Add metadata, paper link, and improve model card

README.md

@@ -1,323 +1,105 @@
-# 1. Model Introduction
-
-**LimiX** is a new class of tabular AI model designed to overcome one of modern machine learning’s longest-standing bottlenecks: structured data. With only **2M parameters**, **LimiX-2M** sets a new state-of-the-art across classification, regression, and missing-value imputation, surpassing XGBoost, CatBoost, AutoGluon, and TabPFN, and approaching the performance level of the larger LimiX-16M. Its lightweight, training-free design makes advanced tabular modeling accessible on ordinary hardware while preserving full transparency and offline deployability.
+---
+license: apache-2.0
+pipeline_tag: other
+tags:
+- tabular
+- foundation-model
+---
 
-![](images/image.png)
+# LimiX-2M
 
+LimiX-2M is a 2M-parameter tabular foundation model designed to mitigate low-rank collapse and attention bottlenecks in structured data. It was introduced in the paper [LimiX-2M: Mitigating Low-Rank Collapse and Attention Bottlenecks in Tabular Foundation Models](https://huggingface.co/papers/2606.04485).
 
+- **GitHub Repository:** [https://github.com/limix-ldm-ai/LimiX](https://github.com/limix-ldm-ai/LimiX)
+- **Project Page:** [https://www.limix.ai/](https://www.limix.ai/)
 
+# 1. Model Introduction
 
+**LimiX** is a new class of tabular AI model designed to overcome one of modern machine learning’s longest-standing bottlenecks: structured data. With only **2M parameters**, **LimiX-2M** sets a new state-of-the-art across classification, regression, and missing-value imputation, surpassing XGBoost, CatBoost, AutoGluon, and TabPFN. Its lightweight, training-free design makes advanced tabular modeling accessible on ordinary hardware while preserving full transparency and offline deployability.
 
 **Key Features**
 
-**Unified Tabular Reasoning：**
-
-End-to-end designed for multi-task tabular intelligence, enabling a single model to handle classification, regression, and imputation without additional tuning, preprocessing, or task-specific fine-tuning.
-
-**Training-Free, Context-Driven Inference：**
-
-Operates directly through context learning: no training, no hyperparameters, no preprocessing pipelines. LimiX automatically interprets and processes raw tabular inputs for immediate use.
-
-**Lightweight & Efficient Deployment：**
-
-A compact 2M-parameter architecture enables fast inference and smooth operation on standard CPUs and laptops, dramatically reducing compute requirements for advanced tabular modeling.
-
-
-
-# 2. Model Architecture & Pretraining Procedures
-
-LimiX adopts a 12-block transformer architecture with axis-wise attention to features and samples, supported by pre-normalized LayerNorm for stable scaling. The LimiX-16M variant uses an asymmetric design, two feature-axis passes and one sample-axis pass per block, to strengthen feature interaction modeling in heterogeneous schemas with minimal overhead.
-
-To learn the joint distribution of tabular variables, LimiX is pretrained through Context-Conditional Masked Modeling (CCMM). By masking table cells and conditioning predictions on a small set of context rows, the model internalizes a wide range of conditional dependencies while adapting to new datasets without training or labels.
-
-![](images/image-5.png)
-
-# 3.  Evaluation Results
-
-## Classification
-
-![](images/image-4.png)
-
-On the BCCO-CLS benchmark, LimiX-16M establishes leading performance by significantly outperforming AutoGluon and all PFN variants in mean AUC, Accuracy, and F1 scores, with substantially better ranks. LimiX-2M also marks a clear lead over these models in most metrics, except for its AUC rank.
-
-## Regression
-
-![](images/image-3.png)
-
-LimiX-16M achieves the best overall scores and rankings on TALENT-REG, with the PFN models and LimiX-2M emerging as close runners-up in both R² and RMSE.
-
-## Missing Value Imputation
-
-LimiX introduces the first training-free, in-context approach for missing-value imputation on entirely new datasets. Across a wide set of real-world benchmarks, LimiX-16M delivers the best performance, achieving lower RMSE and error rates than classical and learned imputers including KNN, MICE, MissForest, GAIN, and MIWAE. Unlike all prior methods, which depend on additional fitting, LimiX performs imputation directly from context with consistently superior accuracy.
-
-![](images/image-1.png)
-
-## Finetune
-
-Using an attention-based retrieval–guided downsampling strategy, LimiX-16M fine-tunes on compact, highly relevant in-context episodes rather than full long contexts, substantially improving sample efficiency and reducing training cost. This approach enables LimiX-16M to significantly outperform strong baselines such as TabDPT and TabPFN-v2, with notable AUC gains across BCCO-CLS datasets.
-
-![](images/image-2.png)
-
-# 4. Deployment
+- **Unified Tabular Reasoning:** End-to-end designed for multi-task tabular intelligence, enabling a single model to handle classification, regression, and imputation without additional tuning or task-specific fine-tuning.
+- **Training-Free, Context-Driven Inference:** Operates directly through in-context learning: no training, no hyperparameters, no preprocessing pipelines. 
+- **Lightweight & Efficient Deployment:** A compact 2M-parameter architecture enables fast inference on standard CPUs and laptops.
 
