-
Use builder.pynb to create the benchmark. The script relies on randomness, so commit the generated benchmark to the repository.
-
Use completions.py to geenerate completions.
python3 completions.py --input benchmark.jsonl --output completions.jsonl --model-name /home/arjun/models/starcoderbase --batch-size 50 --num-completions 1 --max-tokens 8192
-
Use executions.py to execution completions. (Needs an update.)
python3 executions.py --input completions.jsonl --output executions.jsonl
-
Use pass1.ipynb to look at the results. (Needs an update.)
arjunguha / longbench Goto Github PK
View Code? Open in Web Editor NEWLicense: Other