clembench / clembench-runs

All outputs generated by running the benchmark on different versions
MIT License
0 stars 5 forks source link

Add fixed Mistral results for imagegame and referencegame #3

Closed Gnurro closed 6 months ago

Gnurro commented 6 months ago

Prior code failed to run Mistral models due to usage of system messages in imagegame and referencegame, this PR adds the results from re-running with fixed code.