Hugging Face's logo

RaiBP
/
gpt2-openwebtext2-first-30-chunks-ablation-non-english

Text Generation
Transformers
TensorBoard
Safetensors
gpt2
Generated from Trainer
text-generation-inference
Inference Endpoints
Model card Files Metrics Community
RaiBP commited on
Commit
a7f2f3d
·
1 Parent(s): 14bd45a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -10
README.md CHANGED
@@ -49,16 +49,17 @@ The following hyperparameters were used during training:
49
  ### Training results
50
 
51
  ### Evaluation results
52
- Perplexity on random 2000 examples of the target language's [Wikipedia dataset](https://huggingface.co/datasets/wikimedia/wikipedia), using the code provided in the [perplexity docs](https://huggingface.co/docs/transformers/perplexity), with 512 tokes of stride:
53
- | Target language | PPL |
54
- |-----------------|-------------------|
55
- | en |42.175106048583984 |
56
- | de |225.5620574951172 |
57
- | es |184.9262237548828 |
58
- | fr |170.0771026611328 |
59
- |it |238.36192321777344 |
60
- |pt |203.595947265625 |
61
- |nl |225.9720001220703 |
 
62
 
63
  The following script was used for evaluation
64
 
 
49
  ### Training results
50
 
51
  ### Evaluation results
52
+ Perplexity on random 2000 examples of the target language's [Wikipedia dataset](https://huggingface.co/datasets/wikimedia/wikipedia), using the code provided in the [perplexity docs](https://huggingface.co/docs/transformers/perplexity), with 512 tokes of stride.
53
+ Baseline is the result from [OpenAI's GPT-2](https://huggingface.co/gpt2).
54
+ | Target language | PPL | Baseline PPL |
55
+ |-----------------|-------------------|-------------------|
56
+ | en |42.175106048583984 |26.562532424926758 |
57
+ | de |225.5620574951172 ||
58
+ | es |184.9262237548828 ||
59
+ | fr |170.0771026611328 ||
60
+ |it |238.36192321777344 ||
61
+ |pt |203.595947265625 ||
62
+ |nl |225.9720001220703 ||
63
 
64
  The following script was used for evaluation
65