LMFlow
qlora
LMFlow | qlora | |
---|---|---|
10 | 80 | |
8,042 | 9,472 | |
3.5% | - | |
9.6 | 7.4 | |
5 days ago | 7 months ago | |
Python | Jupyter Notebook | |
Apache License 2.0 | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
LMFlow
- Your weekly machine learning digest
-
Any guide/intro to fine-tuning anywhere?
You might want to have a look at LMFlow.
- Robin V2 Launches: Achieves Unparalleled Performance on OpenLLM!
-
[D] Have you tried fine-tuning an open source LLM?
I'd like to recommend LMFlow (https://github.com/OptimalScale/LMFlow), a fast and extensible toolkit for finetuning and inference of large foundation models.
-
[R] DetGPT: Detect What You Need via Reasoning
The "reasoning-based object detection" is a challenging problem because the detector needs to understand and reason about the user's coarse-grained/abstract instructions and analyze the current visual information to locate the target object accurately. In this direction, researchers from the Hong Kong University of Science and Technology and the University of Hong Kong have conducted some preliminary explorations. Specifically, they use a pre-trained visual encoder (BLIP-2) to extract visual features from images and align the visual features to the text space using an alignment function. They use a large-scale language model (Robin/Vicuna) to understand the user's question, combined with the visual information they see, to reason about the objects that users are truly interested in. Then, they provide the object names to the pre-trained detector (Grounding-DINO) for specific location prediction. In this way, the model can analyze the image based on any user instructions and accurately predict the location of the object of interest to the user. It is worth noting that the difficulty here mainly lies in the fact that the model needs to achieve task-specific output formats for different specific tasks as much as possible without damaging the model's original abilities. To guide the language model to follow specific patterns and generate outputs that conform to the object detection format, the research team used ChatGPT to generate cross-modal instruction data to fine-tune the model. Specifically, based on 5000 coco images, they used ChatGPT to create a 30,000 cross-modal image-text fine-tuning dataset. To improve the efficiency of training, they fixed other model parameters and only learned cross-modal linear mapping. Experimental results show that even if only the linear layer is fine-tuned, the language model can understand fine-grained image features and follow specific patterns to perform inference-based image detection tasks, showing excellent performance. This research topic has great potential. Based on this technology, the field of home robots will further shine: people in homes can use abstract or coarse-grained voice instructions to make robots understand, recognize, and locate the objects they need, and provide relevant services. In the field of industrial robots, this technology will bring endless vitality: industrial robots can cooperate more naturally with human workers, accurately understand their instructions and needs, and achieve intelligent decision-making and operations. On the production line, human workers can use coarse-grained voice instructions or text input to allow robots to automatically understand, recognize, and locate the items that need to be processed, thereby improving production efficiency and quality. With object detection models that come with reasoning capabilities, we can develop more intelligent, natural, and efficient robots to provide more convenient, efficient, and humane services to humans. This is a field with broad prospects and deserves more attention and further exploration by more researchers. DetGPT supports multiple language models and has been validated based on two language models, Robin-13B and Vicuna-13B. The Robin series language model is a dialogue model trained by the LMFlow team ( https://github.com/OptimalScale/LMFlow) at the Hong Kong University of Science and Technology, achieving results competitive to Vicuna on multiple language ability evaluation benchmarks (model download: https://github.com/OptimalScale/LMFlow#model-zoo). Previously, the LMFlow team trained a vertical GPT model using a consumer-grade 3090 graphics card in just 5 hours. Today, this team, in collaboration with the NLP Group at the University of Hong Kong, has brought us a multimodal surprise. Welcome to try our demo and open-source code! Online demo: https://detgpt.github.io/ Open-source code: https://github.com/OptimalScale/DetGPT
-
Leaderboard for LLMs? [D]
Hi LMFlow Benchmark (https://github.com/OptimalScale/LMFlow) evaluates 31 open-source LLMs with an automatic metric: negative log likelihood.
-
[R] LMFlow Benchmark: An Automatic Evaluation Framework for Open-Source LLMs
LMFlow: https://github.com/OptimalScale/LMFlow
-
[R] Foundation Model Alignment with RAFT🛶 in LMFlow
Its implementation is available from https://github.com/OptimalScale/LMFlow.
