CoRoVA compresses repository context into compact vectors for code LLMs, reducing TTFT 20-38% versus uncompressed RAG with only a small projector module.
Because these continuous tokens reconstruct to reference texts, we treat them as ground truth for training our projection layer
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CoRoVA: Compressed Representations for Vector-Augmented Code Completion
CoRoVA compresses repository context into compact vectors for code LLMs, reducing TTFT 20-38% versus uncompressed RAG with only a small projector module.