Add MIT licence and a notice separating code from data

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Vishal Khan
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

NOTICE.md

@@ -0,0 +1,27 @@
+# Notice
+
+Two different licences apply here, and they do not cover the same things.
+
+## Code
+
+Everything in this repository is MIT licensed. See [LICENSE](LICENSE).
+Copyright (c) 2026 Vishal Khan.
+
+## Data
+
+The HAM10000 dataset is **not** in this repository and is **not** MIT licensed.
+No image is committed here. `uv run lesion-split fetch` downloads the corpus at
+runtime into a gitignored directory on your own machine, under the dataset's own
+terms rather than this repository's.
+
+HAM10000 is licensed **CC BY-NC 4.0**, which permits non-commercial use only.
+Harvard Dataverse version 4.0, `doi:10.7910/DVN/DBW86T`.
+
+Tschandl, P., Rosendahl, C. and Kittler, H. The HAM10000 dataset, a large
+collection of multi-source dermatoscopic images of common pigmented skin
+lesions. *Scientific Data* 5, 180161 (2018).
+
+The MIT grant on the code does not extend to the data, and the non-commercial
+restriction on the data does not restrict the code. If you run the training
+arms you are using CC BY-NC 4.0 material and its terms apply to what you do
+with it.

README.md

@@ -221,6 +221,9 @@ Nothing downloaded is committed. Non-commercial terms and a public repository do
 not mix, so the images land in a gitignored directory on the reader's machine
 under whatever terms they accepted.
 
+The code in this repository is MIT licensed and the dataset is not. [NOTICE.md](NOTICE.md)
+says which licence covers what.
+
 Kaggle hosts this dataset in at least a dozen forms: balanced, augmented, hair
 removed, resized to 28 by 28. Training on a pre-augmented copy leaks by
 construction, because derived images of one original sit on both sides of any