Datasets:

Modalities:
Tabular
Text
Formats:
parquet
Libraries:
Datasets
pandas
zetavg's picture
Upload README.md with huggingface_hub
c6bebca
|
raw
history blame
1.02 kB
---
dataset_info:
dataset_size: 122922724.1543383
download_size: 35786922
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
- dtype: string
name: preview
- dtype: int64
name: length
- dtype: int64
name: messages_count
splits:
- name: train
num_bytes: 121570482.22576687
num_examples: 8060
- name: test
num_bytes: 1352241.9285714286
num_examples: 83
---
# zh-tw-pythia-ta8000-v1-it1-sg-002
This dataset is a part of the `zh-tw-llm` project.
* Tokenizer: `zh-tw-pythia-tokenizer-a8000-v1`
* Built with: `sharegpt`
* Rows: `train` `8060`, `test` `83`
* Max length: `2048`
* Full config:
```json
{"build_with": ["sharegpt"], "preview_length": 512, "sharegpt_settings": {"source_dataset": "zetavg/ShareGPT-Processed", "train_on_inputs": false, "languages": [{"en": 0.3}, {"zh": 0.2}, "zh_Hant"], "rows_limit": 10000, "test_size": 0.01, "test_split_seed": 42, "test_rows_limit": 100}}
```