Questions about ppl when using gpt2

Hi! I meet some problems when running ConvAI2 evaluation scripts:

I first trained a model from OpenAI GPT. I increased the number of cumulative gradients because I only have one card.
```
python train.py --model_checkpoint /path/to/pretrained/gpt \
--gradient_accumulation_steps=32 --lm_coef=2.0 --max_history=2 \
--n_epochs=1 --num_candidates=4 --personality_permutations=2 \
--train_batch_size=2 --valid_batch_size=2
```
this gives ConvAI2 evalution results:
```
Final Hits@1: 0.761
FINAL F1: 0.1659
FINAL PPL: 20.7
```
Then I tried to train from GPT2-small with the same config:
```
python train.py --model_checkpoint /path/to/pretrained/gpt2 \
--gradient_accumulation_steps=32 --lm_coef=2.0 --max_history=2 \
--n_epochs=1 --num_candidates=4 --personality_permutations=2 \
--train_batch_size=2 --valid_batch_size=2
```
and the evaluation results are:
```
Final Hits@1: 0.737
FINAL F1: 0.1643
FINAL PPL: 178.9
```
The command I used to run convai_evalution.py is:
```
python convai_evaluation.py --eval_type ppl --model_checkpoint /path/to/finetuned/model
```

The ppl of GPT2 is strangely high.

Is there anything that needs to be modified when testing finetuned-gpt2 with convai_evalution.py?

I'm also curious about the best test results and hyperparameters when you finetuned from GPT2. Thank you!

Start with train.py and convai_evaluation.py, focusing on how GPT and GPT-2 checkpoints are loaded and how the ppl evaluation is calculated. Reproduce both commands with the reported settings, then determine whether GPT-2 requires different evaluation handling; done means explaining the PPL discrepancy and documenting any required command or configuration changes.

Written by the indexing model from the issue text.