Search papers, labs, and topics across Lattice.
This paper introduces a novel benchmark and efficient bi-encoder for natural language code retrieval specifically tailored for the 1C:Enterprise ecosystem, which is characterized by its unique Russian syntax and domain-specific terminology. By fine-tuning on a large synthetic dataset generated from public code repositories, the authors achieve significant improvements in retrieval performance, with their model surpassing baseline architectures in balanced macro nDCG@10 and other metrics. The results demonstrate that their approach effectively addresses the challenges posed by limited labeled data while maintaining high retrieval quality with reduced dimensionality.
Fine-tuning on synthetic data allows for a dramatic boost in code retrieval performance, achieving a balanced macro nDCG@10 of 0.5992 in a challenging domain.
Natural language code retrieval is a rapidly evolving task in computer science. However, the 1C:Enterprise ecosystem combines Russian syntax with highly domain-specific terminology, for which open datasets and specialized models have been virtually non-existent. We present a comprehensive pipeline for 1C code retrieval: an open benchmark of 3,413 real-world, PII-scrubbed query-code pairs, a reproducible evaluation harness, and a specialized bi-encoder. To overcome scarce labeled data, we fine-tune on 784,057 synthetic triplets generated by google/gemma-4-26B-A4B-it from public code repositories, using Matryoshka Representation Learning (MRL) and a privacy-aware tokenizer. Because the benchmark subsets differ in size, we report balanced-subset macro, query-weighted micro, and forum-only results. Our model reaches 0.5992 balanced macro nDCG@10, 0.5044 micro, and 0.4617 on forum, versus 0.4932 macro for the baseline architecture and 0.5404 for google/embeddinggemma-300m. Removing every benchmark example flagged by the conservative exact/13-gram overlap audit leaves 0.6011 balanced macro (0.5010 micro), indicating that detected train-benchmark overlap does not explain the headline result. MRL truncation to 256 dimensions preserves 99.9% of retrieval quality while reducing dense-index storage and exact similarity arithmetic by a factor of three.
