{"id":16999568,"url":"https://github.com/nirantk/codellms","last_synced_at":"2026-06-15T21:31:27.810Z","repository":{"id":188070376,"uuid":"648533193","full_name":"NirantK/CodeLLMs","owner":"NirantK","description":"Everything we know about LLMs for CodeGen","archived":false,"fork":false,"pushed_at":"2023-06-05T07:25:43.000Z","size":2,"stargazers_count":1,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-11-09T12:08:29.548Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/NirantK.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2023-06-02T07:35:41.000Z","updated_at":"2025-01-05T07:39:13.000Z","dependencies_parsed_at":"2023-08-13T16:36:49.493Z","dependency_job_id":null,"html_url":"https://github.com/NirantK/CodeLLMs","commit_stats":null,"previous_names":["nirantk/codellms"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/NirantK/CodeLLMs","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NirantK%2FCodeLLMs","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NirantK%2FCodeLLMs/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NirantK%2FCodeLLMs/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NirantK%2FCodeLLMs/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/NirantK","download_url":"https://codeload.github.com/NirantK/CodeLLMs/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NirantK%2FCodeLLMs/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34381758,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-15T02:00:07.085Z","response_time":63,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-10-14T04:09:14.793Z","updated_at":"2026-06-15T21:31:27.789Z","avatar_url":"https://github.com/NirantK.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# CodeLLMs\nEverything I know about LLMs for CodeGen\n\n\nhttps://arxiv.org/pdf/2305.12050.pdf CodeCompose: A Large-Scale Industrial Deployment of AI-assisted Code Authoring\n\n- CodeOps has 3 recurring patterns\n    - Retrieval\n        - Files, classes, lines? \n    - Edit\n        - See examples I had in mind\n    - Verify\n        - CI/CD\n        - Pass the stacktrace back and complete the loop\n- Examples Tasks I had in mind\n    - In a FastAPI app, replace a blocking call in a 2-3 file app with a request id response and a new endpoint to poll, use BackgroundTasks or something else as a queue\n    - Change a Python function from sync to async and await on it in a different file\n    - Rename variable\n    - Refactor a script into functions and classes\n- Naming the building blocks\n    - In-filling\n        - Supported by Incoder, StarCoder, can hack GPT4 too\n    - Fresh Generate\n        - Supported by Incoder, StarCoder, GPT4 \n    - Retrieval\n        - Code Embedding\n            - Maybe something like Replit Code's 1.3b model, finetuned for a specific code? \n        - LLM\n            - Pass the codebase structure e.g. files, classes in prefix, ask the LLM  —  what will change instead of asking it to change? \n                - https://chat.openai.com/share/b21ac3d9-cff1-41b1-b040-a0b7e2d65ee3\n    - Code Markers\n        - Lines and Function prototype\n            - https://chat.openai.com/share/b21ac3d9-cff1-41b1-b040-a0b7e2d65ee3\n        - Syntax Tree\n            - BloopAI uses this with Code Embedding\n            - Works well with small classes, and functions in Python, but gives up with object calls\n        - Generated diff???\n- Ideas that I think we should dive on\n    - Using Developer Edits/git commit diffs\n        - No code, but DIDACT claims to do this\n            - https://ai.googleblog.com/2023/05/large-sequence-models-for-software.html \n            - GPT4 can definitely take Git diff commit and return Git commit messages — indicating some level of code understanding\n        - Works with StarCoder with prompting tricks\n    - AutoPR\n    - BloopAI\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnirantk%2Fcodellms","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fnirantk%2Fcodellms","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnirantk%2Fcodellms/lists"}