|
||||
<a href="doc/LLM/Quickstart/docker_pytorch_inference_gpu.html">Run PyTorch Inference on Intel GPU using Docker (on Linux or WSL)</a>
|
||||
</li>
|
||||
<li>
|
||||
<a href="doc/LLM/Quickstart/chatchat_quickstart.html">Run Local RAG using Langchain-Chatchat on Intel GPU</a>
|
||||
</li>
|
||||
|
|
@ -73,6 +67,21 @@
|
|||
</li>
|
||||
</ul>
|
||||
</li>
|
||||
<li>
|
||||
<strong class="bigdl-quicklinks-section-title">IPEX-LLM Docker Guides</strong>
|
||||
<input id="quicklink-cluster-llm-docker" type="checkbox" class="toctree-checkbox" />
|
||||
<label for="quicklink-cluster-llm-docker" class="toctree-toggle">
|
||||
<i class="fa-solid fa-chevron-down"></i>
|
||||
</label>
|
||||
<ul class="bigdl-quicklinks-section-nav">
|
||||
<li>
|
||||
<a href="doc/LLM/Docker/docker_windows_gpu.html">Overview of IPEX-LLM Containers for Intel GPU</a>
|
||||
</li>
|
||||
<li>
|
||||
<a href="doc/LLM/Docker/docker_pytorch_inference_gpu.html">Run PyTorch Inference on an Intel GPU via Docker</a>
|
||||
</li>
|
||||
</ul>
|
||||
</li>
|
||||
<li>
|
||||
<strong class="bigdl-quicklinks-section-title">IPEX-LLM Installation</strong>
|
||||
<input id="quicklink-cluster-llm-installation" type="checkbox" class="toctree-checkbox" />
|
||||
|
|
|
|||
|
|
@ -15,6 +15,12 @@ subtrees:
|
|||
title: "CPU"
|
||||
- file: doc/LLM/Overview/install_gpu
|
||||
title: "GPU"
|
||||
- file: doc/LLM/Docker/index
|
||||
title: "Docker Guides"
|
||||
subtrees:
|
||||
- entries:
|
||||
- file: doc/LLM/Docker/docker_windows_gpu
|
||||
- file: doc/LLM/Docker/docker_pytorch_inference_gpu
|
||||
- file: doc/LLM/Quickstart/index
|
||||
title: "Quickstart"
|
||||
subtrees:
|
||||
|
|
@ -22,7 +28,6 @@ subtrees:
|
|||
- file: doc/LLM/Quickstart/bigdl_llm_migration
|
||||
- file: doc/LLM/Quickstart/install_linux_gpu
|
||||
- file: doc/LLM/Quickstart/install_windows_gpu
|
||||
- file: doc/LLM/Quickstart/docker_windows_gpu
|
||||
- file: doc/LLM/Quickstart/chatchat_quickstart
|
||||
- file: doc/LLM/Quickstart/webui_quickstart
|
||||
- file: doc/LLM/Quickstart/open_webui_with_ollama_quickstart
|
||||
|
|
|
|||
|
|
@ -1,16 +1,10 @@
|
|||
# Run PyTorch Inference on Intel GPU using Docker (on Linux or WSL)
|
||||
# Run PyTorch Inference on an Intel GPU via Docker
|
||||
|
||||
We can run PyTorch Inference Benchmark, Chat Service and PyTorch Examples on Intel GPUs within Docker (on Linux or WSL).
|
||||
|
||||
## Install Docker
|
||||
|
||||
1. Linux Installation
|
||||
|
||||
Follow the instructions in this [guide](https://www.docker.com/get-started/) to install Docker on Linux.
|
||||
|
||||
2. Windows Installation
|
||||
|
||||
For Windows installation, refer to this [guide](https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Quickstart/docker_windows_gpu.html#install-docker-on-windows).
|
||||
Follow the [Docker installation Guide](./docker_windows_gpu.html#install-docker) to install docker on either Linux or Windows.
