diff --git a/README.md b/README.md index b0486d7..da88f2e 100644 --- a/README.md +++ b/README.md @@ -2,20 +2,18 @@ Lightweight prompt evaluation package. -Use online. Can also be used locally through streamlit. Can use [ollama](https://github.com/ollama/ollama-python) to run LLMs locally if necessary. +Use online [here](https://crucible.streamlit.app/). Can also be used locally through streamlit. Can use [ollama](https://github.com/ollama/ollama-python) to run LLMs locally if necessary. Cost estimation is very rough (input \* 2). -## How to use +## Instructions -1. Set the models in `models.py` -1. Set prompts in `prompts.py` -1. Set variables in `variables.py` -1. Set grading style in `main.py`. +1. Set the models, prompts and variables +1. Set grading style and temperature - `"EXACT"`: is either right or wrong. ignores line breaks and spaces in answer - - `"QUALITATIVE"`: ask gpt4o for feedback -1. Run `python src/crucible/main.py`. -1. Logs from the run will be in `outputs/.yaml`. + - `"QUALITATIVE"`: ask gpt4o for feedback. be mindful of this token usage +1. Click `compile`. Check the price estimation. Click `run`. +1. Results are shown segmented by category. ## Parameters @@ -59,8 +57,8 @@ Cost estimation is very rough (input \* 2). ## TODO -- add tables - add tests +- add instructions ## Resources diff --git a/simulacro_1.jpeg b/simulacro_1.jpeg deleted file mode 100644 index 2dfe1f2..0000000 Binary files a/simulacro_1.jpeg and /dev/null differ diff --git a/src/crucible/streamlit_app.py b/src/crucible/streamlit_app.py index d76463c..ce89db8 100644 --- a/src/crucible/streamlit_app.py +++ b/src/crucible/streamlit_app.py @@ -23,6 +23,7 @@ st.title("Crucible") st.caption("Lightweight prompt evaluation") st.caption("An AUTOMATON tool") +st.caption("[Source](https://github.com/noah-art3mis/crucible)") st.header("Configuration")