SWIFT (Scalable lightWeight Infrastructure for Fine-Tuning)

📖 Table of Contents

Introduction
News
🔥LLM Training and Inference
🔥SCEdit Tuner
Installation
Getting Started
Learn More
License
Contact Us

📝 Introduction

SWIFT (Scalable lightWeight Infrastructure for Fine-Tuning) is an extensible framework designed to facilitate lightweight model fine-tuning and inference. It integrates implementations for various efficient fine-tuning methods, by embracing approaches that is parameter-efficient, memory-efficient, and time-efficient. SWIFT integrates seamlessly into ModelScope ecosystem and offers the capabilities to finetune various models, with a primary emphasis on LLMs and vision models. Additionally, SWIFT is fully compatible with PEFT, enabling users to leverage the familiar Peft interface to finetune ModelScope models.

Currently supported approaches (and counting):

🔥LoRA: LORA: LOW-RANK ADAPTATION OF LARGE LANGUAGE MODELS
🔥LoRA+: LoRA+: Efficient Low Rank Adaptation of Large Models
🔥LLaMA PRO: LLAMA PRO: Progressive LLaMA with Block Expansion
🔥SCEdit: SCEdit: Efficient and Controllable Image Diffusion Generation via Skip Connection Editing < arXiv | Project Page >
🔥NEFTune: Noisy Embeddings Improve Instruction Finetuning
QA-LoRA:Quantization-Aware Low-Rank Adaptation of Large Language Models.
LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models
ROME: Rank-One Editing of Encoder-Decoder Models
Adapter: Parameter-Efficient Transfer Learning for NLP
Prompt Tuning: Visual Prompt Tuning
Side: Side-Tuning: A Baseline for Network Adaptation via Additive Side Networks
Res-Tuning: Res-Tuning: A Flexible and Efficient Tuning Paradigm via Unbinding Tuner from Backbone < arXiv | Project Page | Usage >
All tuners offered on PEFT, like IA3, AdaLoRA

Key features:

By integrating the ModelScope library, models can be readily obtained via a model-id.
Tuners provided by SWIFT can be combined to allow exploration of multiple tuners on a model for best result.
Support calling activate_adapter or deactivate_adapter or set_active_adapters to activate/deactivate tuners. User can inference with one model and multiple tuners in different threads independently.
Support training and inference with scripts/CLI，meanwhile support inference with Web-UI.
Support model deployment(vllm/chatglm.cpp/xinference)，Check Official documentation for details.

Users can check the documentation of SWIFT to get detail tutorials.

🎉 News

🔥2024.03.10: For the end-to-end best practice of fine-tuning to deployment of Qwen1.5-7B-Chat and Qwen1.5-72B-Chat, you can refer to the Qwen1.5 Full Workflow Best Practice.
🔥2024.03.09: Support training and inference of MAMBA series, use this script to begin.
2024.03.09: Support training and inference of AQLM quantized models, use this script to begin.
2024.03.06: Support training and inference of AWQ quantized models, use this Qwen1.5-AWQ script to begin, support training and inference of yi-9b.
🔥2024.02.29: Support LLaMA PRO, use this script to begin.
🔥2024.02.29: Support LoRA+, use this script to begin.
2024.02.25: Support swift export to export models for AWQ/GPTQ quantization and push to ModelScope Hub. For more details, please refer to the document: LLM Quantization Document.
2024.02.22: Support gemma series: gemma-2b, gemma-2b-instruct, gemma-7b, gemma-7b-instruct.
2024.02.16: Support deepseek-math series: deepseek-math-7b, deepseek-math-7b-instruct, deepseek-math-7b-chat.
🔥2024.02.05: Support Qwen1.5 series, To view all supported Qwen1.5 models please check Model List. The qwen1half-7b-chat, qwen1half-7b-chat-int8 fine-tuned scripts are provided.
2024.02.05: Support the training of SDXL, SD, ControlNet, or techniques like DreamBooth, you can check the training scripts for details.
2024.02.01: Support openbmb-minicpm series: openbmb-minicpm-2b-sft-chat, openbmb-minicpm-2b-chat.
🔥2024.02.01: Support dataset mixture to reduce Catastrophic Forgetting. Use --train_dataset_mix_ratio 2.0 to train! We also provide a common knowledge dataset ms-bench.
🔥2024.02.01: Support Agent training! Agent training algorithm comes from this paper. We also introduce the ms-agent dataset. Use this script to begin an agent training!
🔥2024.02.01: Support SFT loss to DPO training to reduce the repeat generation problem caused by the KL-divergence loss.
2024.02.01: Support AdaLoRA and IA3 adapter in SFT.
2024.02.01: Support --merge_lora in AnimateDiff training.

