Dependencies upgrade (#1264)

* Upgrade NLTK

* Upgrade dep

* Update env matrix

* update dep, bump version

* Update py version

* Rmv lock

* Add lock

* Lint

Author: ZoranPandovski

.github/workflows/doc_build.yml

@@ -20,7 +20,7 @@ jobs:
       - name: setup python
         uses: actions/setup-python@v2
         with:
-          python-version: 3.9
+          python-version: 3.11
 
       - name: install all dependencies
         run: |

.github/workflows/lightwood.yml

@@ -14,7 +14,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-latest]
-        python-version: ["3.9", "3.10", "3.11"]
+        python-version: ["3.10", "3.11", "3.12"]
     steps:
       - uses: actions/checkout@v2
       - name: Set up Python ${{ matrix.python-version }}

lightwood/analysis/nc/nc.py

@@ -265,9 +265,9 @@ def apply_inverse(self, nc, significance):
 
 
 class TSAbsErrorErrFunc(TSErrFunc):
-    """Calculates absolute error nonconformity for time series problems.
+    r"""Calculates absolute error nonconformity for time series problems.
 
-        For each forecasted step ``y_h`` for h \in 1..horizon, nonconformity is defined as
+        For each forecasted step ``y_h`` for h \\in 1..horizon, nonconformity is defined as
 
         .. math::
             | y_ni - hat{y}_ni |

lightwood/encoder/text/pretrained.py

@@ -282,7 +282,10 @@ def _tune_model(self, train_dataset, val_dataset, optim, scheduler, n_epochs=1):
                     for vbatch in val_dataset:
                         val_loss += self._call(vbatch).item()
 
-                    log.info(f"Epoch {epoch+1} train batch {bidx+1} - Validation loss: {val_loss/len(val_dataset)}")
+                    log.info(
+                        f"Epoch {epoch + 1} train batch {bidx + 1} - "
+                        f"Validation loss: {val_loss / len(val_dataset)}"
+                    )
                     if val_loss / len(val_dataset) >= best_val_loss:
                         break
 
@@ -318,7 +321,7 @@ def _call(self, batch):
         return loss
 
     def _train_callback(self, epoch, loss):
-        log.info(f"{self.name} at epoch {epoch+1} and loss {loss}!")
+        log.info(f"{self.name} at epoch {epoch + 1} and loss {loss}!")
 
     def encode(self, column_data: Iterable[str]) -> torch.Tensor:
         """

lightwood/ensemble/ts_stacked_ensemble.py

@@ -29,7 +29,7 @@ def __init__(self, target, mixers: List[BaseMixer], data: EncodedDs, dtype_dict:
             raise Exception('This ensemble can only be used to forecast!')
         self.ts_analysis = ts_analysis
         self.horizon = self.ts_analysis['tss'].horizon
-        self.target_cols = [target] + [f'{target}_timestep_{t+1}' for t in range(self.horizon - 1)]
+        self.target_cols = [target] + [f'{target}_timestep_{t + 1}' for t in range(self.horizon - 1)]
         self.agg_dim = 2
         self.opt_max_iter = 1000
 

poetry.lock

@@ -4,17 +4,17 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "lightwood"
-version = "25.3.3.3"
+version = "25.5.2.1"
 description = "Lightwood is Legos for Machine Learning."
 authors = ["MindsDB Inc."]
 license = "GPL-3.0-only"
 readme = "README.md"
 
 [tool.poetry.dependencies]
-python = ">=3.9,<3.12"
-type_infer = ">=0.0.15"
-dataprep_ml = ">=24.5.1.2,<24.6.1.0"
-mindsdb-evaluator = ">=0.0.13"
+python = ">=3.10,<3.13"
+type_infer = "0.0.22"
+dataprep_ml = "0.0.24"
+mindsdb-evaluator = "0.0.17"
 numpy = ">1.23.0"
 nltk = ">=3.9"
 pandas = ">=2.0.0, <2.1.0"

pyproject.toml

@@ -82,5 +82,5 @@ def test_overfit(self):
         results = torch.isclose(answer, correct_answer, atol=1)
         acc = (results.sum() / results.numel()).item()
 
-        print(f'Transformer correctly reconstructed {round(100*acc, 2)}%')
+        print(f'Transformer correctly reconstructed {round(100 * acc, 2)}%')
         assert acc >= 0.5