Causalnlp

New:
- N/A

Changed
- updated dependencies

Fixed:
- N/A

New:
- Added `model_name` parameter to `CausalBertModel` to support other DistilBert models (e.g., multilingual)

Changed
- N/A

Fixed:
- N/A

New:
- Added support for `CausalBert`

Changed
- Added `p` parameter to `CausalInferenceModel.fit` and `CausalInferenceModel.predict` for user-supplied propensity scores in X-Learner and R-Learner.
- Removed CV from propensity score computations in X-Learner and R-Learner and increase default `max_iter` to 10000

Fixed:
- Resolved problem with `CausalInferenceModel.tune_and_use_default_learner` when outcome is continuous
- Changed to `max_iter=10000` for default `LogisticRegression` base learner

New:
- N/A

Changed
- Use `LinearRegression` and `LogisticRegression` as default base learners for `s-learner`.
- changed parameter name of `metalearner_type` to `method` in `CausalInferenceModel`.

Fixed:
- Resolved mis-references in `_balance` method (renamed from `_minimize_bias`).
- Fixed convergence issues and factored out propensity score computations to `CausalInferenceModel.compute_propensity_scores`.

New:
- N/A

Changed
- N/A

Fixed:
- Added `sample_size` parameter to `CausalInferenceModel.evalute_robustness`

New:
- Added `CausalInferenceModel.evaluate_robustness` method to assess robustness of causal estimates using sensitivity analysis

Changed
- reduced dependencies with local metalearner implementations

Fixed:
- N/A