2024.01.30: Support internlm-xcomposer2-7b-chat.
🔥2024.01.30: Support ZeRO-3, just need to specify --deepspeed default-zero3.
2024.01.29: Support internlm2-math series: internlm2-math-7b, internlm2-math-7b-chat, internlm2-math-20b, internlm2-math-20b-chat.
🔥2024.01.26: Support yi-vl-6b-chat, yi-vl-34b-chat.
2024.01.24: Support codefuse-codegeex2-6b-chat, codefuse-qwen-14b-chat.
2024.01.23: Support orion series: orion-14b, orion-14b-chat.
2024.01.20: Support xverse-13b-256k, xverse-65b-v2, xverse-65b-chat.
🔥2024.01.17: Support internlm2 series: internlm2-7b-base, internlm2-7b, internlm2-7b-sft-chat, internlm2-7b-chat, internlm2-20b-base, internlm2-20b, internlm2-20b-sft-chat, internlm2-20b-chat.
2024.1.15: Support yuan series: yuan2-2b-instruct, yuan2-2b-janus-instruct, yuan2-51b-instruct, yuan2-102b-instruct.
🔥2024.01.12: Support deepseek-moe series: deepseek-moe-16b, deepseek-moe-16b-chat.
🔥2024.01.04: Support for VLLM deployment, compatible with the OpenAI API style. For more details, please refer to VLLM Inference Acceleration and Deployment
2024.01.04: Update Benchmark to facilitate viewing the training speed and GPU memory required for different models.
🔥 2023.12.29: Support web-ui for training and inference, use swift web-ui after the installation of ms-swift.
🔥 2023.12.29: Support DPO RLHF(Reinforcement Learning from Human Feedback) and three datasets: AI-ModelScope/stack-exchange-paired and AI-ModelScope/hh-rlhf and AI-ModelScope/hh_rlhf_cn for this task. Check this documentation to start training!
🔥 2023.12.28: Support SCEdit! This framework can easily reduce memory usage in training and inference, and replace ControlNet for controllable image generating scenarios, view the following chapter for details.
2023.12.23: Support codegeex2-6b.
2023.12.19: Support phi2-3b.
2023.12.18: Support for VLLM for inference acceleration.
2023.12.15: Support deepseek, deepseek-coder series: deepseek-7b, deepseek-7b-chat, deepseek-67b, deepseek-67b-chat, openbuddy-deepseek-67b-chat, deepseek-coder-1_3b, deepseek-coder-1_3b-instruct, deepseek-coder-6_7b, deepseek-coder-6_7b-instruct, deepseek-coder-33b, deepseek-coder-33b-instruct.
2023.12.13: Support mistral-7b-instruct-v2, mixtral-moe-7b, mixtral-moe-7b-instruct.
2023.12.09: Support the freeze_parameters parameter as a compromise between LoRA and full parameter. Corresponding shell scripts can be found at full_freeze_ddp. Support disable_tqdm, lazy_tokenize, preprocess_num_proc parameters, for details please refer to Command-Line parameters.
2023.12.08: Support sus-34b-chat, support yi-6b-200k, yi-34b-200k.
2023.12.07: Support Multi-Node DDP training.
2023.12.04: Supported models: zephyr-7b-beta-chat, openbuddy-zephyr-7b-chat. Supported datasets: hc3-zh, hc3-en.
🔥 2023.12.02: Best Practice for Self-cognition Fine-tuning, 10 minutes for self-cognition fine-tuning for LLM, creating a LLM that is specific to oneself.
🔥 2023.11.30: Support for training and inference of the qwen-1_8b, qwen-72b, and qwen-audio model series. The corresponding shell scripts can be viewed at qwen_1_8b_chat, qwen_72b_chat, qwen_audio_chat.
🔥 2023.11.29: Support the training and inference for AnimateDiff
🔥 2023.11.24: Support for yi-34b-chat, codefuse-codellama-34b-chat: The corresponding shell script can be found in yi_34b_chat, codefuse_codellama_34b_chat.
🔥 2023.11.18: Support for tongyi-finance-14b series models: tongyi-finance-14b, tongyi-finance-14b-chat, tongyi-finance-14b-chat-int4. The corresponding shell script can be found in tongyi_finance_14b_chat_int4.
2023.11.16: Added support for more models in flash attn: qwen series, qwen-vl series, llama series, openbuddy series, mistral series, yi series, ziya series. Please use the use_flash_attn parameter.
🔥 2023.11.11: NEFTune Supported, Use is with Swift.prepare_model(model, NEFTuneConfig())
🔥 2023.11.11: Support training and inference with CLI, and inference with Web-UI. Check the Run using Swift CLI chapter for details.
🔥 2023.11.11: Support model deployment(vllm/chatglm.cpp/xinference)，Check Official documentation for details.
🔥 2023.11.10: Support for bluelm series models: bluelm-7b, bluelm-7b-chat, bluelm-7b-32k, bluelm-7b-chat-32k. The corresponding shell script can be found in bluelm_7b_chat.
🔥 2023.11.08: Support the finetuning of xverse-65b model, scripts can be found at: xverse_65b.
🔥 2023.11.07: Support the finetuning of yi-6b, yi-34b model, scripts can be found at: yi_6b, yi_34b.
🔥 2023.10.30: Support QA-LoRA and LongLoRA to decrease memory usage in training.
🔥 2023.10.30: Support ROME(Rank One Model Editing) to add/modify knowledges, training is not needed!
2023.10.30: Support for skywork-13b series models: skywork-13b, skywork-13b-chat. The corresponding shell script can be found in skywork_13b.
🔥 2023.10.27: Support for chatglm3 series models: chatglm3-6b-base, chatglm3-6b, chatglm3-6b-32k. The corresponding shell script can be found in chatglm3_6b.
🔥 2023.10.17: Supported int4, int8 models: qwen-7b-chat-int4, qwen-14b-chat-int4, qwen-vl-chat-int4, baichuan2-7b-chat-int4, baichuan2-13b-chat-int4, qwen-7b-chat-int8, qwen-14b-chat-int8.
2023.10.15: Supported ziya2-13b model series: ziya2-13b, ziya2-13b-chat.
2023.10.12: Supported mistral-7b model series: openbuddy-mistral-7b-chat, mistral-7b, mistral-7b-instruct.
🔥 2023.10.07: Supported DeepSpeed ZeRO-2, enabling LoRA (not just QLoRA) to run DDP on 2*A10.
2023.10.04: Supported datasets in the fields of mathematics, law, SQL, and coding: blossom-math-zh, school-math-zh, text2sql-en, sql-create-context-en, lawyer-llama-zh, tigerbot-law-zh, leetcode-python-en.
🔥 2023.09.25: Supported qwen-14b model series: qwen-14b, qwen-14b-chat.
2023.09.18: Supported internlm-20b model series: internlm-20b, internlm-20b-chat.
2023.09.12: Supported training with MP+DDP to accelerate full-parameter fine-tuning speed.
2023.09.05: Supported openbuddy-llama2-70b-chat model.
2023.09.03: Supported baichuan2 model series: baichuan2-7b, baichuan2-7b-chat, baichuan2-13b, baichuan2-13b-chat.

