---
dataset_info:
  features:
  - name: id
    dtype: int64
  - name: seed
    dtype: int64
  - name: temp
    dtype: float64
  - name: top_p
    dtype: float64
  - name: text
    dtype: string
  - name: tokens
    dtype: int64
  splits:
  - name: train
    num_bytes: 490450904
    num_examples: 262668
  download_size: 305146673
  dataset_size: 490450904
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---
# Token Statistics

- Total sequences: 262668
- Total tokens: 110395870
- Unique tokens used in all positions: 122995
- Unique tokens in first position: 11749
- Mean token count: 420.28671174257215
- Token count std: 366.30608313535305
- Min sequence length: 1
- Max sequence length: 1096

# Distribution Analysis:
- Tokenizer vocabulary size: 128000
- Unique tokens in dataset first position: 11749
- KL(Model -> Dataset): 0.1909
- KL(Dataset -> Model): 0.0632
- Jensen-Shannon divergence: 0.1271
- Cross-entropy: 5.2445


![image/png](https://cdn-uploads.huggingface.co/production/uploads/64105805928400b416439f10/cxqwEtaxEawjAyRmbByUm.png)