|
||||
|
||||
## Launch Docker
|
||||
|
||||
|
|
@ -20,26 +14,52 @@ docker pull intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
|||
```
|
||||
|
||||
Start ipex-llm-xpu Docker Container:
|
||||
```bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
```eval_rst
|
||||
.. tabs::
|
||||
.. tab:: Linux
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
docker run -itd \
|
||||
--net=host \
|
||||
--device=/dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/models \
|
||||
$DOCKER_IMAGE
|
||||
|
||||
.. tab:: Windows WSL
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--privileged \
|
||||
--device /dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/llm-models \
|
||||
-v /usr/lib/wsl:/usr/lib/wsl \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
|
||||
|
||||
Access the container:
|
||||
```
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
|
||||
To verify the device is successfully mapped into the container, run `sycl-ls` to check the result. In a machine with Arc A770, the sampled output is:
|
||||
|
||||
```bash
|
||||
115
docs/readthedocs/source/doc/LLM/Docker/docker_windows_gpu.md
Normal file
115
docs/readthedocs/source/doc/LLM/Docker/docker_windows_gpu.md
Normal file
|
|
@ -0,0 +1,115 @@
|
|||
# Overview of IPEX-LLM Containers for Intel GPU
|
||||
|
||||
|
||||
An IPEX-LLM container is a pre-configured environment that includes all necessary dependencies for running LLMs on Intel GPUs.
|
||||
|
||||
This guide provides general instructions for setting up the IPEX-LLM Docker containers with Intel GPU. It begins with instructions and tips for Docker installation, and then introduce the available IPEX-LLM containers and their uses.
|
||||
|
||||
## Install Docker
|
||||
|
||||
### Linux
|
||||
|
||||
Follow the instructions in the [Offcial Docker Guide](https://www.docker.com/get-started/) to install Docker on Linux.
|
||||
|
||||
|
||||
### Windows
|
||||
|
||||
```eval_rst
|
||||
.. tip::
|
||||
|
||||
The installation requires at least 35GB of free disk space on C drive.
|
||||
|
||||
```
|
||||
```eval_rst
|
||||
.. note::
|
||||
|
||||
Detailed installation instructions for Windows, including steps for enabling WSL2, can be found on the [Docker Desktop for Windows installation page](https://docs.docker.com/desktop/install/windows-install/).
|
||||
|
||||
```
|
||||
|
||||
#### Install Docker Desktop for Windows
|
||||
Follow the instructions in [this guide](https://docs.docker.com/desktop/install/windows-install/) to install **Docker Desktop for Windows**. Restart you machine after the installation is complete.
|
||||
|
||||
#### Install WSL2
|
||||
|
||||
Follow the instructions in [this guide](https://docs.microsoft.com/en-us/windows/wsl/install) to install **Windows Subsystem for Linux 2 (WSL2)**.
|
||||
|
||||
```eval_rst
|
||||
.. tip::
|
||||
|
||||
You may verify WSL2 installation by running the command `wsl --list` in PowerShell or Command Prompt. If WSL2 is installed, you will see a list of installed Linux distributions.
|
||||
```
|
||||
|
||||
#### Enable Docker integration with WSL2
|
||||
|
||||
Open **Docker desktop**, and select `Settings`->`Resources`->`WSL integration`->turn on `Ubuntu` button->`Apply & restart`.