✨ LLM Training and Inference

If you want to learn more about best practices for fine-tuning LLM or tutorials on inference, fine-tuning, quantization, and deployment of LLM, you can refer to the LLM series documentation.

WEB UI training and inference

After installation, you can use web-ui training/inference like:

SWIFT_UI_LANG=en swift web-ui

Supported environment variables:

WEBUI_SHARE=1 Share the gradio or not SWIFT_UI_LANG=en/zh The language of radio WEBUI_SERVER server_name， web-ui host ip，0.0.0.0 means all routes are allowed，127.0.0.1 means only localhost can visit the web WEBUI_PORT The port of web-ui

Here is a simple introduction of web-ui:

Quick Start

You can test if the environment is installed correctly by running the following code.

# pip install ms-swift[llm] -U

# Experimental environment: A10, 3090, V100, ...
# 8GB GPU memory
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'

import torch

from swift.llm import (
    DatasetName, InferArguments, ModelType, SftArguments,
    infer_main, sft_main, app_ui_main, merge_lora
)

model_type = ModelType.qwen1half_0_5b
sft_args = SftArguments(
    model_type=model_type,
    train_dataset_sample=2000,
    dataset=[DatasetName.jd_sentiment_zh],
    output_dir='output')
result = sft_main(sft_args)
best_model_checkpoint = result['best_model_checkpoint']
print(f'best_model_checkpoint: {best_model_checkpoint}')
torch.cuda.empty_cache()

infer_args = InferArguments(
    ckpt_dir=best_model_checkpoint,
    load_dataset_config=True,
    val_dataset_sample=10)
merge_lora(infer_args, device_map='cpu')
result = infer_main(infer_args)
torch.cuda.empty_cache()

app_ui_main(infer_args)

Training Scripts

You can refer to the following scripts to customize your own training script.

full: qwen1half-7b-chat (A100), qwen-7b-chat (2*A100)
full+ddp+zero2: qwen-7b-chat (4*A100)
full+ddp+zero3: qwen-14b-chat (4*A100)
lora: chatglm3-6b (3090), baichuan2-13b-chat (2*3090), yi-34b-chat (A100), qwen-72b-chat (2*A100)
lora+ddp: chatglm3-6b (2*3090)
lora+ddp+zero3: qwen-14b-chat (4*3090), qwen-72b-chat (4*A100)
qlora(gptq-int4): qwen-7b-chat-int4 (3090)
qlora(gptq-int8): qwen1half-7b-chat-int8 (3090)
qlora(bnb-int4): qwen-7b-chat (3090)

