Narrative Similarity on SemEval Task 4 Track B 2026 (test)
68.75Similarity ScoreTrack B multi-view pipeline
Evaluation Results
| Method | Links | |
|---|---|---|
| Track B multi-view pipelineEmbedding Model=all-mpnet-base-v2, LLM Extractor=o3-mini2026.06 | 68.75 | |
| Track B multi-view pipelineEmbedding Model=LaBSE, LLM Extractor=gpt-4o-mini2026.06 | 65.41 | |
| Track B multi-view pipelineEmbedding Model=LaBSE, LLM Extractor=gpt-4.12026.06 | 64.66 | |
| Track B multi-view pipelineEmbedding Model=all-mpnet-base-v2, LLM Extractor=gpt-4.12026.06 | 63.91 | |
| Track B multi-view pipelineEmbedding Model=all-mpnet-base-v2, LLM Extractor=gpt-4o-mini2026.06 | 62.91 | |
| Track B multi-view pipelineEmbedding Model=LaBSE, LLM Extractor=o3-mini2026.06 | 62.91 | |
| Track B multi-view pipelineEmbedding Model=para-mul-mpnet-v2, LLM Extractor=gpt-4.12026.06 | 59.4 | |
| Track B multi-view pipelineEmbedding Model=para-mul-mpnet-v2, LLM Extractor=o3-mini2026.06 | 59.15 | |
| Track B multi-view pipelineEmbedding Model=all-MiniLM-L6-v2, LLM Extractor=gpt-4.12026.06 | 56.64 | |
| Track B multi-view pipelineEmbedding Model=para-mul-mpnet-v2, LLM Extractor=gpt-4o-mini2026.06 | 56.14 | |
| Track B multi-view pipelineEmbedding Model=all-MiniLM-L6-v2, LLM Extractor=o3-mini2026.06 | 54.14 | |
| Track B multi-view pipelineEmbedding Model=all-MiniLM-L6-v2, LLM Extractor=gpt-4o-mini2026.06 | 53.38 |