{"id":18885542,"url":"https://github.com/togethercomputer/opendatahub","last_synced_at":"2026-01-30T17:31:24.835Z","repository":{"id":142620906,"uuid":"601276481","full_name":"togethercomputer/OpenDataHub","owner":"togethercomputer","description":null,"archived":false,"fork":false,"pushed_at":"2023-04-26T23:49:43.000Z","size":41,"stargazers_count":128,"open_issues_count":14,"forks_count":29,"subscribers_count":9,"default_branch":"main","last_synced_at":"2025-06-05T03:26:30.847Z","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/togethercomputer.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,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2023-02-13T18:22:17.000Z","updated_at":"2025-05-14T23:15:54.000Z","dependencies_parsed_at":null,"dependency_job_id":"fa4646e5-ef59-4be6-81e9-80ba564008d2","html_url":"https://github.com/togethercomputer/OpenDataHub","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/togethercomputer/OpenDataHub","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/togethercomputer%2FOpenDataHub","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/togethercomputer%2FOpenDataHub/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/togethercomputer%2FOpenDataHub/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/togethercomputer%2FOpenDataHub/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/togethercomputer","download_url":"https://codeload.github.com/togethercomputer/OpenDataHub/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/togethercomputer%2FOpenDataHub/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":28916075,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-01-30T16:37:38.804Z","status":"ssl_error","status_checked_at":"2026-01-30T16:37:37.878Z","response_time":66,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":"2024-11-08T07:19:43.591Z","updated_at":"2026-01-30T17:31:24.815Z","avatar_url":"https://github.com/togethercomputer.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# OpenDataHub\n\nThis repository contains the current snapshot of the OpenChatKit bot. You can find all training data in `data`,\nthe hyperparameters used for training in `training.yaml`, training log in `training_log`,\nand the pointer to the model at `model.yaml`.\n\nYou can find in different branches different specialized versions of this bot.\n\nYou can make it better by contributing data! \n\n## Data Model\n\nHow should we think about the training data for OpenChatKit bots? A _training set_ is a _set_ of _slices_, \nwhere each _slice_ contains a set of (input, output) pairs. Each slice corresponds to one file \nin the `data` folder.\n\nFor example, if the data folder contains\n```\ndata\n|- pile.yaml\n|- soda.yaml\n```\nduring training, the training set will contain the union of both `pile` and `soda`.\nNote that different slices can be weighted differently, which will be specified in \nthe file `training.yaml` (see \"Model Training\" for details)\n\n### Data Format\n\nYou can provide data in various formats. \n\n1. You can provide a collection of input/output pairs\n```\nIOPairs:\n- input: INPUT TEXT STRING\n  output: OUTPUT TEXT STRING\n- input: INPUT TEXT STRING\n  output: OUTPUT TEXT STRING\n...\n```\nor pure text\n```\nText:\n- text: TEXT STRING\n- text: TEXT STRING\n...\n```\n\n2. You can provide us the link to your dataset on HuggingFace\n```\nHuggingFace:\n- link: LINK TO YOUR DATASET\n```\n\n3. You can prepare your dataset as in OpenAI jsonl format (https://platform.openai.com/docs/guides/fine-tuning) \nand put it in a link that we can `wget` or `curl`\n```\nOpenAIJsonl:\n- link: LINK TO YOUR DATASET\n```\n\n## Model Training\n\nEach merged pull request will trigger (currently manually) to the training of a model. \nHyper-parameters, including the specific mixture of data, will be specified in `training.yaml`:\n```\nTraining:\n  - lr: 0.0001\n  - momentum: 0.99\nMixture:\n  - pile: 0.5\n  - soda: 0.5\n```\nAfter training, a file `training_log` will be committed to the repository. And a file \n`model.yaml` will be made available in the repository specifying where to find this model\nand (optionally) Together API end-point to query such a model.\n\n## How to Contribute?\n\nYou can help us to make OpenChatKit better in three ways.\n\n### Finding \"Bugs\"\n\nIf you realize that the bug is not performing well, please open an issue, specifying \nyour input, the bot's output, and a description of what is wrong with it (potentially with the right answer).\n\n### Fixing \"Bugs\"\n\nIf you have data that you believe could be useful to fix some of the issues, please\nadd your data into the `data` folder and make a pull request associated with the issue\nthat you think this will fix.\n\nWe will review these pull requests, train a model, and merge them.\n\n### Specialization\n\nYou don't have to always merge into the main branch. If you have specific things to \ntry out (e.g., a `text2sql` bot), feel free to open a new branch work there!\n\nLet's work together to make the best open-source bot!\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftogethercomputer%2Fopendatahub","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ftogethercomputer%2Fopendatahub","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftogethercomputer%2Fopendatahub/lists"}