Features

Supported SFT Methods: lora, qlora, longlora, qalora, full parameter fine-tuning, partial parameter fine-tuning.
Supported Features: quantization, DDP, model parallelism, gradient checkpointing, pushing to modelscope hub, custom datasets, multimodal and agent SFT, mutli-round chat, DPO, self-cognition fine-tuning, ...
Supported Models: [Detailed Info]
- Multi-Modal:
  - qwen-vl series: qwen-vl, qwen-vl-chat, qwen-vl-chat-int4.
  - qwen-audio series: qwen-audio, qwen-audio-chat.
  - yi-vl series: yi-vl-6b-chat, yi-vl-34b-chat.
  - cogagent series: cogagent-18b-chat, cogagent-18b-instruct.
  - internlm-xcomposer2 series: internlm-xcomposer2-7b-chat.
- General:
  - qwen series:
    - qwen-1_8b, qwen-1_8b-chat, qwen-1_8b-chat-int4, qwen-1_8b-chat-int8.
    - qwen-7b, qwen-7b-chat, qwen-7b-chat-int4, qwen-7b-chat-int8.
    - qwen-14b, qwen-14b-chat, qwen-14b-chat-int4, qwen-14b-chat-int8.
    - qwen-72b, qwen-72b-chat, qwen-72b-chat-int4, qwen-72b-chat-int8.
  - qwen1.5 series:
    - qwen1half-0_5b, qwen1half-0_5b-chat, qwen1half-0_5b-chat-int4, qwen1half-0_5b-chat-int8, qwen1half-0_5b-chat-awq.
    - qwen1half-1_8b, qwen1half-1_8b-chat, qwen1half-1_8b-chat-int4, qwen1half-1_8b-chat-int8, qwen1half-1_8b-chat-awq.
    - qwen1half-4b, qwen1half-4b-chat, qwen1half-4b-chat-int4, qwen1half-4b-chat-int8, qwen1half-4b-chat-awq.
    - qwen1half-7b, qwen1half-7b-chat, qwen1half-7b-chat-int4, qwen1half-7b-chat-int8, qwen1half-7b-chat-awq.
    - qwen1half-14b, qwen1half-14b-chat, qwen1half-14b-chat-int4, qwen1half-14b-chat-int8, qwen1half-14b-chat-awq.
    - qwen1half-72b, qwen1half-72b-chat, qwen1half-72b-chat-int4, qwen1half-72b-chat-int8, qwen1half-72b-chat-awq.
  - chatglm series: chatglm2-6b, chatglm2-6b-32k, chatglm3-6b-base, chatglm3-6b, chatglm3-6b-32k.
  - llama series: llama2-7b, llama2-7b-chat, llama2-13b, llama2-13b-chat, llama2-70b, llama2-70b-chat.
  - yi series: yi-6b, yi-6b-200k, yi-6b-chat, yi-9b, yi-34b, yi-34b-200k, yi-34b-chat.
  - internlm series:
    - internlm-7b, internlm-7b-chat, internlm-7b-chat-8k, internlm-20b, internlm-20b-chat.
    - internlm2-1_8b, internlm2-1_8b-sft-chat, internlm2-1_8b-chat, internlm2-7b-base, internlm2-7b, internlm2-7b-sft-chat, internlm2-7b-chat, internlm2-20b-base, internlm2-20b, internlm2-20b-sft-chat, internlm2-20b-chat.
  - deepseek series: deepseek-7b, deepseek-7b-chat, deepseek-67b, deepseek-67b-chat, deepseek-moe-16b, deepseek-moe-16b-chat.
  - gemma series: gemma-2b, gemma-2b-instruct, gemma-7b, gemma-7b-instruct.
  - openbmb-minicpm series: openbmb-minicpm-2b-sft-chat, openbmb-minicpm-2b-chat.
  - openbuddy series: openbuddy-llama2-13b-chat, openbuddy-llama-65b-chat, openbuddy-llama2-70b-chat, openbuddy-mistral-7b-chat, openbuddy-zephyr-7b-chat, openbuddy-deepseek-67b-chat, openbuddy-mixtral-moe-7b-chat.
  - mistral series: mistral-7b, mistral-7b-instruct, mistral-7b-instruct-v2.
  - mixtral series: mixtral-moe-7b, mixtral-moe-7b-instruct.
  - baichuan series: baichuan-7b, baichuan-13b, baichuan-13b-chat, baichuan2-7b, baichuan2-7b-chat, baichuan2-13b, baichuan2-13b-chat, baichuan2-7b-chat-int4, baichuan2-13b-chat-int4.
