Causilo 1.0.0

.gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,40 @@
+Causilo License v1.0
+
+Nums AI Inc. ("Company") makes Causilo available under this license ("License"). "Model" means the Causilo weights, parameters, checkpoints, code, and related materials Company makes available under this License. Separately licensed components remain subject to their respective licenses.
+
+By accessing, using, modifying, or Distributing the Model or any Derivative, whether received from Company or a third party, you agree to this License. If acting for an entity, you represent that you have authority to bind it. If you do not agree, you may not exercise the rights granted here.
+
+1. Definitions
+a. "Derivative" means a modified version of the Model, including fine-tuned, retrained, quantized, or pruned versions, learned adapters (including LoRA weights), weight deltas, and any work incorporating or derived from the Model's weights, parameters, or code. A model trained or adapted by distilling the Model or any Derivative, including through their Outputs, is also a Derivative. Outputs and independently authored papers, research results, or code are not Derivatives merely because they use, evaluate, or describe the Model, provided they do not contain Model components. This exclusion does not apply to distilled models.
+b. "Distribution" or "Distribute" means providing or making available, by any means, a copy of the Model or a Derivative.
+c. "Non-Commercial Purpose" means research, testing, evaluation, or experimentation not directed toward commercial gain, revenue generation, commercial product development, business decision-making, client deliverables, or production deployment. This includes new research, benchmarking, replication, and reproducibility. Institutional affiliation, salaries, and research grants do not by themselves make an activity commercial. Research collaboration and hosting or downloading files for Distribution under Section 3 do not by themselves constitute production deployment. Payment for general-purpose computing or storage does not by itself make otherwise permitted research commercial.
+d. "Outputs" means predictions, scores, explanations, or other results generated by the Model or a Derivative, excluding Model or Derivative components such as weights or learned adapters.
+
+2. License Grant and Use
+a. Subject to compliance with this License, Company grants you, to the extent of its licensable rights, a non-exclusive, worldwide, non-transferable, non-sublicensable, irrevocable (except under Section 7), royalty-free license to access, use, copy, modify, and create Derivatives of the Model and lawfully received Derivatives for Non-Commercial Purposes, and to Distribute them under Section 3. All rights not expressly granted are reserved. You may not assign or sublicense Company's rights without its written consent; compliant Distribution under Section 3 is not an assignment or sublicense.
+b. Except as permitted in Section 2(c), commercial or production use of the Model, Derivatives, or Outputs, including use in paid products or services or to train, fine-tune, or distill models for commercial use, requires a separate license from Company and any other relevant rights holders. You must not knowingly permit, assist, or cause any third party to engage in such use without the required licenses.
+c. Notwithstanding Sections 1(c) and 2(b), you may publish, present, and share scholarly papers and research results, including Outputs generated in compliance with this License, for scholarly communication, including through academic venues or publishers that charge publication, conference, or subscription fees. Company claims no ownership in your independently authored papers or Outputs. Such papers and results need not be licensed under this License; any included Model or Derivative components remain subject to Section 3. This permission does not authorize commercial or production use of Outputs outside scholarly communication.
+d. You must comply with applicable law when using or Distributing the Model, Derivatives, or Outputs.
+e. You may not provide, host, or make available the Model or any Derivative as part of a hosted, managed, API, or SaaS service, whether paid or free, without a separate license from Company and any other relevant rights holders. This restriction does not prohibit hosting downloadable model files for Distribution under Section 3 or using general-purpose computing resources to run your own permitted activities without providing such a service.
+
+3. Research Redistribution
+You may Distribute unmodified Model copies and Derivatives you create or lawfully receive, in whole or in part, solely for Non-Commercial Purposes supporting research, evaluation, reproducibility, peer review, or scholarly archiving, including through public repositories. No paper or prior Company permission is required, provided that:
+a. You include a copy of this License, retain copyright, attribution, and disclaimer notices, and identify the original Model and its version, if supplied.
+b. You clearly identify modifications, the person or entity responsible for them, and a brief description of the changes, retaining previous modification notices.
+c. Company directly offers every recipient the same rights in the Model, including as incorporated in a Derivative, under this License. For your contributions to a distributed Derivative, you must have authority to grant, and hereby grant, every direct or indirect recipient of those contributions the same permissions, subject to the same restrictions and conditions, as this License grants for the Model. You must preserve permissions and notices from prior contributors and must not impose terms restricting the rights granted under this License.
+d. You Distribute without charge and do not condition receipt on purchasing another product or service.
+This License does not require you to publish any Model, Derivative, paper, code, or training data.
+
+4. Disclaimer of Warranties
+TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, THE MODEL AND DERIVATIVES ARE PROVIDED "AS IS", WITHOUT WARRANTIES OF ANY KIND, INCLUDING MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
+
+5. Limitation of Liability
+TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER COMPANY NOR CONTRIBUTORS SHALL BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT, OR OTHERWISE, ARISING FROM OR IN CONNECTION WITH THE MODEL, DERIVATIVES, OR THEIR USE OR DISTRIBUTION.
+
+6. Trademarks
+You may use the names "Nums AI Inc." and "Causilo" for accurate attribution, citation, and description. No other trademark rights are granted, and you may not imply Company's endorsement of a Derivative or use Company's logos without its written permission.
+
+7. Termination
+Your rights under this License terminate automatically if you breach it. They are automatically reinstated if you cure the breach within 30 days after discovering it; otherwise, reinstatement requires the relevant rights holder's written approval. Reinstatement does not affect remedies for the breach.
+
+While your rights are terminated, you must cease use, modification, and Distribution of the Model and Derivatives. You may retain copies solely for legal compliance or institutional archiving, without further use or Distribution. Termination does not affect the rights of recipients who remain in compliance, or the permissions in Section 2(c) for papers and results generated in compliance before termination. Restrictions, notice obligations, disclaimers, and limitations of liability survive termination.

