SantaCoder: don't reach for the stars!
read the original abstract
The BigCode project is an open-scientific collaboration working on the responsible development of large language models for code. This tech report describes the progress of the collaboration until December 2022, outlining the current state of the Personally Identifiable Information (PII) redaction pipeline, the experiments conducted to de-risk the model architecture, and the experiments investigating better preprocessing methods for the training data. We train 1.1B parameter models on the Java, JavaScript, and Python subsets of The Stack and evaluate them on the MultiPL-E text-to-code benchmark. We find that more aggressive filtering of near-duplicates can further boost performance and, surprisingly, that selecting files from repositories with 5+ GitHub stars deteriorates performance significantly. Our best model outperforms previous open-source multilingual code generation models (InCoder-6.7B and CodeGen-Multi-2.7B) in both left-to-right generation and infilling on the Java, JavaScript, and Python portions of MultiPL-E, despite being a substantially smaller model. All models are released under an OpenRAIL license at https://hf.co/bigcode.
This paper has not been read by Pith yet.
Forward citations
Cited by 24 Pith papers
-
PrivCode++: Latent-Conditioned Differentially Private Code Generation for Comprehensive Guarantees
PrivCode++ introduces the first DP code generation method protecting both prompts and code via latent-conditioned two-stage training, claiming higher utility and stronger privacy than prior baselines.
-
Beyond Pass Rate: A Multilingual, Execution-Grounded Evaluation of Open Code LLMs
Multilingual execution-grounded benchmark finds top open code LLM at 23.64% correctness versus 57.2% human baseline, with compile errors dominating 63% of failures.
-
C-Pack: Packed Resources For General Chinese Embeddings
C-Pack releases a new Chinese embedding benchmark, large training dataset, and optimized models that outperform priors by up to 10% on C-MTEB while also delivering English SOTA results.
-
RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems
RepoBench is a new benchmark with retrieval, completion, and pipeline tasks to evaluate code auto-completion systems on entire repositories instead of single files.
-
Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generation
EvalPlus augments HumanEval with 80x more tests via LLM and mutation strategies, exposing up to 28.9% more incorrect LLM-generated code and reversing some model performance rankings.
-
When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs
Experiments across code LLMs show no-review collapses fastest, human-gated filters slow collapse, and AI self-gates lose effect over time, degenerating to ungated self-training under self-confirming acceptance as prov...
-
CodegenBench: Can LLMs Write Efficient Code Across Architectures?
CodegenBench shows LLMs generate optimized code well for x86_64 but exhibit significant performance degradation on Sunway and Kunpeng due to limited documentation and training data.
-
No Test Cases, No Problem: Distillation-Driven Code Generation for Scientific Workflows
MOSAIC generates executable scientific code without I/O test cases by combining student-teacher distillation with a consolidated context window to reduce hallucinations across subproblems.
-
DuCodeMark: Dual-Purpose Code Dataset Watermarking via Style-Aware Watermark-Poison Design
DuCodeMark watermarks code datasets using AST style transformations and repressible poisons for both source-code and decompilation tasks, verified by t-test, with high stealth and a 28.6% performance drop if removed.
-
MOSAIC: Multi-agent Orchestration for Task-Intelligent Scientific Coding
MOSAIC is a training-free multi-agent LLM framework with rationale, coding, reflection, and debugging agents plus a consolidated context window that outperforms prior methods on scientific coding benchmarks.
-
DataComp-LM: In search of the next generation of training sets for language models
DCLM-Baseline dataset lets a 7B model reach 64% 5-shot MMLU accuracy after 2.6T tokens, beating prior open-data models by 6.6 points on MMLU with 40% less compute.
-
DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence
An open-source MoE code model matches GPT-4 Turbo on coding and math benchmarks while expanding to 338 languages and 128K context length.
-
LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code
LiveCodeBench collects 400 recent contest problems to create a contamination-free benchmark evaluating LLMs on code generation and related capabilities like self-repair and execution.
-
StarCoder 2 and The Stack v2: The Next Generation
StarCoder2-15B matches or beats CodeLlama-34B on code tasks despite being smaller, and StarCoder2-3B outperforms prior 15B models, with open weights and exact training data identifiers released.
-
Textbooks Are All You Need
A 1.3B-parameter code model trained on 7B tokens of curated textbook and synthetic data achieves 50.6% on HumanEval, indicating data quality can enable strong performance at small scale.
-
Scaling Data-Constrained Language Models
Repeating training data up to 4 epochs yields negligible loss increase versus unique data for fixed compute, and a new scaling law accounts for the decaying value of repeated tokens and excess parameters.
-
Characterizing initial human-AI proof formalization workflows
A controlled user study and qualitative survey find that AI assistance raises formalization accuracy for math proofs, with users flexibly combining multiple tools while retaining oversight.
-
Hide to Guide: Learning via Semantic Masking
SMEPO applies fine-grained semantic masking to expert guidance in RLVR, turning hard problems into fill-in-the-blank tasks while preserving structure, yielding up to 3.2 point accuracy gains and 4.2x faster training.
-
DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence
DeepSeek-Coder open-source models trained on 2T code tokens with fill-in-the-blank pretraining achieve SOTA results among open models and surpass closed-source Codex and GPT-3.5 on code benchmarks.
-
An Empirical Study on Influence-Based Pretraining Data Selection for Code Large Language Models
Data-influence-score filtering using validation-set loss on downstream coding tasks improves Code-LLM performance, with the most beneficial training data varying significantly across different programming tasks.
-
IssueGuard: Real-Time Secret Leak Prevention Tool for GitHub Issue Reports
IssueGuard delivers real-time secret detection in GitHub issues via regex and CodeBERT, reaching 92.7% F1-score and outperforming pure regex scanners.
-
Retrieval-Augmented Code Generation: A Survey with Focus on Repository-Level Approaches
The paper organizes repository-level retrieval-augmented code generation into a unified framework covering retrieval substrate, control regime, and evaluation setting while summarizing strategies, datasets, and challenges.
-
Qwen2.5-Coder Technical Report
Qwen2.5-Coder models claim state-of-the-art results on over 10 code benchmarks, outperforming larger models of similar size.
-
A Survey on Large Language Models for Code Generation
A systematic literature review that organizes recent work on LLMs for code generation into a taxonomy covering data curation, model advances, evaluations, ethics, environmental impact, and applications, with benchmark...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.