Merge pull request #52 from martinscooper/first_model_negative_threshold-config

Add first_model_negative_threshold to the documentation

Author: yannisk2

docs/docs/dev/configuration.md

@@ -26,7 +26,8 @@ The following parameters can be set in the configuration file:
 | Parameter | Description |
 |-----------|-------------|
 | `first_model_positive_threshold`  | Number of elements that must be assigned a positive label for the category in order to trigger the training of a classification model. <br /> <br /> _See also:_ The [training invocation](model_training.md#training-invocation) documentation. |
-| `changed_element_threshold`       | Number of changes in user labels for the category -- relative to the last trained model -- that are required to trigger the training of a new model. A change can be a assigning a label (positive or negative) to an element, or changing an existing label. Note that `first_model_positive_threshold` must also be met for the training to be triggered. <br /> <br /> _See also:_ The [training invocation](model_training.md#training-invocation) documentation. |
+| `first_model_negative_threshold`  | Number of elements that must be assigned a negative label for the category in order to trigger the training of a classification model. <br /> <br /> _See also:_ The [training invocation](model_training.md#training-invocation) documentation. |
+| `changed_element_threshold`       | Number of changes in user labels for the category -- relative to the last trained model -- that are required to trigger the training of a new model. A change can be a assigning a label (positive or negative) to an element, or changing an existing label. Note that both, `first_model_positive_threshold` and `first_model_negative_threshold`, must also be met for the training to be triggered. <br /> <br /> _See also:_ The [training invocation](model_training.md#training-invocation) documentation. |
 | `training_set_selection_strategy` | Strategy to be used from [TrainingSetSelectionStrategy](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/training_set_selector/train_set_selector_api.py). A TrainingSetSelectionStrategy determines which examples will be sent to the classification models at training time - these will not necessarily be identical to the set of elements labeled by the user. For currently supported implementations see the [get_training_set_selector()](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/training_set_selector/training_set_selector_factory.py) function. <br /> <br /> _See also:_ The [training set selection](model_training.md#training-set-selection) documentation. |
 | `model_policy`                    | Policy to be used from [ModelPolicies](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/models/core/model_policies.py). A [ModelPolicy](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/models/policy/model_policy.py) determines which type of classification model(s) will be used, and _when_ (e.g. always / only after a specific number of iterations / etc.). <br /> <br /> _See also:_ The [model selection](model_training.md#model-selection) documentation. |
 | `active_learning_strategy`        | Strategy to be used from [ActiveLearningCatalog](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/active_learning/core/catalog.py). An [ActiveLearner](https://github.com/label-sleuth/label-sleuth/blob/main/label_sleuth/active_learning/core/active_learning_api.py) module implements the strategy for recommending the next elements to be labeled by the user, aiming to increase the efficiency of the annotation process. <br /> <br /> _See also:_ The [active learning](active_learning.md) documentation. |

docs/docs/dev/model_training.md

@@ -17,6 +17,7 @@ Label Sleuth starts a new model training iteration whenever the following two co
 |   Condition on | Description | Default |
 |---|---|---|
 | **Number of positive labels** | The user has to provide a minimum number of positive labels. The threshold of required positive labels can be configured by setting the value of the `first_model_positive_threshold` parameter in the system's [configuration file](configuration.md). | 20 |
+| **Number of negative labels** | The user has to provide a minimum number of negative labels. The threshold of required negative labels can be configured by setting the value of the `first_model_negative_threshold` parameter in the system's [configuration file](configuration.md). | 0 |
 | **Number of label changes** | The user has to change a minimum number of labels since the last model training iteration (unless it is the first iteration). A change can be assigning a label (positive or negative) to an element, or changing an existing label. The threshold of required label changes can be configured by setting the value of the `changed_element_threshold` parameter in the system's [configuration file](configuration.md). | 20 |
 
 ## Training set selection