Fix

Fix numpy

Fix altair

Merge branch 'main' into hf-launch

Fix app

Add llama 2 model

Merge remote-tracking branch 'origin/main' into hf-launch

Add falcon 7b model

Add falcon 40b model

Add gpt4all-13b-snoozy model

Add text-davinci-003 results on HellaSwag and WinoGrande zero-shot

Add koala results on HellaSwag and WinoGrande zero-shot

Add stablelm results on HellaSwag and WinoGrande zero-shot

Add oasst/pythia-12b HellaSwag and WinoGrande zero-shot results

Add Pythia models WinoGrande (zero shot)

Add alpaca 7b model

Add dolly-v2-12b results

Option to not filter emptry entries in table

Fix hf spaces coloring

Test styling in HF

Move over to hf spaces

Remove action to check file sizes

Add github actions to upload to hf spaces

Fix streamlit theme

Add python version for HF spaces

Prepare HF spaces launch

Remove prompted StartCoder

Notes on definition of "open" model

Remove codeT results for code-davinci-002 as not comparable to other HumanEval results, due to additional explicit testing of outputs

Add link to hf space

Replace commercial column with open

Drop rows without any benchmarks

Add WinoGrande zero-shot and results

Add WinoGrande few shot results for gpt4 and 3.5

Drop columns after filtering for which there are no entries

Shown values in categorical filter now sorted

For now set PaLM2 commercial use to no until clear

Starting to add PaLM2 benchmark results

Add download button for table + update todos

Check for bool dtype at table filter

Add filtering by values

Add column for publisher

Add further results from HELM

Add / modify gpt models according to HELM benchmark

Clarifying gpt model names

Text work

Fix GPT -3 commercial use

Add special thanks

Add replit code

Add HellaSwag few shot