{"id":24818869,"url":"https://github.com/gmongaras/wizard_qlora_finetuning","last_synced_at":"2025-04-12T00:53:02.172Z","repository":{"id":194889728,"uuid":"691745907","full_name":"gmongaras/Wizard_QLoRA_Finetuning","owner":"gmongaras","description":"Finetuning Some Wizard Models With QLoRA","archived":false,"fork":false,"pushed_at":"2023-09-17T13:18:53.000Z","size":78,"stargazers_count":7,"open_issues_count":0,"forks_count":9,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-04-12T00:52:58.157Z","etag":null,"topics":["finetuning","llama","llm","llm-finetuning","lora","qlora","wizard"],"latest_commit_sha":null,"homepage":"","language":"Python","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/gmongaras.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":"2023-09-14T19:55:50.000Z","updated_at":"2024-04-19T20:27:13.000Z","dependencies_parsed_at":"2025-03-25T20:37:53.178Z","dependency_job_id":null,"html_url":"https://github.com/gmongaras/Wizard_QLoRA_Finetuning","commit_stats":null,"previous_names":["gmongaras/wizard_qlora_finetuning"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/gmongaras%2FWizard_QLoRA_Finetuning","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/gmongaras%2FWizard_QLoRA_Finetuning/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/gmongaras%2FWizard_QLoRA_Finetuning/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/gmongaras%2FWizard_QLoRA_Finetuning/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/gmongaras","download_url":"https://codeload.github.com/gmongaras/Wizard_QLoRA_Finetuning/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248501881,"owners_count":21114682,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","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":["finetuning","llama","llm","llm-finetuning","lora","qlora","wizard"],"created_at":"2025-01-30T17:55:05.874Z","updated_at":"2025-04-12T00:53:02.166Z","avatar_url":"https://github.com/gmongaras.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Wizard_QLoRA_Finetuning\nFinetuning Some Wizard Models With QLoRA\n\n# Demo\n[https://youtu.be/hkt5Nz0buso?si=HNmYLp_z5SGZlMbM](https://youtu.be/hkt5Nz0buso?si=HNmYLp_z5SGZlMbM)\n\n# Pipeline\n## Model fine-tuning\nFinetuning can be done with the `finetune.py` script. In this script, a model will be downloaded and finetuned on one of the datasets in 4-bit precision. \nAs finetuning progress is being made, checkpoints are saved to the specified output directory.\n\n## Merging\nAfter the model is trained, one of the checkpoint files should be merged so that the LoRA weights and old weights are combined into a single weight matrix, \nmaking inference more efficient than if you had them split. `merge.py` does the merge given a specified checkpoint file and the specified model type.\n\n## Inference\nInference has a few scripts. `infer.py` and `infer.ipynb` are similar and just run straight inference on a given model. \n`infer_interface.ipynb` has an additional interface using Gradio.\n\n## Uploading/Saving Models\n`upload.py` can be used to upload huggingface models to the hub easily given a repo name to upload. Make sure to get a `write` token from huggingface to upload properly.\n\n## Data Creation\n`data_creation.ipynb` is a simple example of data creation.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgmongaras%2Fwizard_qlora_finetuning","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fgmongaras%2Fwizard_qlora_finetuning","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgmongaras%2Fwizard_qlora_finetuning/lists"}