ResearchBenchmarksInstance-aware Caption Generation on MyVLM benchmarkInstance-aware Caption Generation on MyVLM benchmarkFollow27.06CLIP Image SimilarityIIR-VLM24.085624.857825.6326.4022Jan 20, 2026Evaluation resultsMethodMethodLinksCLIP Image SimilarityCLIP Text SimilarityIIR-VLMExamples=1Examples=12026.0127.0669.81MyVLMExamples=4Examples=42026.0125.4262.61MyVLMExamples=2Examples=22026.0124.9161.01MyVLMExamples=1Examples=12026.0124.257.37