ወደ ዜና ተመለስ
ፈጠራAI Understanding አጭር መግለጫ

Natural Language Code Retrieval for 1C:Enterprise: An Open Benchmark and Efficient Bi-Encoder

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.

5 min readRead the primary source
Source-page capture accompanying Natural Language Code Retrieval for 1C:Enterprise: An Open Benchmark and Efficient Bi-Encoder
ዋና-ምንጭ ሰነድምንጭ ተመዝግቧል
አታሚ
arxiv.org
ምንጭ አገናኝ
arxiv.orghttps://arxiv.org/abs/2608.19957
የምንጭ ዓይነት
ዋና ሰነድ - ኦፊሴላዊ ማስታወቂያ ፣ ወረቀት ፣ ፋይል ወይም የመጀመሪያ ወገን ገጽ በቀጥታ እናነባለን።
አውድይህንን በ60 ሰከንድ ውስጥ ይረዱት።

እዚ ጀምር

ቁልፍ ቃላት

ቢ-ኢንኮደር
መጠይቆችን እና ሰነዶችን ወደ ተለያዩ ቬክተሮች የሚያስገባ ሞዴል በፍጥነት በመጠን ሊነጻጸሩ ይችላሉ።
ቤንችማርክ
የሞዴል አፈጻጸምን ለመለካት እና ለማነፃፀር የሚያገለግል ደረጃውን የጠበቀ ሙከራ ወይም የውሂብ ስብስብ።
መልሶ ማግኘት
ተዛማጅ ሰነዶችን ወይም መዝገቦችን ከእውቀት ምንጭ ለጥያቄ መፈለግ።
እራስህን ፈትን።AI ምንድን ነው? ጥያቄ

ምን ተፈጠረ

The authors present a comprehensive pipeline for 1C code : an open of 3,413 real-world, PII-scrubbed query-code pairs, a reproducible evaluation harness, and a specialized . To overcome scarce labeled data, they 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.

The authors present a comprehensive pipeline for 1C code , consisting of an open , a reproducible evaluation harness, and a specialized .

The open consists of 3,413 real-world, PII-scrubbed query-code pairs, which are used to evaluate the performance of the proposed pipeline.

The reproducible evaluation harness ensures that the evaluation process is consistent and can be reproduced by others.

The specialized is designed to efficiently retrieve code snippets from the 1C:Enterprise ecosystem, which combines Russian syntax with highly domain-specific terminology.

To overcome scarce labeled data, the authors 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.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The use of the open and reproducible evaluation harness ensures that the evaluation process is consistent and can be reproduced by others.

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

የምንጭ ዝርዝሮች: arxiv.org

ለምን አስፈላጊ ነው።

The proposed pipeline addresses the lack of open datasets and specialized models for 1C code , enabling the development of more accurate and efficient retrieval systems. The use of MRL and a privacy-aware tokenizer also ensures the protection of sensitive information.

The proposed pipeline addresses the lack of open datasets and specialized models for 1C code , enabling the development of more accurate and efficient retrieval systems.

The use of MRL and a privacy-aware tokenizer ensures the protection of sensitive information, making the proposed pipeline more reliable and trustworthy.

The proposed pipeline has the potential to improve the development of 1C code systems, which is essential for various applications, such as software development and maintenance.

The use of synthetic triplets generated from public code repositories reduces the need for labeled data, making the proposed pipeline more efficient and cost-effective.

The proposed pipeline can be used as a starting point for further research and development of more advanced 1C code systems.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The use of the open and reproducible evaluation harness ensures that the evaluation process is consistent and can be reproduced by others.

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

Interactive Mechanism

በይነተገናኝ ሜካኒዝም፡ በትክክል እንዴት እንደሚሰራ

ከዚህ ልማት በስተጀርባ ያለውን ቴክኖሎጂ በይነተገናኝ ያስሱ።

System Requirements:
Best ArchitecturePure RAGRecommended pattern
Hallucination RiskVery LowGrounding efficacy
Update Cost$0 (Vector sync)Ongoing maintenance
Core takeaway: Fine-tuning teaches models how to speak (form, style, syntax); RAG teaches models what to say (verifiable facts). Never use fine-tuning alone for factual memory.
በይነተገናኝ ጽንሰ-ሐሳብ ቼክ+10 Points
What is AI? Quiz

As use of AI scales up across an organization, what tends to matter most?

ቀጥሎ ምን እንደሚታይ

The authors' approach to fine-tuning on synthetic triplets generated from public code repositories, and the use of MRL and a privacy-aware tokenizer, are key aspects to watch in this research.

The authors' approach to fine-tuning on synthetic triplets generated from public code repositories is a key aspect to watch in this research.

The use of MRL and a privacy-aware tokenizer is another important aspect to watch, as it ensures the protection of sensitive information.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is also worth watching, as it has the potential to improve the development of 1C code systems.

The use of the open and reproducible evaluation harness ensures that the evaluation process is consistent and can be reproduced by others.

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The authors' use of synthetic triplets generated from public code repositories reduces the need for labeled data, making the proposed pipeline more efficient and cost-effective.

The proposed pipeline can be used as a starting point for further research and development of more advanced 1C code systems.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

The proposed pipeline's ability to efficiently retrieve code snippets from the 1C:Enterprise ecosystem is a significant advancement in the field of natural language code .

The proposed pipeline's potential to improve the development of 1C code systems makes it an important area of research and development.

ተዛማጅ መመሪያዎች እና ጥያቄዎች

AI ምንድን ነው?ChatGPT እና LLMsየAI ሥነ ምግባርAI ወኪሎችAI ሞዴሎች ተብራርተዋልትራንስፎርመሮችየAI መጪው ጊዜAI ስልጠናPrompt Engineeringየሚያውቁትን ይሞክሩ - ነፃ የ AI ጥያቄዎችን ይሞክሩበእኛ የቃላት መፍቻ ውስጥ የ AI ቃልን ይፈልጉ
ይህ ጠቃሚ ሆኖ ተገኝቷል?