Atla Selene Mini

[NinaCalvi]

Atla Selene Mini

to run it:

python scripts/run_generative.py --model="AtlaAI/Selene-1-Mini-Llama-3.1-8B"

The json results are:

{
  "alpacaeval-easy": 0.93,
  "alpacaeval-hard": 0.9263157894736842,
  "alpacaeval-length": 0.9368421052631579,
  "donotanswer": 0.8235294117647058,
  "hep-cpp": 0.9207317073170732,
  "hep-go": 0.9085365853658537,
  "hep-java": 0.8902439024390244,
  "hep-js": 0.9085365853658537,
  "hep-python": 0.926829268292683,
  "hep-rust": 0.8475609756097561,
  "llmbar-adver-GPTInst": 0.9021739130434783,
  "llmbar-adver-GPTOut": 0.5957446808510638,
  "llmbar-adver-manual": 0.6521739130434783,
  "llmbar-adver-neighbor": 0.7611940298507462,
  "llmbar-natural": 0.92,
  "math-prm": 0.9854586129753915,
  "mt-bench-easy": 1.0,
  "mt-bench-hard": 0.7297297297297297,
  "mt-bench-med": 0.925,
  "refusals-dangerous": 0.95,
  "refusals-offensive": 0.985,
  "xstest-should-refuse": 1.0,
  "xstest-should-respond": 0.804
  "model": "AtlaAI/Selene-1-Mini-Llama-3.1-8B ",
  "model_type": "generative"
}

[natolambert]

@NinaCalvi -- please run

make style
make quality

[NinaCalvi]

@NinaCalvi -- please run

make style
make quality

addressed!