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_new.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_new.png" width=100%; />
|
||||
</a>
|
||||
|
||||
```eval_rst
|
||||
.. tip::
|
||||
|
||||
If you encounter **Docker Engine stopped** when opening Docker Desktop, you can reopen it in administrator mode.
|
||||
```
|
||||
|
||||
#### Verify Docker is enabled in WSL2
|
||||
|
||||
Execute the following commands in PowerShell or Command Prompt to verify that Docker is enabled in WSL2:
|
||||
```bash
|
||||
wsl -d Ubuntu # Run Ubuntu WSL distribution
|
||||
docker version # Check if Docker is enabled in WSL
|
||||
```
|
||||
|
||||
You can see the output similar to the following:
|
||||
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_wsl.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_wsl.png" width=100%; />
|
||||
</a>
|
||||
|
||||
```eval_rst
|
||||
.. tip::
|
||||
|
||||
During the use of Docker in WSL, Docker Desktop needs to be kept open all the time.
|
||||
```
|
||||
|
||||
|
||||
## IPEX-LLM Docker Containers
|
||||
|
||||
We have several docker images available for running LLMs on Intel GPUs. The following table lists the available images and their uses:
|
||||
|
||||
| Image Name | Description | Use Case |
|
||||
|------------|-------------|----------|
|
||||
| intelanalytics/ipex-llm-cpu:2.1.0-SNAPSHOT | CPU Inference |For development and running LLMs using llama.cpp, Ollama and Python|
|
||||
| intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT | GPU Inference |For development and running LLMs using llama.cpp, Ollama and Python|
|
||||
| intelanalytics/ipex-llm-serving-cpu:2.1.0-SNAPSHOT | CPU Serving|For serving multiple users/requests through REST APIs using vLLM/FastChat|
|
||||
| intelanalytics/ipex-llm-serving-xpu:2.1.0-SNAPSHOT | GPU Serving|For serving multiple users/requests through REST APIs using vLLM/FastChat|
|
||||
| intelanalytics/ipex-llm-finetune-qlora-cpu-standalone:2.1.0-SNAPSHOT | CPU Finetuning via Docker|For fine-tuning LLMs using QLora/Lora, etc. |
|
||||
|intelanalytics/ipex-llm-finetune-qlora-cpu-k8s:2.1.0-SNAPSHOT|CPU Finetuning via Kubernetes|For fine-tuning LLMs using QLora/Lora, etc. |
|
||||
| intelanalytics/ipex-llm-finetune-qlora-xpu:2.1.0-SNAPSHOT| GPU Finetuning|For fine-tuning LLMs using QLora/Lora, etc.|
|
||||
|
||||
We have also provided several quickstarts for various usage scenarios:
|
||||
- [Run and develop LLM applications in PyTorch](./docker_pytorch_inference_gpu.html)
|
||||
|
||||
... to be added soon.
|
||||
|
||||
## Troubleshooting
|
||||
|
||||
|
||||
If your machine has both an integrated GPU (iGPU) and a dedicated GPU (dGPU) like ARC, you may encounter the following issue:
|
||||
|
||||
```bash
|
||||
Abort was called at 62 line in file:
|
||||
./shared/source/os_interface/os_interface.h
|
||||
LIBXSMM_VERSION: main_stable-1.17-3651 (25693763)
|
||||
LIBXSMM_TARGET: adl [Intel(R) Core(TM) i7-14700K]
|
||||
Registry and code: 13 MB
|
||||
Command: python chat.py --model-path /llm/llm-models/chatglm2-6b/
|
||||
Uptime: 29.349235 s
|
||||
Aborted
|
||||
```
|
||||
To resolve this problem, you can disabling the iGPU in Device Manager on Windows as follows:
|
||||
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/disable_iGPU.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/disable_iGPU.png" width=100%; />
|
||||
</a>
|
||||
8
docs/readthedocs/source/doc/LLM/Docker/index.rst
Normal file
8
docs/readthedocs/source/doc/LLM/Docker/index.rst
Normal file
|
|
@ -0,0 +1,8 @@
|
|||
IPEX-LLM Docker Container User Guides
|
||||
=====================================
|
||||
|
||||
In this section, you will find guides related to using IPEX-LLM with Docker, covering how to:
|
||||
|
||||
|
||||
* `Overview of IPEX-LLM Containers for Intel GPU <./docker_windows_gpu.html>`_
|
||||
* `Run PyTorch Inference on an Intel GPU via Docker <./docker_pytorch_inference_gpu.html>`_
|
||||
|
|
@ -1,135 +0,0 @@
|
|||
# Install IPEX-LLM in Docker on Windows with Intel GPU
|
||||
|
||||
This guide demonstrates how to install IPEX-LLM in Docker on Windows with Intel GPUs.