-1. **Environment Preparation**
+# 2. Model Usage
 
-   * Recommended to deploy with Docker. Download the [Dockerfile](https://github.com/limix-ldm/LimiX/blob/main/Dockerfile) from the repository and execute the following command to build the image:
+To use LimiX-2M, you need to clone the official repository and install the dependencies listed in the [Deployment](#4-deployment) section.
 
-     ```bash
-     docker build --network=host -t limix/infe:v1 --build-arg FROM_IMAGES=nvidia/cuda:12.2.0-base-ubuntu22.04 -f Dockerfile .
-     ```
+### Classification Task Example
 
-   * For manual deployment, install dependencies:
+```python
+from sklearn.datasets import load_breast_cancer
+from sklearn.metrics import accuracy_score, roc_auc_score
+from sklearn.model_selection import train_test_split
+from huggingface_hub import hf_hub_download
+import numpy as np
+import os, sys
 
-     ```bash
-     # Download precompiled flash_attn file
-     wget -O flash_attn-2.8.0.post2+cu12torch2.7cxx11abiTRUE-cp312-cp312-linux_x86_64.whl https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.0.post2/flash_attn-2.8.0.post2+cu12torch2.7cxx11abiTRUE-cp312-cp312-linux_x86_64.whl
-     # Install basic dependencies
-     pip install python==3.12.7 torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1
-     pip install flash_attn-2.8.0.post2+cu12torch2.7cxx11abiTRUE-cp312-cp312-linux_x86_64.whl
-     pip install scikit-learn einops huggingface-hub matplotlib networkx numpy pandas scipy tqdm typing_extensions xgboost kditransform hyperopt
-     ```
+# Setup environment for inference
+os.environ["RANK"] = "0"
+os.environ["WORLD_SIZE"] = "1"
+os.environ["MASTER_ADDR"] = "127.0.0.1"
+os.environ["MASTER_PORT"] = "29500"
 
-2. **Model Download**
+# Assuming the LimiX repository is cloned and in the python path
+# from inference.predictor import LimiXPredictor
 
-   * Download model weights via Hugging Face Hub:
+X, y = load_breast_cancer(return_X_y=True)
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5, random_state=42)
 
-     ```python
-     from huggingface_hub import hf_hub_download
-     model_file = hf_hub_download(repo_id="stableai-org/LimiX-16M", filename="LimiX-16M.ckpt", local_dir="./cache")
-     ```
+# Download model weights
+model_file = hf_hub_download(repo_id="stableai-org/LimiX-2M", filename="LimiX-2M.ckpt", local_dir="./cache")
 
-***
+# Initialize predictor (Requires local inference code from GitHub)
+# clf = LimiXPredictor(device='cuda', model_path=model_file, inference_config='config/cls_default_retrieval.json')
+# prediction = clf.predict(X_train, y_train, X_test)
 
