Morph MORPH
SIGN IN SIGN UP
dhruvhead / open-instruct UNCLAIMED
0 0 0 Python

COMMITS

/ eval/bbh/run_eval.py
docs
June 24, 2024
H
add to other scripts
Hamish Ivison committed
3c335be
February 25, 2024
H
Merge pull request #122 from liujch1998/eval_bbh_tydiqa
Hamish Ivison committed
59245eb
February 24, 2024
J
Eval: BBH always stops at \n\n; TydiQA reports recall
Jiacheng Liu committed
827204c
H
clean up, unneeded checks
Hamish Ivison committed
7cdf729
H
typo
Hamish Ivison committed
e4e6983
February 23, 2024
H
refactor: load model sep. from tokenizer
Hamish Ivison committed
adae32c
February 21, 2024
H
tokenizer util
Hamish Ivison committed
7f7acd4
February 13, 2024
H
Beaker improvements & HF formatting (#110)
Hamish Ivison committed
8dadfdf
October 21, 2023
Y
Fix the max_num_batched_tokens error in vllm.
yizhongw committed
c93a41e
October 1, 2023
Y
Refine the prompting and parsing for BBH.
yizhongw committed
720159a
September 24, 2023
Y
Support using vllm for running BBH
yizhongw committed
2a13f1b
September 22, 2023
Y
Use fast tokenizer by default.
yizhongw committed
28d5366
September 4, 2023
Y
Use import_module to dynamically import the chat formatting function, allowing more flexibility for user-defined formatting.
yizhongw committed
1a35973
September 3, 2023
Y
Use `chat_formatting_function` to specify which prompt encoding function the evals should use.
yizhongw committed
5d6264a
September 2, 2023
Y
Support LLaMa2 chat format.
yizhongw committed
ed2cded
August 14, 2023
Y
Fix a bug: when loading models, the code now will use `auto` torch_dtype, and only convert to half if the user explicitly specify that.
yizhongw committed
9cad3d2
June 9, 2023
H
initial commit
Hamish Ivison committed
e44fcf5