sync with AZ internal version

.gitattributes

@@ -1 +1,2 @@
-*.prior binary
+*.prior filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text

CHANGELOG.md

@@ -6,10 +6,120 @@ This follows the guideline on [keep a changelog](https://keepachangelog.com/)
 ## [Unreleased]
 
 ### Changed
-- Fragment generators using transformers
 - CAZP scoring component
 
 
+## [4.3.5] 2024-04-18
+
+### Changed
+
+- Code clean-up in create\_adapter()
+
+### Fixed
+
+- Import mol2mol vocabulary rather than copying the file
+
+
+## [4.3.4] 2024-04-16
+
+### Added
+
+- Write invalid SMILES unchanged to RL CSV
+
+
+## [4.3.3] 2024-04-16
+
+### Added
+
+- Notebook: demo on how to analyse RL CSV
+
+
+## [4.3.2] 2024-04-15
+
+### Added
+
+- Dataclass validation for scoring component parameters
+
+### Fixed
+
+- Datatype in MatchingSubstructure's Parameters: only single SMARTS is allowed
+
+
+## [4.3.1] 2024-04-15
+
+### Added
+
+- Notebook to demo simple RL run, TensorBoard visualisation and TensorBoard data extraction.
+
+
+## [4.3.0] 2024-04-15
+
+### Added
+
+- Linkinvent based on unified Transformer model supported by RL and sampling.  Both beam search and multinomial sampling are implemented.
+
+
+## [4.2.13] 2024-04-12
+
+### Fixed
+
+- downgraded Chemprop to 1.5.2 and sklearn to 1.2.2 to retain backward compatibility
+
+
+## [4.2.12] 2024-04-10
+
+### Changed
+
+- New default torch device setup from PyTorch 2.x
+
+### Added
+
+- Config parameter "device" to explicitly set torch device e.g. "cuda:0"
+
+
+## [4.2.11] 2024-04-08
+
+### Fixed
+
+- Fixed unknown token handling for Mol2mol TL
+
+
+## [4.2.10] 2024-04-05
+
+### Fixed
+
+- Fixed dataloader for TL to use incomplete batch 
+
+
+## [4.2.9] 2024-04-04
+
+### Fixed
+
+- Skip hash check in metadata if no metadata in model file
+
+
+## [4.2.8] 2024-04-03
+
+### Added
+
+- Mol2Mol supports unknown tokens for all the priors
+
+
+
+## [4.2.7] 2024-03-27
+
+### Added
+
+- Optional randomization in all TL epochs for Reinvent
+
+
+## [4.2.6] 2024-03-21
+
+### Fixed
+
+- Return from make\_grid\_image()
+
+
 ## [4.2.5] 2024-03-20
 
 ### Added

NEWS.md

@@ -1,3 +1,20 @@
+New in REINVENT 4.3
+===================
+
+For details see CHANGELOG.md.
+
+* Upgrade to **PyTorch 2.2**: rerun `pip install -r requirements-linux-64.lock`
+* 2 new **notebooks** demoing Reinvent with reinforcement learning and also transfer learning, includes TensorBoard visualisation and basic analysis
+* New Linkinvent model code based on unified transformer
+* New PubChem Mol2Mol prior
+* Unknown token support for PubChem based transformer models
+* New "device" config parameter to allow for explicit device e.g. "cuda:0"
+* Optional SMILES randomization in every TL epoch for Reinvent
+* Dataclass parameter validation for most scoring components
+* Invalid SMILES are now written to the reinforcement learning CSV
+* Code improvements and fixes
+
+
 New in REINVENT 4.2
 ===================
 

README.md

@@ -57,7 +57,7 @@ Installation
     ```
 4. Optional: if you want to use **AMD GPUs** on Linux you would need to install the [ROCm PyTorch version](https://pytorch.org/get-started/locally/) manually _after_ installation of the dependencies in point 3, e.g.
    ```shell
-   pip install torch==1.13.1+rocm5.2 torchvision==0.14.1+rocm5.2 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/rocm5.2
+   pip install torch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 --index-url https://download.pytorch.org/whl/rocm5.7
    ```
 5. Install the tool. The dependencies were already installed in the previous step, so there is no need to install them again (flag `--no-deps).  If you want to install in editable mode (changes to the code are automatically picked up) add -e before the dot.
     ```shell
@@ -90,30 +90,31 @@ appropriate run mode depending on the research problem you are trying to address
 There is additional information in `config/toml` in several `*.md` files with
 instructions on how to configure the TOML file.
 
-<!--- For concrete examples, you can check out the Jupyter notebook examples in the ReinventCommunity repo.
-Running each example will result in a template file.There are templates for many running modes. 
-Each running mode can be executed by `python input.py some\_running\_mode.json` after activating the environment.
-    
-Templates can be manually edited before using. The only thing that needs modification for a standard run are the file 
-and folder paths. Most running modes produce logs that can be monitored by tensorboard. --->
-
 
 Tutorials / `Jupyter` notebooks
 -------------------------------
 
-NOTE: these will be updated at a later time!
+Basic instructions can be found in the comments in the config examples in `config/toml`.
+
+Notebooks will be provided in the `notebook/` directory.  Please note that we provide the notebooks in jupytext "light script" format.  To work with the light scripts you will need to install jupytext.  A few other packages will come in handy too.
+
+```shell
+pip install jupytext mols2grid seaborn
+```
+
+The Python files in `notebook/` can then be converted to a notebook e.g.
 
-<!--- There is another repository containing useful `jupyter` notebooks related to `REINVENT` 
-called [ReinventCommunity](https://github.com/MolecularAI/ReinventCommunity). Note, that it uses a
-different `conda` environment to execute, so you have to set up a separate environment. --->
+```shell
+jupytext --to ipynb -o Reinvent_demo.ipynb Reinvent_demo.py
+```
 
 
 Updating dependencies
 ---------------------
 
 Update the lock files with [pip-tools](https://pypi.org/project/pip-tools/) (please, do not edit the files manually):
 ```shell
-pip-compile --extra-index-url=https://download.pytorch.org/whl/cu113 --extra-index-url=https://pypi.anaconda.org/OpenEye/simple --resolver=backtracking pyproject.toml
+pip-compile --extra-index-url=https://download.pytorch.org/whl/cu121 --extra-index-url=https://pypi.anaconda.org/OpenEye/simple --resolver=backtracking pyproject.toml
 ```
 To update a single package, use `pip-compile --upgrade-package somepackage`
 (see the documentation for pip-tools).