-
-
-# 5. Model Usage
-
-1. **Classification Task Example**
-
-   ```python
-   from sklearn.datasets import load_breast_cancer
-   from sklearn.metrics import accuracy_score, roc_auc_score
-   from sklearn.model_selection import train_test_split
-   from huggingface_hub import hf_hub_download
-   import numpy as np
-   import os, sys
-
-   os.environ["RANK"] = "0"
-   os.environ["WORLD_SIZE"] = "1"
-   os.environ["MASTER_ADDR"] = "127.0.0.1"
-   os.environ["MASTER_PORT"] = "29500"
-
-   ROOT_DIR = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
-   if ROOT_DIR not in sys.path:
-       sys.path.insert(0, ROOT_DIR)
-   from inference.predictor import LimiXPredictor
-
-   X, y = load_breast_cancer(return_X_y=True)
-   X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5, random_state=42)
-
-   model_file = hf_hub_download(repo_id="stableai-org/LimiX-16M", filename="LimiX-16M.ckpt", local_dir="./cache")
-
-   clf = LimiXPredictor(device='cuda', model_path=model_file, inference_config='config/cls_default_retrieval.json')
-   prediction = clf.predict(X_train, y_train, X_test)
-
-   print("roc_auc_score:", roc_auc_score(y_test, prediction[:, 1]))
-   print("accuracy_score:", accuracy_score(y_test, np.argmax(prediction, axis=1)))
-   ```
-
-2. **Regression Task Example**
-
-   ```python
-   from functools import partial
-
-   from sklearn.datasets import fetch_california_housing
-   from sklearn.model_selection import train_test_split
-   from sklearn.metrics import r2_score
-   from huggingface_hub import hf_hub_download
-   try:
-       from sklearn.metrics import root_mean_squared_error as mean_squared_error
-   except:
-       from sklearn.metrics import mean_squared_error
-       mean_squared_error = partial(mean_squared_error, squared=False)
-   import os, sys
-
-   os.environ["RANK"] = "0"
-   os.environ["WORLD_SIZE"] = "1"
-   os.environ["MASTER_ADDR"] = "127.0.0.1"
-   os.environ["MASTER_PORT"] = "29500"
-
-   ROOT_DIR = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
-   if ROOT_DIR not in sys.path:
-       sys.path.insert(0, ROOT_DIR)
-   from inference.predictor import LimiXPredictor
-
-   house_data = fetch_california_housing()
-   X, y = house_data.data, house_data.target
-   X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)
-
-   y_mean = y_train.mean()
-   y_std = y_train.std()
-   y_train_normalized = (y_train - y_mean) / y_std
-   y_test_normalized = (y_test - y_mean) / y_std
-
-   model_path = hf_hub_download(repo_id="stableai-org/LimiX-16M", filename="LimiX-16M.ckpt", local_dir="./cache")
-
-   model = LimiXPredictor(device='cuda', model_path=model_path, inference_config='config/reg_default_retrieval.json')
-   y_pred = model.predict(X_train, y_train_normalized, X_test)    
-
-   # Compute RMSE and R²
-   y_pred = y_pred.to('cpu').numpy()
-   rmse = mean_squared_error(y_test_normalized, y_pred)
-   r2 = r2_score(y_test_normalized, y_pred)
-
-   print(f'RMSE: {rmse}')
-   print(f'R2: {r2}')
-   ```
-
-## Ensemble Inference Based on Sample Retrieval
-
-For a detailed technical introduction to Ensemble Inference Based on Sample Retrieval, please refer to the [technical report](https://github.com/limix-ldm/LimiX/blob/main/LimiX_Technical_Report.pdf).
-
-Considering inference speed and memory requirements, ensemble inference based on sample retrieval currently only supports hardware with specifications higher than the NVIDIA RTX 4090 GPU.
-
-### Classification Task
-
-```plain text
-python inference_classifier.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data
+# print("roc_auc_score:", roc_auc_score(y_test, prediction[:, 1]))
+# print("accuracy_score:", accuracy_score(y_test, np.argmax(prediction, axis=1)))
 ```
 
-### Regression Task
-
-```plain text
-python inference_regression.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data
-```
+# 3. Model Architecture & Pretraining
 
-### Customizing Data Preprocessing for Inference Tasks
+LimiX adopts a 12-block transformer architecture with axis-wise attention to features and samples. LimiX-2M specifically utilizes the **RaBEL** framework, which expands each scalar into compact localized RBF features to improve conditioning and shallow-layer effective rank. 
 
-#### First, Generate the Inference Configuration File
+The model is pretrained through **Context-Conditional Masked Modeling (CCMM)**. By masking table cells and conditioning predictions on context rows, the model learns a wide range of conditional dependencies, allowing it to adapt to new datasets without task-specific training.
 
-generate\_inference\_config()
-
-### Classification Task
-
-#### Single GPU or CPU
-
-```plain text
-python  inference_classifier.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data
-```
-
-#### Multi-GPU Distributed Inference
-
-```plain text
-torchrun --nproc_per_node=8  inference_classifier.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data --inference_with_DDP
-```
-
-### Regression Task
-
-#### Single GPU or CPU
-
-```plain text
-python  inference_regression.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data
-```
-
-#### Multi-GPU Distributed Inference
-
-```plain text
-torchrun --nproc_per_node=8  inference_regression.py --save_name your_save_name --inference_config_path path_to_retrieval_config --data_dir path_to_data --inference_with_DDP
-```
-
-### Retrieval Optimization Project
-
-This project implements an optimized retrieval system. To achieve the best performance, we utilize Optuna for hyperparameter tuning of retrieval parameters.
-
-#### Installation
-
-Ensure you have the required dependencies installed:
-
-```plain text
-pip install optuna
-```
-
-#### Usage
+# 4. Deployment
 
-For standard inference using pre-optimized parameters, refer to the code below:
+For manual deployment, install the following dependencies:
 
-```plain text
-searchInference = RetrievalSearchHyperparameters(
-           dict(device_id=0,model_path=model_path), X_train, y_train, X_test, y_test,
-)
-config, result = searchInference.search(n_trials=10, metric="AUC",
-              inference_config='config/cls_default_retrieval.json',task_type="cls")
+```bash
+pip install python==3.12.7 torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1
+pip install scikit-learn einops huggingface-hub matplotlib networkx numpy pandas scipy tqdm typing_extensions xgboost kditransform hyperopt
 ```
 
