Tristan commited on
Commit
e4f5e40
1 Parent(s): 5d5b4cd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -2
README.md CHANGED
@@ -94,5 +94,7 @@ The model achieves the following results without any fine-tuning (zero-shot):
94
  |arc_easy |acc/acc_norm|0.4381/0.3948 |**0.4651**/**0.4247** |**0.0082**/**0.0029** |
95
  |arc_challenge|acc/acc_norm|0.1903/0.2270 |0.1997/0.2329 |0.4132/0.6256 |
96
 
97
- To get these results, we used the Eleuther AI evaluation harness [here](https://github.com/EleutherAI/lm-evaluation-harness),
98
- which can produce results different than those reported in the GPT2 paper. The p-values come from the stderr from the evaluation harness, plus a normal distribution assumption.
 
 
 
94
  |arc_easy |acc/acc_norm|0.4381/0.3948 |**0.4651**/**0.4247** |**0.0082**/**0.0029** |
95
  |arc_challenge|acc/acc_norm|0.1903/0.2270 |0.1997/0.2329 |0.4132/0.6256 |
96
 
97
+ To get these results, we used commit `4f0410a4be0049729078376ce36a42dc308b6e38` of the Eleuther AI evaluation harness [here](https://github.com/EleutherAI/lm-evaluation-harness),
98
+ which can produce results different than those reported in the GPT2 paper.
99
+ We added a change [here](https://github.com/EleutherAI/lm-evaluation-harness/compare/master...mathemakitten:lm-evaluation-harness:master) to enable evaluation of the OLM GPT2, which has a very slightly different vocab size.
100
+ The p-values come from the stderr from the evaluation harness, plus a normal distribution assumption.