|
||||
|
||||
It applies to Intel Core Core 12 - 14 gen integrated GPUs (iGPUs) and Intel Arc Series GPU.
|
||||
|
||||
> Note:
|
||||
> - WSL2 support is required during the installation process.
|
||||
> - This installation method requires at least 35GB of free disk space on C drive.
|
||||
|
||||
## Install Docker on Windows
|
||||
**Getting Started with Docker:**
|
||||
1. **For New Users:**
|
||||
- Begin by visiting the [official Docker Get Started page](https://www.docker.com/get-started/) for a comprehensive introduction and installation guide.
|
||||
- Detailed installation instructions for Windows, including steps for enabling WSL2, can be found on the [Docker Desktop for Windows installation page](https://docs.docker.com/desktop/install/windows-install/).
|
||||
|
||||
2. **Detailed installation Steps for Windows Users:**
|
||||
- **Download and install [Docker Desktop for Windows](https://docs.docker.com/desktop/install/windows-install/)**: The installation could take around 5 minutes, and a reboot will be required after installation.
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_install.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_install.png" width=100%; />
|
||||
</a>
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_install_finish.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_install_finish.png" width=100%; />
|
||||
</a>
|
||||
|
||||
|
||||
- **Install WSL2**: Open PowerShell or Windows Command Prompt in administrator mode by right-clicking and selecting "Run as administrator", enter:
|
||||
```bash
|
||||
wsl --install
|
||||
```
|
||||
The installation could take around 5 minutes. Then restart your machine and open PowerShell or Windows Command Prompt, enter `wsl --list`:
|
||||
|
||||
You can see the similar result like this:
|
||||
```bash
|
||||
C:\Users\arda>wsl --list
|
||||
* Ubuntu
|
||||
docker-desktop-data
|
||||
docker-desktop
|
||||
```
|
||||
- **Enable integration with WSL2 in Docker desktop**: Open Docker desktop and click `Settings`->`Resources`->`WSL integration`->turn on `Ubuntu` button->`Apply & restart`.
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_new.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_desktop_new.png" width=100%; />
|
||||
</a>
|
||||
>Note: If you encounter **Docker Engine stopped** when opening Docker Desktop, you can reopen it in administrator mode.
|
||||
|
||||
- **Check Docker is enabled in WSL2**: Enter `wsl -d Ubuntu` in PowerShell or Windows Command Prompt to run the Ubuntu wsl distribution. Then enter `docker version` to check if docker could be used in WSL. You can see the similar result like this:
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_wsl.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_wsl.png" width=100%; />
|
||||
</a>
|
||||
>Note: During the use of Docker in WSL, Docker Desktop needs to be kept open all the time.
|
||||
|
||||
|
||||
## IPEX LLM Inference with XPU on Windows
|
||||
### 1. Prepare ipex-llm-xpu Docker Image
|
||||
Run the following command in WSL:
|
||||
```bash
|
||||
docker pull intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
```
|
||||
This step will take around 20 minutes depending on your network.