  - yuan series: yuan2-2b-instruct, yuan2-2b-janus-instruct, yuan2-51b-instruct, yuan2-102b-instruct.
  - xverse series: xverse-7b, xverse-7b-chat, xverse-13b, xverse-13b-chat, xverse-65b, xverse-65b-v2, xverse-65b-chat, xverse-13b-256k.
  - orion series: orion-14b, orion-14b-chat.
  - bluelm series: bluelm-7b, bluelm-7b-chat, bluelm-7b-32k, bluelm-7b-chat-32k.
  - zephyr series: zephyr-7b-beta-chat.
  - ziya series: ziya2-13b, ziya2-13b-chat.
  - skywork series: skywork-13b, skywork-13b-chat.
  - other: polylm-13b, seqgpt-560m, sus-34b-chat.
- Financial:
  - tongyi-finance series: tongyi-finance-14b, tongyi-finance-14b-chat, tongyi-finance-14b-chat-int4.
- Coding:
  - codefuse series: codefuse-codellama-34b-chat, codefuse-codegeex2-6b-chat, codefuse-qwen-14b-chat.
  - deepseek-coder series: deepseek-coder-1_3b, deepseek-coder-1_3b-instruct, deepseek-coder-6_7b, deepseek-coder-6_7b-instruct, deepseek-coder-33b, deepseek-coder-33b-instruct.
  - codegeex2 series: codegeex2-6b.
  - phi series: phi2-3b.
- Math:
  - internlm2-math series: internlm2-math-7b, internlm2-math-7b-chat, internlm2-math-20b, internlm2-math-20b-chat.
  - deepseek-math series: deepseek-math-7b, deepseek-math-7b-instruct, deepseek-math-7b-chat.
Supported Datasets: [Detailed Info]
- NLP:
  - General: 🔥ms-bench, 🔥ms-bench-mini, 🔥alpaca-en(gpt4), 🔥alpaca-zh(gpt4), multi-alpaca-all, instinwild-en, instinwild-zh, cot-en, cot-zh, firefly-all-zh, instruct-en, gpt4all-en, sharegpt-en, sharegpt-zh, tulu-v2-sft-mixture, wikipedia-zh, open-orca, open-orca-gpt4, sharegpt-gpt4, 🔥sharegpt-gpt4-mini.
  - Agent: 🔥ms-agent, damo-mini-agent-zh, damo-agent-zh, agent-instruct-all-en.
  - RLHF: 🔥hh-rlhf-cn, stack-exchange-paired, hh-rlhf-harmless-base, hh-rlhf-helpful-base, hh-rlhf-helpful-online, hh-rlhf-helpful-rejection-sampled, hh-rlhf-red-team-attempts, hh-rlhf-cn-harmless-base-cn, hh-rlhf-cn-helpful-base-cn, hh-rlhf-cn-harmless-base-en, hh-rlhf-cn-helpful-base-en.
  - Coding: code-alpaca-en, 🔥leetcode-python-en, 🔥codefuse-python-en, 🔥codefuse-evol-instruction-zh.
  - Medical: medical-en, medical-zh, medical-mini-zh, 🔥disc-med-sft-zh.
  - Law: lawyer-llama-zh, tigerbot-law-zh, 🔥disc-law-sft-zh.
  - Math: 🔥blossom-math-zh, school-math-zh, open-platypus-en.
  - SQL: text2sql-en, 🔥sql-create-context-en.
  - Text Generation: 🔥advertise-gen-zh, 🔥dureader-robust-zh.
  - Classification: cmnli-zh, 🔥cmnli-mini-zh, 🔥jd-sentiment-zh, 🔥hc3-zh, 🔥hc3-en.
  - AWQ: pileval.
  - Other: finance-en, poetry-zh, webnovel-zh, generated-chat-zh, cls-fudan-news-zh, ner-jave-zh.
- Multi-Modal:
  - Vision: coco-en, 🔥coco-mini-en, coco-mini-en-2, capcha-images.
  - Audio: aishell1-zh, 🔥aishell1-mini-zh.
- Custom Dataset
Supported Templates:
- Text Generation: default-generation, default-generation-bos, chatglm-generation, qwen-audio-generation.
- Chat: default, qwen, qwen-audio, baichuan, chatglm2, chatglm3, llama, openbuddy, internlm, internlm2, internlm-xcomposer2, yi, yi-vl, yuan, xverse, ziya, skywork, bluelm, zephyr, sus, deepseek, deepseek-coder, codefuse-codellama, codefuse, cogagent-chat, cogagent-instruct, orion, openbmb, gemma, chatml.

