README.md

---
configs:
- config_name: recon_codeparrot_train
  data_files:
  - split: train
    path: recon_codeparrot_train/**/*.parquet
- config_name: recon_finewiki_en
  data_files:
  - split: train
    path: recon_finewiki_en/**/*.parquet
- config_name: recon_latex_formulas_en_10m
  data_files:
  - split: train
    path: recon_latex_formulas_en_10m/**/*.parquet
- config_name: recon_nemotron_cc_v2_hq
  data_files:
  - split: train
    path: recon_nemotron_cc_v2_hq/**/*.parquet
- config_name: recon_nemotron_cc_v2_hq_synth
  data_files:
  - split: train
    path: recon_nemotron_cc_v2_hq_synth/**/*.parquet
- config_name: recon_nemotron_math_4plus
  data_files:
  - split: train
    path: recon_nemotron_math_4plus/**/*.parquet
- config_name: recon_redpajama_github
  data_files:
  - split: train
    path: recon_redpajama_github/**/*.parquet
- config_name: sft_deepseek-r1-0528
  data_files:
  - split: train
    path: sft_DeepSeek-R1-0528/**/*.parquet
- config_name: sft_deepseek-r1
  data_files:
  - split: train
    path: sft_DeepSeek-R1/**/*.parquet
- config_name: sft_deepseek-v3_deepseek-v3-0324
  data_files:
  - split: train
    path: sft_DeepSeek-V3_DeepSeek-V3-0324/**/*.parquet
- config_name: sft_deepseek-v3_mixtral_8x22b_qwen_2-5_72b
  data_files:
  - split: train
    path: sft_DeepSeek-V3_Mixtral_8x22B_Qwen_2.5_72B/**/*.parquet
- config_name: sft_deepseek_r1_deepseek-v3_qwen_2-5_72b_deepseek-v3-0324
  data_files:
  - split: train
    path: sft_Deepseek_R1_DeepSeek-V3_Qwen_2.5_72B_DeepSeek-V3-0324/**/*.parquet
- config_name: sft_llama-nemotron-post-training-dataset-chat
  data_files:
  - split: train
    path: sft_Llama-Nemotron-Post-Training-Dataset-chat/**/*.parquet
- config_name: sft_llama-nemotron-post-training-dataset-code
  data_files:
  - split: train
    path: sft_Llama-Nemotron-Post-Training-Dataset-code/**/*.parquet
- config_name: sft_llama-nemotron-post-training-dataset-math
  data_files:
  - split: train
    path: sft_Llama-Nemotron-Post-Training-Dataset-math/**/*.parquet
- config_name: sft_llama-nemotron-post-training-dataset-science
  data_files:
  - split: train
    path: sft_Llama-Nemotron-Post-Training-Dataset-science/**/*.parquet
- config_name: sft_magpie-qwen2-5-pro-1m-v0-1
  data_files:
  - split: train
    path: sft_Magpie-Qwen2.5-Pro-1M-v0.1/**/*.parquet
- config_name: sft_mixtral-8x22b-v0-1
  data_files:
  - split: train
    path: sft_Mixtral-8x22B-v0.1/**/*.parquet
- config_name: sft_mixtral-8x22b-v0-1_nemotron_4_340b
  data_files:
  - split: train
    path: sft_Mixtral-8x22B-v0.1_Nemotron_4_340B/**/*.parquet
- config_name: sft_mixtral-8x22b-v0-1_code
  data_files:
  - split: train
    path: sft_Mixtral-8x22B-v0.1_code/**/*.parquet
- config_name: sft_na
  data_files:
  - split: train
    path: sft_NA/**/*.parquet
- config_name: sft_nemotron_4_340b
  data_files:
  - split: train
    path: sft_Nemotron_4_340B/**/*.parquet
- config_name: sft_qwen2-5-72b-instruct
  data_files:
  - split: train
    path: sft_Qwen2.5-72B-Instruct/**/*.parquet
- config_name: sft_qwen3-30b-a3b_qwen3-235b-a22b
  data_files:
  - split: train
    path: sft_Qwen3-30B-A3B_Qwen3-235B-A22B/**/*.parquet
- config_name: sft_qwen3
  data_files:
  - split: train
    path: sft_Qwen3/**/*.parquet
- config_name: sft_cosmopedia-v2
  data_files:
  - split: train
    path: sft_cosmopedia-v2/**/*.parquet
- config_name: sft_ifeval-like-data
  data_files:
  - split: train
    path: sft_ifeval-like-data/**/*.parquet
- config_name: sft_stem
  data_files:
  - split: train
    path: sft_stem/**/*.parquet
- config_name: sft_thinking-code
  data_files:
  - split: train
    path: sft_thinking-code/**/*.parquet
- config_name: sft_thinking-math
  data_files:
  - split: train
    path: sft_thinking-math/**/*.parquet
- config_name: sft_thinking-sft
  data_files:
  - split: train
    path: sft_thinking-sft/**/*.parquet
- config_name: longmino_2e13
  data_files:
  - split: train
    path: longmino_2e13/**/*.parquet
- config_name: longmino_2e14
  data_files:
  - split: train
    path: longmino_2e14/**/*.parquet
- config_name: longmino_2e15
  data_files:
  - split: train
    path: longmino_2e15/**/*.parquet
- config_name: longmino_2e16
  data_files:
  - split: train
    path: longmino_2e16/**/*.parquet
- config_name: longmino_2e17
  data_files:
  - split: train
    path: longmino_2e17/**/*.parquet
- config_name: nemotron_code
  data_files:
  - split: train
    path: nemotron_code/**/*.parquet
- config_name: nemotron_reasoning
  data_files:
  - split: train
    path: nemotron_reasoning/**/*.parquet
- config_name: nemotron_text
  data_files:
  - split: train
    path: nemotron_text/**/*.parquet
---

# Dataset

Each example contains `prompt` (chat format) and `target` fields.

```python
from datasets import load_dataset
ds = load_dataset("leonli66/compression-pretraining-data", "<config_name>")
```