- LMFlow – Toolkit for Finetuning and Inference of Large Foundation Models
qlora
- FLaNK Stack Weekly for 30 Oct 2023
-
I released Marx 3B V3.
Marx 3B V3 is StableLM 3B 4E1T instruction tuned on EverythingLM Data V3(ShareGPT Format) for 2 epochs using QLoRA.
-
Tuning and Testing Llama 2, Flan-T5, and GPT-J with LoRA, Sematic, and Gradio
https://github.com/artidoro/qlora
The tools and mechanisms to get a model to do what you want is ever so changing, ever so quickly. Build and understand a notebook yourself, and reduce dependencies. You will need to switch them.
-
Yet another QLoRA tutorial
My own project right now is still in raw generated form, and this now makes me think about trying qlora's scripts since this gives me some confidence I should be able to get it to turn out now that someone else has carved a path and charted the map. I was going to target llamatune which was mentioned here the other day.
-
Creating a new Finetuned model
Most papers I did read showed at least a thousand, even 10000 at several cases, so I assumed that to be the trend in the case of Low rank adapter(PEFT) training.(source: [2305.14314] QLoRA: Efficient Finetuning of Quantized LLMs (arxiv.org) , Stanford CRFM (Alpaca) and the minimum being openchat/openchat · Hugging Face ; There are a lot more examples)
-
[R] LaVIN-lite: Training your own Multimodal Large Language Models on one single GPU with competitive performance! (Technical Details)
4-bit quantization training mainly refers to qlora. Simply put, qlora quantizes the weights of the LLM into 4-bit for storage, while dequantizing them into 16-bit during the training process to ensure training precision. This method significantly reduces GPU memory overhead during training (the training speed should not vary much). This approach is highly suitable to be combined with parameter-efficient methods. However, the original paper was designed for single-modal LLMs and the code has already been wrapped in HuggingFace's library. Therefore, we extracted the core code from HuggingFace's library and migrated it into LaVIN's code. The main principle is to replace all linear layers in LLM with 4-bit quantized layers. Those interested can refer to our implementation in quantization.py and mm_adaptation.py, which is roughly a dozen lines of code.
-
[D] To all the machine learning engineers: most difficult model task/type you’ve ever had to work with?
There have been some new development like QLora which help fine-tune LLMs without updating all the weights.
-
Finetune MPT-30B using QLORA
This might be helpful: https://github.com/artidoro/qlora/issues/10
-
is lora fine-tuning on 13B/33B/65B comparable to full fine-tuning?
curious, since qlora paper only reports lora/qlora comparison for full fine-tuning for small 7B models.for 13B/33B/65B, it does not do so (table 4 in paper)it would be helpful if anyone can please provide links where I can read more on efficacy of lora or disadvantages of lora?
-
Need a detailed tutorial on how to create and use a dataset for QLoRA fine-tuning.
This might not be appropriate answer but did you take a look at this repository? https://github.com/artidoro/qlora With artidoro's repository it's pretty easy to train qlora. You just prepare your own dataset and run the following command: python qlora.py --model_name_or_path --dataset="path/to/your/dataset" --dataset_format="self-instruct" This is only available for several dataset formats. But every dataset format has to have input-output pairs. So the dataset json format has to be like this [ { “input”: “something ”, “output”:“something ” }, { “input”: “something ”, “output”:“something ” } ]
What are some alternatives?
axolotl - Go ahead and axolotl questions
alpaca-lora - Instruct-tune LLaMA on consumer hardware
CogVLM - a state-of-the-art-level open visual language model | 多模态预训练模型
GPTQ-for-LLaMa - 4 bits quantization of LLaMA using GPTQ
chatgpt_macro_for_texstudio - The ChatGPT Macro for TeXstudio is a user-friendly integration that connects TeXstudio with OpenAI's API.
bitsandbytes - Accessible large language models via k-bit quantization for PyTorch.
llm-foundry - LLM training code for Databricks foundation models
ggml - Tensor library for machine learning
const_layout - Official implementation of the MM'21 paper "Constrained Graphic Layout Generation via Latent Optimization" (LayoutGAN++, CLG-LO, and Layout evaluation)
alpaca_lora_4bit
giskard - 🐢 Open-Source Evaluation & Testing framework for LLMs and ML models