[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:h2o-llmstudio":3},"\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fh2oai\u002Fh2o-llmstudio\u002FHEAD\u002Fllm_studio\u002Fapp_utils\u002Fstatic\u002Fllm-studio-logo-light.png#gh-dark-mode-only\" \u002F>\u003C\u002Fp>\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fh2oai\u002Fh2o-llmstudio\u002FHEAD\u002Fllm_studio\u002Fapp_utils\u002Fstatic\u002Fllm-studio-logo.png#gh-light-mode-only\" \u002F>\u003C\u002Fp>\u003Ch3>\n    \u003Cp>Welcome to H2O LLM Studio, a framework and no-code GUI designed for\u003Cbr \u002F>\n    fine-tuning state-of-the-art large language models (LLMs).\n\u003C\u002Fp>\n\u003C\u002Fh3>\u003Cp>\u003Ca href=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F1069138\u002F233859311-32aa1f8c-4d68-47ac-8cd9-9313171ff9f9.png\" rel=\"nofollow ugc noopener\">\u003Cimg width=\"50%\" alt=\"home\" src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F1069138\u002F233859311-32aa1f8c-4d68-47ac-8cd9-9313171ff9f9.png\" \u002F>\u003C\u002Fa>\u003Ca href=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F1069138\u002F233859315-e6928aa7-28d2-420b-8366-bc7323c368ca.png\" rel=\"nofollow ugc noopener\">\u003Cimg width=\"50%\" alt=\"logs\" src=\"https:\u002F\u002Fuser-images.githubusercontent.com\u002F1069138\u002F233859315-e6928aa7-28d2-420b-8366-bc7323c368ca.png\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch2>Jump to\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ca href=\"#with-h2o-llm-studio-you-can\" rel=\"nofollow ugc noopener\">With H2O LLM Studio, you can\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#quickstart\" rel=\"nofollow ugc noopener\">Quickstart\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#whats-new\" rel=\"nofollow ugc noopener\">What's New\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#setup\" rel=\"nofollow ugc noopener\">Setup\u003C\u002Fa>\u003Cul>\n\u003Cli>\u003Ca href=\"#recommended-install\" rel=\"nofollow ugc noopener\">Recommended Install\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#virtual-environments\" rel=\"nofollow ugc noopener\">Virtual Environments\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#run-h2o-llm-studio-gui\" rel=\"nofollow ugc noopener\">Run H2O LLM Studio GUI\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#run-h2o-llm-studio-gui-using-docker\" rel=\"nofollow ugc noopener\">Run H2O LLM Studio GUI using Docker\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#run-h2o-llm-studio-with-command-line-interface-cli\" rel=\"nofollow ugc noopener\">Run H2O LLM Studio with command line interface (CLI)\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#troubleshooting\" rel=\"nofollow ugc noopener\">Troubleshooting\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#data-format-and-example-data\" rel=\"nofollow ugc noopener\">Data format and example data\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#training-your-model\" rel=\"nofollow ugc noopener\">Training your model\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#example-run-on-oasst-data-via-cli\" rel=\"nofollow ugc noopener\">Example: Run on OASST data via CLI\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#model-checkpoints\" rel=\"nofollow ugc noopener\">Model checkpoints\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#documentation\" rel=\"nofollow ugc noopener\">Documentation\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#contributing\" rel=\"nofollow ugc noopener\">Contributing\u003C\u002Fa>\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"#license\" rel=\"nofollow ugc noopener\">License\u003C\u002Fa>\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>With H2O LLM Studio, you can\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>easily and effectively fine-tune LLMs \u003Cstrong>without the need for any coding experience\u003C\u002Fstrong>.\u003C\u002Fli>\n\u003Cli>use a \u003Cstrong>graphical user interface (GUI)\u003C\u002Fstrong> specially designed for large language models.\u003C\u002Fli>\n\u003Cli>fine-tune any LLM using a large variety of hyperparameters.\u003C\u002Fli>\n\u003Cli>use recent fine-tuning techniques such as \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.09685\" rel=\"nofollow ugc noopener\">Low-Rank Adaptation (LoRA)\u003C\u002Fa> and 8-bit model training with a low memory footprint.\u003C\u002Fli>\n\u003Cli>use Reinforcement Learning (RL) to fine-tune your model (experimental).\u003C\u002Fli>\n\u003Cli>use advanced evaluation metrics to judge generated answers by the model.\u003C\u002Fli>\n\u003Cli>track and compare your model performance visually. In addition, \u003Ca href=\"https:\u002F\u002Fwandb.ai\u002F\" rel=\"nofollow ugc noopener\">W&B\u003C\u002Fa> integration can be used.\u003C\u002Fli>\n\u003Cli>chat with your model and get instant feedback on your model performance.\u003C\u002Fli>\n\u003Cli>easily export your model to the \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002F\" rel=\"nofollow ugc noopener\">Hugging Face Hub\u003C\u002Fa> and share it with the community.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Quickstart\u003C\u002Fh2>\n\u003Cp>For questions, discussing, or just hanging out, come and join our \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FWKhYMWcVbq\" rel=\"nofollow ugc noopener\">Discord\u003C\u002Fa>!\u003C\u002Fp>\n\u003Cp>Use cloud-based runpod.io instance to run the latest version of H2O LLM Studio with GUI.\u003C\u002Fp>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fwww.runpod.io\u002Fconsole\u002Fdeploy?template=vf9ppiy56z\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F0dffd945-0be0-4ef0-85cd-4e6f260d4e6c\" alt=\"open_in_runpod\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Cp>Using CLI for fine-tuning LLMs:\u003C\u002Fp>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fwww.kaggle.com\u002Fcode\u002Filu000\u002Fh2o-llm-studio-cli\u002F\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fkaggle.com\u002Fstatic\u002Fimages\u002Fopen-in-kaggle.svg\" alt=\"Kaggle\" \u002F>\u003C\u002Fa> \u003Ca href=\"https:\u002F\u002Fcolab.research.google.com\u002Fdrive\u002F1soqfJjwDJwjjH-VzZYO_pUeLx5xY4N1K?usp=sharing\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg\" alt=\"Open in Colab\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch2>What's New\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fh2oai\u002Fh2o-llmstudio\u002Fpull\u002F788\" rel=\"nofollow ugc noopener\">PR 788\u003C\u002Fa> New problem type for Causal Regression Modeling allows to train single target regression data using LLMs.\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fh2oai\u002Fh2o-llmstudio\u002Fpull\u002F747\" rel=\"nofollow ugc noopener\">PR 747\u003C\u002Fa> Fully removed RLHF in favor of DPO\u002FIPO\u002FKTO optimization.\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fh2oai\u002Fh2o-llmstudio\u002Fpull\u002F741\" rel=\"nofollow ugc noopener\">PR 741\u003C\u002Fa> Removing separate max length settings for prompt and answer in favor of a single \u003Ccode>max_length\u003C\u002Fcode> settings better resembling \u003Ccode>chat_template\u003C\u002Fcode> functionality from \u003Ccode>transformers\u003C\u002Fcode>.\u003C\u002Fli>\n\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fh2oai\u002Fh2o-llmstudio\u002Fpull\u002F599\" rel=\"nofollow ugc noopener\">PR 592\u003C\u002Fa> Added \u003Ccode>KTOPairLoss\u003C\u002Fcode> for DPO modeling allowing\u003C\u002Fli>\n\u003C\u002Ful>\n",1784240407315]