bfa1367149
* Change installation address Change former address: "https://docs.conda.io/en/latest/miniconda.html#" to new address: "https://conda-forge.org/download/" for 63 occurrences under python\llm\example * Change Prompt Change "Anaconda Prompt" to "Miniforge Prompt" for 1 occurrence * Create and update model minicpm * Update model minicpm Update model minicpm under GPU/PyTorch-Models * Update readme and generate.py change "prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=False)" and delete "pip install transformers==4.37.0 " * Update comments for minicpm GPU Update comments for generate.py at minicpm GPU * Add CPU example for MiniCPM * Update minicpm README for CPU * Update README for MiniCPM and Llama3 * Update Readme for Llama3 CPU Pytorch * Update and fix comments for MiniCPM |
||
|---|---|---|
| .. | ||
| generate.py | ||
| README.md | ||
MiniCPM
In this directory, you will find examples on how you could apply IPEX-LLM INT4 optimizations on MiniCPM models. For illustration purposes, we utilize the openbmb/MiniCPM-2B-sft-bf16 as a reference MiniCPM model.
0. Requirements
To run these examples with IPEX-LLM, we have some recommended requirements for your machine, please refer to here for more information.
Example: Predict Tokens using generate() API
In the example generate.py, we show a basic use case for a MiniCPM model to predict the next N tokens using generate() API, with IPEX-LLM INT4 optimizations.
1. Install
We suggest using conda to manage environment:
On Linux:
conda create -n llm python=3.11
conda activate llm
# install ipex-llm with 'all' option
pip install --pre --upgrade ipex-llm[all] --extra-index-url https://download.pytorch.org/whl/cpu
On Windows:
conda create -n llm python=3.11
conda activate llm
pip install --pre --upgrade ipex-llm[all]
2. Run
python ./generate.py --repo-id-or-model-path REPO_ID_OR_MODEL_PATH --prompt PROMPT --n-predict N_PREDICT
Arguments info:
--repo-id-or-model-path REPO_ID_OR_MODEL_PATH: argument defining the huggingface repo id for the MiniCPM model (e.g.openbmb/MiniCPM-2B-sft-bf16) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be'openbmb/MiniCPM-2B-sft-bf16'.--prompt PROMPT: argument defining the prompt to be infered (with integrated prompt format for chat). It is default to be'What is AI?'.--n-predict N_PREDICT: argument defining the max number of tokens to predict. It is default to be32.
Note
: When loading the model in 4-bit, IPEX-LLM converts linear layers in the model into INT4 format. In theory, a XB model saved in 16-bit will requires approximately 2X GB of memory for loading, and ~0.5X GB memory for further inference.
Please select the appropriate size of the MiniCPM model based on the capabilities of your machine.
2.1 Client
On client Windows machine, it is recommended to run directly with full utilization of all cores:
python ./generate.py
2.2 Server
For optimal performance on server, it is recommended to set several environment variables (refer to here for more information), and run the example with all the physical cores of a single socket.
E.g. on Linux,
# set IPEX-LLM env variables
source ipex-llm-init
# e.g. for a server with 48 cores per socket
export OMP_NUM_THREADS=48
numactl -C 0-47 -m 0 python ./generate.py
2.3 Sample Output
openbmb/MiniCPM-2B-sft-bf16
Inference time: xxxx s
-------------------- Prompt --------------------
<用户>what is AI?<AI>
-------------------- Output --------------------
<s> <用户>what is AI?<AI> AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It is a broad field of computer