test_config.yaml

pipeline_settings:
  - __impl__: PipelineSettings
    combine_gnn_strategy: none
    norm_fn: none
    skip_graph_creation: False

semantic_graph_context_generator:
  - __impl__: SemanticGraphContextGenerator
    generator_id: self-instruction
    model_id: gpt-3.5-turbo
    temperature: 0
    max_tokens: 500
    with_tree: False

gnn_heads:
  - __impl__: GNNHead
    generator_id: message_passing
    k_hop: 2
    is_sparse: True
    embbedding_id: sbert
    embbedding_model: all-MiniLM-L6-v2
    kernel_id: cosine_similarity
    graph_id: adjacency
    connection_threshold: 0.7

topic_model:
  - __impl__: TopicModel
    topic_model_id: topic_model
    embedding_model: all-MiniLM-L6-v2
    clustering_model:
      HDBSCAN(min_cluster_size=15, metric='euclidean',
      cluster_selection_method='eom', prediction_data=True)
    dimensionality_reduction:
      umap_model = UMAP(n_neighbors=15, n_components=5, min_dist=0.0,
      metric='cosine')
    vectorizer_model: CountVectorizer(stop_words="english")
    representation_model:
      OpenAI(model="gpt-3.5-turbo", delay_in_seconds=3, chat=True)
    hierarchical_linkage_fn: sch.linkage(x, 'single', optimal_ordering=True)
    with_tree: True
    auto_cluster: True