---
license: cc-by-4.0
tags:
- chemistry
- instruction-tuning
configs:
- config_name: forward_synthesis
  data_files:
  - split: train
    path: forward_synthesis/train.parquet
  - split: test
    path: forward_synthesis/test.parquet
  - split: test_full
    path: forward_synthesis/test_full.parquet
- config_name: retrosynthesis
  data_files:
  - split: train
    path: retrosynthesis/train.parquet
  - split: test
    path: retrosynthesis/test.parquet
  - split: test_full
    path: retrosynthesis/test_full.parquet
- config_name: molecule_captioning
  data_files:
  - split: train
    path: molecule_captioning/train.parquet
  - split: test
    path: molecule_captioning/test.parquet
  - split: test_full
    path: molecule_captioning/test_full.parquet
- config_name: molecule_generation
  data_files:
  - split: train
    path: molecule_generation/train.parquet
  - split: test
    path: molecule_generation/test.parquet
  - split: test_full
    path: molecule_generation/test_full.parquet
- config_name: name_conversion-i2f
  data_files:
  - split: train
    path: name_conversion-i2f/train.parquet
  - split: test
    path: name_conversion-i2f/test.parquet
  - split: test_full
    path: name_conversion-i2f/test_full.parquet
- config_name: name_conversion-i2s
  data_files:
  - split: train
    path: name_conversion-i2s/train.parquet
  - split: test
    path: name_conversion-i2s/test.parquet
  - split: test_full
    path: name_conversion-i2s/test_full.parquet
- config_name: name_conversion-s2f
  data_files:
  - split: train
    path: name_conversion-s2f/train.parquet
  - split: test
    path: name_conversion-s2f/test.parquet
  - split: test_full
    path: name_conversion-s2f/test_full.parquet
- config_name: name_conversion-s2i
  data_files:
  - split: train
    path: name_conversion-s2i/train.parquet
  - split: test
    path: name_conversion-s2i/test.parquet
  - split: test_full
    path: name_conversion-s2i/test_full.parquet
- config_name: property_prediction-esol
  data_files:
  - split: train
    path: property_prediction-esol/train.parquet
  - split: test
    path: property_prediction-esol/test.parquet
  - split: test_full
    path: property_prediction-esol/test_full.parquet
- config_name: property_prediction-lipo
  data_files:
  - split: train
    path: property_prediction-lipo/train.parquet
  - split: test
    path: property_prediction-lipo/test.parquet
  - split: test_full
    path: property_prediction-lipo/test_full.parquet
- config_name: property_prediction-bbbp
  data_files:
  - split: train
    path: property_prediction-bbbp/train.parquet
  - split: test
    path: property_prediction-bbbp/test.parquet
  - split: test_full
    path: property_prediction-bbbp/test_full.parquet
- config_name: property_prediction-clintox
  data_files:
  - split: train
    path: property_prediction-clintox/train.parquet
  - split: test
    path: property_prediction-clintox/test.parquet
  - split: test_full
    path: property_prediction-clintox/test_full.parquet
- config_name: property_prediction-hiv
  data_files:
  - split: train
    path: property_prediction-hiv/train.parquet
  - split: test
    path: property_prediction-hiv/test.parquet
  - split: test_full
    path: property_prediction-hiv/test_full.parquet
- config_name: property_prediction-sider
  data_files:
  - split: train
    path: property_prediction-sider/train.parquet
  - split: test
    path: property_prediction-sider/test.parquet
  - split: test_full
    path: property_prediction-sider/test_full.parquet
- config_name: all
  data_files:
  - split: train
    path:
    - forward_synthesis/train.parquet
    - retrosynthesis/train.parquet
    - molecule_captioning/train.parquet
    - molecule_generation/train.parquet
    - name_conversion-i2f/train.parquet
    - name_conversion-i2s/train.parquet
    - name_conversion-s2f/train.parquet
    - name_conversion-s2i/train.parquet
    - property_prediction-esol/train.parquet
    - property_prediction-lipo/train.parquet
    - property_prediction-bbbp/train.parquet
    - property_prediction-clintox/train.parquet
    - property_prediction-hiv/train.parquet
    - property_prediction-sider/train.parquet
  - split: test
    path:
    - forward_synthesis/test.parquet
    - retrosynthesis/test.parquet
    - molecule_captioning/test.parquet
    - molecule_generation/test.parquet
    - name_conversion-i2f/test.parquet
    - name_conversion-i2s/test.parquet
    - name_conversion-s2f/test.parquet
    - name_conversion-s2i/test.parquet
    - property_prediction-esol/test.parquet
    - property_prediction-lipo/test.parquet
    - property_prediction-bbbp/test.parquet
    - property_prediction-clintox/test.parquet
    - property_prediction-hiv/test.parquet
    - property_prediction-sider/test.parquet
  - split: test_full
    path:
    - forward_synthesis/test_full.parquet
    - retrosynthesis/test_full.parquet
    - molecule_captioning/test_full.parquet
    - molecule_generation/test_full.parquet
    - name_conversion-i2f/test_full.parquet
    - name_conversion-i2s/test_full.parquet
    - name_conversion-s2f/test_full.parquet
    - name_conversion-s2i/test_full.parquet
    - property_prediction-esol/test_full.parquet
    - property_prediction-lipo/test_full.parquet
    - property_prediction-bbbp/test_full.parquet
    - property_prediction-clintox/test_full.parquet
    - property_prediction-hiv/test_full.parquet
    - property_prediction-sider/test_full.parquet
---

# SMolInstruct, hub-native

`osunlp/SMolInstruct` (`b3e028007c6abc043f62ed0aef4824dc9e86bacc`), the official instruction view (`insert_core_tags=True`) of all 14 tasks, one subset per task, with the unified columns `id`, `query`, `ground_truth`, `response`, `task`, `subtask` and `scoring_context` added beside every original one. `query` is `input` plus one fixed sentence per task naming its reply tag. `test` is the official 200-per-task subset; `test_full` is the whole test split. See `src/scripts/datasets/README.md` in the AutoDataSci repository for what those columns mean.

Row counts: {"train": 3288855, "test": 2653, "test_full": 33061}.

```json
{
  "forward_synthesis": {
    "train": 971809,
    "test": 200,
    "test_full": 4062
  },
  "retrosynthesis": {
    "train": 941735,
    "test": 200,
    "test_full": 4156
  },
  "molecule_captioning": {
    "train": 56498,
    "test": 200,
    "test_full": 2538
  },
  "molecule_generation": {
    "train": 56498,
    "test": 200,
    "test_full": 2493
  },
  "name_conversion-i2f": {
    "train": 300000,
    "test": 200,
    "test_full": 2993
  },
  "name_conversion-i2s": {
    "train": 299890,
    "test": 200,
    "test_full": 2993
  },
  "name_conversion-s2f": {
    "train": 299890,
    "test": 200,
    "test_full": 2993
  },
  "name_conversion-s2i": {
    "train": 299890,
    "test": 200,
    "test_full": 2993
  },
  "property_prediction-esol": {
    "train": 888,
    "test": 112,
    "test_full": 112
  },
  "property_prediction-lipo": {
    "train": 3360,
    "test": 200,
    "test_full": 420
  },
  "property_prediction-bbbp": {
    "train": 1569,
    "test": 197,
    "test_full": 197
  },
  "property_prediction-clintox": {
    "train": 1144,
    "test": 144,
    "test_full": 144
  },
  "property_prediction-hiv": {
    "train": 32864,
    "test": 200,
    "test_full": 4107
  },
  "property_prediction-sider": {
    "train": 22820,
    "test": 200,
    "test_full": 2860
  }
}
```