|
||||
|
||||
### 2. Start ipex-llm-xpu Docker Container
|
||||
|
||||
To map the xpu into the container, an example (docker_setup.sh) could be:
|
||||
```bash
|
||||
#/bin/bash
|
||||
export DOCKER_IMAGE=intelanalytics/ipex-llm-xpu:2.1.0-SNAPSHOT
|
||||
export CONTAINER_NAME=my_container
|
||||
export MODEL_PATH=/llm/models[change to your model path]
|
||||
|
||||
sudo docker run -itd \
|
||||
--net=host \
|
||||
--privileged \
|
||||
--device /dev/dri \
|
||||
--memory="32G" \
|
||||
--name=$CONTAINER_NAME \
|
||||
--shm-size="16g" \
|
||||
-v $MODEL_PATH:/llm/llm-models \
|
||||
-v /usr/lib/wsl:/usr/lib/wsl \
|
||||
$DOCKER_IMAGE
|
||||
```
|
||||
Then run this bash in WSL:
|
||||
```
|
||||
./docker_setup.sh
|
||||
```
|
||||
After the container is booted, you could get into the container through `docker exec`.
|
||||
```
|
||||
export CONTAINER_NAME=my_container
|
||||
docker exec -it $CONTAINER_NAME bash
|
||||
```
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/start_docker_in_wsl.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/start_docker_in_wsl.png" width=100%; />
|
||||
</a>
|
||||
To verify the device is successfully mapped into the container, run `sycl-ls` in the docker container to check the result. In a machine with iGPU (Intel(R) UHD Graphics 770), the sampled output is:
|
||||
|
||||
```bash
|
||||
root@docker-desktop:/# sycl-ls
|
||||
[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device OpenCL 1.2 [2023.16.12.0.12_195853.xmain-hotfix]
|
||||
[opencl:cpu:1] Intel(R) OpenCL, Intel(R) Core(TM) i7-14700K OpenCL 3.0 (Build 0) [2023.16.12.0.12_195853.xmain-hotfix]
|
||||
[opencl:gpu:2] Intel(R) OpenCL Graphics, Intel(R) Graphics [0xa780] OpenCL 3.0 NEO [23.35.27191.42]
|
||||
[ext_oneapi_level_zero:gpu:0] Intel(R) Level-Zero, Intel(R) Graphics [0xa780] 1.3 [1.3.26241]
|
||||
```
|
||||
|
||||
>Note: If you want to exit this Docker container, you can enter `exit`, which will exit the container and return to WSL. You can use `docker ps` to check that the container created in this case will still be running. If you want to enter this container again, you can type `docker exec -it $CONTAINER_NAME bash` again in WSL.
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/docker_exit.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/docker_exit.png" width=100%; />
|
||||
</a>
|
||||
|
||||
### 3. Start Inference
|
||||
**Chat Interface**: Use `chat.py` for conversational AI. For example, enter below command in docker container:
|
||||
```bash
|
||||
cd /llm
|
||||
python chat.py --model-path /llm/llm-models/chatglm2-6b
|
||||
```
|
||||
The output is similar like this:
|
||||
```bash
|
||||
Human: What is AI?
|
||||
IPEX-LLM:
|
||||
AI, or Artificial Intelligence, refers to the development of computer systems or machines that can perform tasks that typically require human intelligence. These systems are designed to learn from data and make decisions, or take actions, based on that data.
|
||||
```
|
||||
|
||||
If your machine has both an integrated GPU (iGPU) and a dedicated GPU (dGPU) like ARC, you may encounter the following issue:
|
||||
```bash
|
||||
Abort was called at 62 line in file:
|
||||
./shared/source/os_interface/os_interface.h
|
||||
LIBXSMM_VERSION: main_stable-1.17-3651 (25693763)
|
||||
LIBXSMM_TARGET: adl [Intel(R) Core(TM) i7-14700K]
|
||||
Registry and code: 13 MB
|
||||
Command: python chat.py --model-path /llm/llm-models/chatglm2-6b/
|
||||
Uptime: 29.349235 s
|
||||
Aborted
|
||||
```
|
||||
Disabling the iGPU in Device Manager can resolve this problem.
|
||||
<a href="https://llm-assets.readthedocs.io/en/latest/_images/disable_iGPU.png">
|
||||
<img src="https://llm-assets.readthedocs.io/en/latest/_images/disable_iGPU.png" width=100%; />
|
||||
</a>
|
||||
Loading…
Reference in a new issue