-This will launch an Optuna study to find the best combination of retrieval parameters for your specific dataset and use case.
-
-***
-
-
-
-# 6. Tool Invocation
-
-The LimiX model can integrate with various toolchains for extended functionality:
-
-* **Data Processing Tools**: Integrates with `pandas` and `scikit-learn` for data cleaning, feature engineering, and result evaluation (e.g., `r2_score`, `mean_squared_error`).
+*Note: Flash Attention 2 is recommended for optimal performance.*
 
-* **Hyperparameter Optimization Tools**: Optimize retrieval parameters via the `hyperopt` library, example as follows:
-
-  ```python
-  # Hyperparameter search example (refer to inference_regression.py)
-  from utils.inference_utils import sample_inferece_params
-  hyperopt_config, base_config = sample_inferece_params(rng, 2, 4)
-  model.set_inference_config(inference_config=hyperopt_config, **base_config)
-  ```
-
-* **Distributed Inference**: Supports DDP (Distributed Data Parallel) mode for multi-GPU acceleration via `torch.distributed`.
-
-***
-
-
-
-# 7. License
-
-1. **Code License**: The repository code is licensed under the \[Apache-2.0 License]\(LICENSE.txt), allowing commercial use and secondary development with retention of the original copyright notice.
+# 5. License
 
+1. **Code License**: The repository code is licensed under the [Apache-2.0 License](LICENSE.txt).
 2. **Model Weight License**: The use of LimiX model weights is subject to a separate Model License:
-
-   * Fully open for academic research without authorization required.
-
-   * Commercial use requires official authorization (refer to the license application process on the [StableAI official website](https://www.stable-ai.ai/)).
-
-***
-
-
-
-# 8. Third-Party Notices
-
-This project uses the following third-party components, whose usage is governed by their respective licenses:
-
-* **PyTorch**: BSD-style license
-
-* **scikit-learn**: BSD license
-
-* **flash-attention**: MIT License
-
-* **Hugging Face Hub**: Apache-2.0 License
-
-* For the complete list of dependencies and license information, refer to `requirements.txt` and the official documentation of each component.
-
-***
-
-
-
-# 9. Contact Us
-
-* **Official Documentation**: <https://www.limix.ai/doc/>
-
-* **GitHub Repository**: <https://github.com/limix-ldm/LimiX> (Submit issues for questions)
-
-* **Official Website**: <https://www.stable-ai.ai/> (For commercial cooperation and license inquiries)
-
-* **Technical Report**: [LimiX: Unleashing Structured-Data Modeling Capability for Generalist Intelligence](https://arxiv.org/abs/2509.03505)
-
-***
+   - Fully open for academic research.
+   - Commercial use requires official authorization from [StableAI](https://www.stable-ai.ai/).
+
+# 6. Citation
+
+If you use LimiX-2M in your research, please cite the following:
+
+```bibtex
+@article{zhang2025limix,
+  title={Limix: Unleashing structured-data modeling capability for generalist intelligence},
+  author={Zhang, Xingxuan and Ren, Gang and Yu, Han and Yuan, Hao and Wang, Hui and Li, Jiansheng and Wu, Jiayun and Mo, Lang and Mao, Li and Hao, Mingchao and others},
+  journal={arXiv preprint arXiv:2509.03505},
+  year={2025}
+}
+
+@article{limix2m2026,
+  title={LimiX-2M: Mitigating Low-Rank Collapse and Attention Bottlenecks in Tabular Foundation Models},
+  author={Zhang, Xingxuan and others},
+  journal={arXiv preprint arXiv:2606.04485},
+  year={2026}
+}
+```