h2o-llmstudio
h2o-llmstudio copied to clipboard
h2oai
H2O LLM Studio - a framework and no-code GUI for fine-tuning LLMs. Documentation: https://docs.h2o.ai/h2o-llmstudio/
common error is to set learning rate with e notation. For example 3e-4. Yaml parses it as a string
### 🚀 Feature Allow epoch to be optionally used as the x-axis of training/eval charts for easier comparison between runs with different amounts of training pairs. ### Motivation I'm often...
### 🚀 Feature Additional to the hardcoded models, add support to use local models as judges for evaluation. Can be simplified to require the OpenAI API. Should be basically an...
### 🚀 Feature In the Validation Prediction Insights, include the Zero-Epoch Response (i.e. how would the original foundation model response) as well as the BLEU (or other validation score) on...
- Move the local_rank == 0 check into the logger initialization to simplify code - Adds an option to log all ranks for an experiment. closes #584
In this PR we add multi-target support for both classification and regression. The following code-adaptation have been made: - For classification and regression the answer column can now be a...
### 🐛 Bug S3 import from public bucket fails on first try. It works after clicking continue and ignoring error message. This is confirmed in version 1.8.0. The functionality worked...
### 🐛 Bug  ### To Reproduce In some cases, the training progress is not rounded, and shows as a very long float value. ### LLM Studio version v1.8.0+ ?
### 🚀 Feature When starting from docker, the demo datasets could be prepackaged and then copied to the workdir to save the download and preparation time at the initial startup....
### 🚀 Feature Support multilabel causal classification