{
    "bomFormat": "CycloneDX",
    "specVersion": "1.6",
    "serialNumber": "urn:uuid:19c1f27e-d53a-45dc-94ba-a0f9fb30f79a",
    "version": 1,
    "metadata": {
        "timestamp": "2025-07-10T08:49:17.278628+00:00",
        "component": {
            "type": "machine-learning-model",
            "bom-ref": "bigcode/starcoder2-15b-instruct-v0.1-e7aea605-679e-5203-beb5-6bb2732ebe54",
            "name": "bigcode/starcoder2-15b-instruct-v0.1",
            "externalReferences": [
                {
                    "url": "https://huggingface.co/bigcode/starcoder2-15b-instruct-v0.1",
                    "type": "documentation"
                }
            ],
            "modelCard": {
                "modelParameters": {
                    "task": "text-generation",
                    "architectureFamily": "starcoder2",
                    "modelArchitecture": "Starcoder2ForCausalLM",
                    "datasets": [
                        {
                            "ref": "bigcode/self-oss-instruct-sc2-exec-filter-50k-87afcdf4-328e-5009-ba04-8150239e25f2"
                        }
                    ]
                },
                "properties": [
                    {
                        "name": "library_name",
                        "value": "transformers"
                    },
                    {
                        "name": "base_model",
                        "value": "bigcode/starcoder2-15b"
                    }
                ],
                "quantitativeAnalysis": {
                    "performanceMetrics": [
                        {
                            "slice": "dataset: livecodebench-codegeneration",
                            "type": "pass@1",
                            "value": 20.4
                        },
                        {
                            "slice": "dataset: livecodebench-selfrepair",
                            "type": "pass@1",
                            "value": 20.9
                        },
                        {
                            "slice": "dataset: livecodebench-testoutputprediction",
                            "type": "pass@1",
                            "value": 29.8
                        },
                        {
                            "slice": "dataset: livecodebench-codeexecution",
                            "type": "pass@1",
                            "value": 28.1
                        },
                        {
                            "slice": "dataset: humaneval",
                            "type": "pass@1",
                            "value": 72.6
                        },
                        {
                            "slice": "dataset: humanevalplus",
                            "type": "pass@1",
                            "value": 63.4
                        },
                        {
                            "slice": "dataset: mbpp",
                            "type": "pass@1",
                            "value": 75.2
                        },
                        {
                            "slice": "dataset: mbppplus",
                            "type": "pass@1",
                            "value": 61.2
                        },
                        {
                            "slice": "dataset: ds-1000",
                            "type": "pass@1",
                            "value": 40.6
                        }
                    ]
                }
            },
            "authors": [
                {
                    "name": "bigcode"
                }
            ],
            "licenses": [
                {
                    "license": {
                        "name": "bigcode-openrail-m"
                    }
                }
            ],
            "description": "We introduce StarCoder2-15B-Instruct-v0.1, the very first entirely self-aligned code Large Language Model (LLM) trained with a fully permissive and transparent pipeline. Our open-source pipeline uses StarCoder2-15B to generate thousands of instruction-response pairs, which are then used to fine-tune StarCoder-15B itself without any human annotations or distilled data from huge and proprietary LLMs.- **Model:** [bigcode/starcoder2-15b-instruct-v0.1](https://huggingface.co/bigcode/starcoder2-instruct-15b-v0.1)- **Code:** [bigcode-project/starcoder2-self-align](https://github.com/bigcode-project/starcoder2-self-align)- **Dataset:** [bigcode/self-oss-instruct-sc2-exec-filter-50k](https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k/)- **Authors:**[Yuxiang Wei](https://yuxiang.cs.illinois.edu),[Federico Cassano](https://federico.codes/),[Jiawei Liu](https://jw-liu.xyz),[Yifeng Ding](https://yifeng-ding.com),[Naman Jain](https://naman-ntc.github.io),[Harm de Vries](https://www.harmdevries.com),[Leandro von Werra](https://twitter.com/lvwerra),[Arjun Guha](https://www.khoury.northeastern.edu/home/arjunguha/main/home/),[Lingming Zhang](https://lingming.cs.illinois.edu).![self-alignment pipeline](https://huggingface.co/datasets/bigcode/starcoder2-instruct-assets/resolve/main/method.png)",
            "tags": [
                "transformers",
                "safetensors",
                "starcoder2",
                "text-generation",
                "code",
                "conversational",
                "dataset:bigcode/self-oss-instruct-sc2-exec-filter-50k",
                "arxiv:2410.24198",
                "base_model:bigcode/starcoder2-15b",
                "base_model:finetune:bigcode/starcoder2-15b",
                "license:bigcode-openrail-m",
                "model-index",
                "autotrain_compatible",
                "text-generation-inference",
                "endpoints_compatible",
                "region:us"
            ]
        }
    },
    "components": [
        {
            "type": "data",
            "bom-ref": "bigcode/self-oss-instruct-sc2-exec-filter-50k-87afcdf4-328e-5009-ba04-8150239e25f2",
            "name": "bigcode/self-oss-instruct-sc2-exec-filter-50k",
            "data": [
                {
                    "type": "dataset",
                    "bom-ref": "bigcode/self-oss-instruct-sc2-exec-filter-50k-87afcdf4-328e-5009-ba04-8150239e25f2",
                    "name": "bigcode/self-oss-instruct-sc2-exec-filter-50k",
                    "contents": {
                        "url": "https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k",
                        "properties": [
                            {
                                "name": "pretty_name",
                                "value": "StarCoder2-15b Self-Alignment Dataset (50K)"
                            },
                            {
                                "name": "configs",
                                "value": "Name of the dataset subset: default {\"split\": \"train\", \"path\": \"data/train-*\"}"
                            },
                            {
                                "name": "license",
                                "value": "odc-by"
                            }
                        ]
                    },
                    "governance": {
                        "owners": [
                            {
                                "organization": {
                                    "name": "bigcode",
                                    "url": "https://huggingface.co/bigcode"
                                }
                            }
                        ]
                    },
                    "description": "Final self-alignment training dataset for StarCoder2-Instruct. \n\nseed: Contains the seed Python function\nconcepts: Contains the concepts generated from the seed\ninstruction: Contains the instruction generated from the concepts\nresponse: Contains the execution-validated response to the instruction\n\nThis dataset utilizes seed Python functions derived from the MultiPL-T pipeline.\n"
                }
            ]
        }
    ]
}