README.md

@@ -0,0 +1,56 @@
+---
+license: other
+license_name: causilo-1.0
+license_link: LICENSE
+library_name: causilo
+extra_gated_prompt: "Use is subject to Causilo License v1.0. Non-commercial research and free research redistribution are permitted under its conditions. Commercial or production use, and hosted/API/SaaS services whether paid or free, require separate licenses."
+extra_gated_fields:
+  I accept the model license: checkbox
+extra_gated_button_content: "Accept and access"
+---
+
+# Causilo
+
+Causilo is a pretrained tabular foundation model from Nums AI Inc., supporting classification and regression through a scikit-learn interface.
+
+## Files
+
+- `classifier/config.json` and `classifier/model.safetensors`: classification model.
+- `regressor/config.json` and `regressor/model.safetensors`: regression model.
+
+## Installation
+
+Python 3.10–3.12 and PyTorch 2.13+ are required.
+
+```bash
+pip install causilo
+hf auth login
+```
+
+Accept the license on this page, then authenticate with the same Hugging Face account. The first fit downloads and caches the checkpoint.
+
+```python
+from causilo import CausiloClassifier, CausiloRegressor
+
+classifier = CausiloClassifier()
+classifier.fit(X_train, y_train)
+probabilities = classifier.predict_proba(X_test)
+
+regressor = CausiloRegressor()
+regressor.fit(X_train, y_train)
+predictions = regressor.predict(X_test)
+```
+
+See the [code repository](https://github.com/nums-ai/causilo) for usage and [benchmarks](https://github.com/nums-ai/causilo#benchmarks). The architecture is `causilo-v1.0`; checkpoint and configuration formats are version 1.
+
+## Licenses
+
+- Python source and wheel: Apache-2.0; see [the code repository](https://github.com/nums-ai/causilo).
+- Model weights: [Causilo License v1.0](LICENSE), including the checkpoint revision pinned by the package.
+- Non-commercial research, testing, evaluation, experimentation and modifications are permitted subject to the License. Institutional affiliation alone does not determine whether a use is non-commercial.
+- Free research redistribution of original copies and Derivatives is permitted under Section 3 without prior Company permission, with the required license, notices and modification information.
+- Commercial or production use of the Model, Derivatives or Outputs requires separate licenses, subject to the scholarly-communication permission in Section 2(c).
+- Hosted, managed, API or SaaS services require separate licenses whether paid or free. Hosting downloadable files under Section 3 is permitted.
+- These points summarize the License; the full text controls.
+
+Licensor: Nums AI Inc. Contact: contact@nums.world.

classifier/config.json

@@ -0,0 +1,27 @@
+{
+  "architecture": "causilo-v1.0",
+  "format_version": 1,
+  "config_version": 1,
+  "model": {
+    "task": "classification",
+    "width": 128,
+    "expansion": 2,
+    "group_size": 3,
+    "frequencies": 16,
+    "column_latents": 128,
+    "column_heads": 4,
+    "column_depths": [
+      3,
+      3
+    ],
+    "row_heads": 8,
+    "row_latents": 4,
+    "row_depths": [
+      3,
+      3
+    ],
+    "prediction_heads": 4,
+    "prediction_depth": 12,
+    "outputs": 10
+  }
+}

classifier/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fe50487b33ef6b4ed65d1dab1e48888b86f9ed604317a4b596fea4dd462b7f6
+size 144385448

regressor/config.json

@@ -0,0 +1,27 @@
+{
+  "architecture": "causilo-v1.0",
+  "format_version": 1,
+  "config_version": 1,
+  "model": {
+    "task": "regression",
+    "width": 128,
+    "expansion": 2,
+    "group_size": 3,
+    "frequencies": 16,
+    "column_latents": 128,
+    "column_heads": 4,
+    "column_depths": [
+      3,
+      3
+    ],
+    "row_heads": 8,
+    "row_latents": 4,
+    "row_depths": [
+      3,
+      3
+    ],
+    "prediction_heads": 4,
+    "prediction_depth": 12,
+    "outputs": 999
+  }
+}

regressor/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a4197d7c1060d187c19b62d4d83c74ce986c1f3911825576eb58c2ae4f0057d
+size 148417316