🔥SCEdit

SCEdit is an efficient generative fine-tuning framework proposed by Alibaba TongYi Vision Intelligence Lab. This framework enhances the fine-tuning capabilities for text-to-image generation downstream tasks and enables quick adaptation to specific generative scenarios, saving 30%-50% of training memory costs compared to LoRA. Furthermore, it can be directly extended to controllable image generation tasks, requiring only 7.9% of the parameters that ControlNet needs for conditional generation and saving 30% of memory usage. It supports various conditional generation tasks including edge maps, depth maps, segmentation maps, poses, color maps, and image completion.

We using 3D style data from the style transfer dataset for training, and testing with the same Prompt: A boy in a camouflage jacket with a scarf. The qualitative and quantitative results are as follows:

Method	bs	ep	Target Module	Param. (M)	Mem. (MiB)
LoRA/r=64	1	50	q/k/v/out/mlp	23.94 (2.20%)	8440MiB
SCEdit	1	50	up_blocks	19.68 (1.81%)	7556MiB
LoRA/r=64	10	100	q/k/v/out/mlp	23.94 (2.20%)	26300MiB
SCEdit	10	100	up_blocks	19.68 (1.81%)	18634MiB
LoRA/r=64	30	200	q/k/v/out/mlp	23.94 (2.20%)	69554MiB
SCEdit	30	200	up_blocks	19.68 (1.81%)	43350MiB

The benchmark listed above can be reproduced by：

# Install swift by the next chapter
cd examples/pytorch/multi_modal/notebook
python text_to_image_synthesis.py

🛠️ Installation

SWIFT is running in Python environment. Please make sure your python version is higher than 3.8.

Install SWIFT by the pip command:

# full ability
pip install ms-swift[all] -U
# only use llm
pip install ms-swift[llm] -U
# only use aigc
pip install ms-swift[aigc] -U
# only use adapters
pip install ms-swift -U

Install SWIFT by source code(for running sft/infer examples), please run:

git clone https://github.com/modelscope/swift.git
cd swift
pip install -e .[llm]

SWIFT requires torch>=1.13.

Use SWIFT in our docker image:

docker pull registry.cn-hangzhou.aliyuncs.com/modelscope-repo/modelscope:ubuntu20.04-cuda11.8.0-py38-torch2.0.1-tf2.13.0-1.9.1

🚀 Getting Started

SWIFT supports multiple tuners, as well as tuners provided by PEFT. To use these tuners, simply call:

from swift import Swift, LoRAConfig
config = LoRAConfig(...)
model = Swift.prepare_model(model, config, extra_state_keys=['...'])

The code snippet above initialized the tuner randomly. The input model is an instance of torch.nn.Module, the config is a subclass instance of SwiftConfig or PeftConfig. extra_state_keys is the extra module weights(like the linear head) to be trained and stored in the output dir.

