obversarystudios/failure-geometry-demo
0
1�
2 �`�i� � �j � d Z ddlmZ ddlZddlmZ ddlmZ ddl m3Z4 dZ d d�Z e f d d �Z
d
d5�Zy)a� 6TF-IDF + Truncated SVD embedding for failure records.7 8No network downloads; no GPU required. Runs entirely with scikit-learn.9 10Feature construction combines:11- TF-IDF on a failure-context string (input text, expected label, prediction, reasoning type)12- TruncatedSVD to produce a dense, low-dimensional representation13 14Adapted from failure-induced-benchmarks/src/failure_geometry/embedding.py15(engineered sparse features) but extended to dense SVD projections for16KMeans and scatter visualisation.17� )� annotationsN)� TruncatedSVD)�TfidfVectorizer)� normalize� c �B � d| d � d| d � d| d � d| d � d | d18 � �19S ) Nzinput: �xz expected: �yz
prediction: �20predictionz type: �reasoning_typez model: �model_id� )�failures �d/mnt/ai_lab_os/AI_LAB/infra/09_practice/github_ready/huggingface/failure-geometry-demo/core/embed.py�
_failure_textr sV � �21�'�#�,� � � �S�\�N� #� �|�,�-� .��)�*�+� ,��*�%�&� (�� c � � | st j d|f� S | D �cg c]
}t |� �� }}t |� }t ddd�� }|j |� }t
||j d dz22 |dz23 � }|dk r.|j � }t |d d �d t |d� �f � S t |d� � } | j |� }24t |25� S c c}w )26Nr i )� � T)� max_features� ngram_range� sublinear_tfr r �* )� n_components� random_state) �np�emptyr �lenr �
fit_transform�min�shape�toarrayr �maxr ) �failuresr �f�texts�n�27vectorizer�tfidf�effective_dims�arr�svd�denses r �embed_failuresr. $ s� � � ��x�x��L�)�*�*�'/� 0�!�]�1�
� 0�E� 0� �E�28�A� �����J�29
� $� $�U� +�E�� �u�{�{�1�~��'9�1�q�5�A�N�����m�m�o����Q� 8�#�n�a�"8� 8� 8�9�:�:�30�N��31D�C� � � �e� $�E� �U� ���#
1s �Cc �x � t | � dk r t j t | � df� S t | d�� S )z8Return exactly 2 components for PCA-style scatter plots.� r )r )r r �zerosr. )r$ s r �embed_for_scatterr2 ? s3 � �32�8�}�q���x�x��X� ��*�+�+� �(�� 3�3r )r �dict�return�str)r$ �33list[dict]r �intr4 �34np.ndarray)r$ r6 r4 r8 )�__doc__�35__future__r �numpyr �sklearn.decompositionr � sklearn.feature_extraction.textr �sklearn.preprocessingr �36EMBED_DIMSr r. r2 r r r �<module>r@ sP �� � #� � .� ;� +� �37�� #� �� �� � � �64r