You may combine multiple tuners by:

from swift import Swift, LoRAConfig, PromptConfig
model = Swift.prepare_model(model, {'lora': LoRAConfig(...), 'prompt': PromptConfig(...)})

Call save_pretrained and push_to_hub after finetuning:

from swift import push_to_hub
model.save_pretrained('some-output-folder')
push_to_hub('my-group/some-repo-id-modelscope', 'some-output-folder', token='some-ms-token')

Assume my-group/some-repo-id-modelscope is the model-id in the hub, and some-ms-token is the token for uploading.

Using the model-id to do later inference:

from swift import Swift
model = Swift.from_pretrained(model, 'my-group/some-repo-id-modelscope')

Here shows a runnable example:

import os
import tempfile

# Please install modelscope by `pip install modelscope`
from modelscope import Model

from swift import LoRAConfig, SwiftModel, Swift, push_to_hub

tmp_dir = tempfile.TemporaryDirectory().name
if not os.path.exists(tmp_dir):
    os.makedirs(tmp_dir)


model = Model.from_pretrained('modelscope/Llama-2-7b-ms', device_map='auto')
lora_config = LoRAConfig(target_modules=['q_proj', 'k_proj', 'v_proj'])
model: SwiftModel = Swift.prepare_model(model, lora_config)
# Do some finetuning here
model.save_pretrained(tmp_dir)

push_to_hub('my-group/swift_llama2', output_dir=tmp_dir)
model = Model.from_pretrained('modelscope/Llama-2-7b-ms', device_map='auto')
model = SwiftModel.from_pretrained(model, 'my-group/swift_llama2', device_map='auto')

This is a example that uses transformers for model creation uses SWIFT for efficient tuning.

from swift import Swift, LoRAConfig, AdapterConfig, PromptConfig
from transformers import AutoModelForImageClassification

# init vit model
model = AutoModelForImageClassification.from_pretrained("google/vit-base-patch16-224")

# init lora tuner config
lora_config = LoRAConfig(
    r=10,  # the rank of the LoRA module
    target_modules=['query', 'key', 'value'],  # the modules to be replaced with the end of the module name
    merge_weights=False  # whether to merge weights
)

# init adapter tuner config
adapter_config = AdapterConfig(
    dim=768,  # the dimension of the hidden states
    hidden_pos=0,  # the position of the hidden state to passed into the adapter
    target_modules=r'.*attention.output.dense$',  # the modules to be replaced with regular expression
    adapter_length=10  # the length of the adapter length
)

# init prompt tuner config
prompt_config = PromptConfig(
    dim=768,  # the dimension of the hidden states
    target_modules=r'.*layer\.\d+$',  # the modules to be replaced with regular expression
    embedding_pos=0,    # the position of the embedding tensor
    prompt_length=10,   # the length of the prompt tokens
    attach_front=False  # Whether prompt is attached in front of the embedding
)

# create model with swift. In practice, you can use any of these tuners or a combination of them.
model = Swift.prepare_model(model, {"lora_tuner": lora_config, "adapter_tuner": adapter_config, "prompt_tuner": prompt_config})

# get the trainable parameters of model
model.get_trainable_parameters()
# 'trainable params: 838,776 || all params: 87,406,432 || trainable%: 0.9596273189597764'

You can use the features offered by Peft in SWIFT:

from swift import LoraConfig, Swift
from peft import TaskType
lora_config = LoraConfig(target_modules=['query', 'key', 'value'], task_type=TaskType.CAUSAL_LM)
model_wrapped = Swift.prepare_model(model, lora_config)

# or call from_pretrained to load weights in the modelhub
model_wrapped = Swift.from_pretrained(model, 'some-id-in-the-modelscope-modelhub')

The saving strategy between Swift tuners and Peft tuners are slightly different. You can name a tuner by:

model = Swift.prepare_model(model, {'default': LoRAConfig(...)})
model.save_pretrained('./output')

In the output dir, you will have a dir structure like this:

output
    |-- default
        |-- adapter_config.json
        |-- adapter_model.bin
    |-- adapter_config.json
    |-- adapter_model.bin

The config/weights stored in the output dir is the config of extra_state_keys and the weights of it. This is different from PEFT, which stores the weights and config of the default tuner.

🔍 Learn More

ModelScope library

ModelScope Library is the model library of ModelScope project, which contains a large number of popular models.
Contribute your own model to ModelScope

liushuchun / swift