Initial upload

README.md

@@ -0,0 +1,124 @@
+---
+license: mit
+task_categories:
+  - text-generation
+tags:
+  - chemistry
+  - drug-discovery
+  - smiles
+  - molecules
+  - instruction-tuning
+size_categories:
+  - 1M<n<10M
+configs:
+  - config_name: all
+    data_files: all/*.parquet
+    default: true
+  - config_name: smileyllama
+    data_files: smileyllama/*.parquet
+  - config_name: linkllama
+    data_files: linkllama/*.parquet
+dataset_info:
+  features:
+    - name: instruction
+      dtype: string
+    - name: input
+      dtype: string
+    - name: output
+      dtype: string
+    - name: source
+      dtype: string
+---
+
+# CADD-Instruct
+
+An instruction-tuning dataset for computer-aided drug design (CADD).
+
+## Format
+
+Every row follows the standard instruction / input / output schema:
+
+| field | description |
+| --- | --- |
+| `instruction` | The system-level role/task description. |
+| `input` | The user request, listing the target molecular properties. |
+| `output` | A SMILES string satisfying the request. |
+| `source` | Origin of the row (`smileyllama` or `linkllama`). |
+
+### Example
+
+```json
+{
+  "instruction": "You love and excel at generating SMILES strings of drug-like molecules",
+  "input": "Output a SMILES string for a drug-like molecule with the following properties: <= 90 TPSA, lacks bad SMARTS, has covalent warheads (acrylamides), <= 5 LogP, <= 0.4 Fraction sp3, <= 10 Rotatable bonds:",
+  "output": "n1cc(/C=C/C(=O)Nc2ccc(cc2)COC)c(cc1)-c1cnn(C)c1",
+  "source": "smileyllama"
+}
+```
+
+## Configurations
+
+| config | rows | description |
+| --- | --- | --- |
+| `all` (default) | ~3.86M | Combined dataset (all sources). |
+| `smileyllama` | ~2.33M | SmileyLlama SFT data. |
+| `linkllama` | ~1.53M | LinkLlama cap50 training data. |
+
+## Usage
+
+Load the full dataset:
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("cadd-instruct")          # default "all" config
+print(ds["train"][0])
+```
+
+Filter the combined dataset by `source`:
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("cadd-instruct", split="train")
+smiley = ds.filter(lambda ex: ex["source"] == "smileyllama")
+```
+
+Or load a specific source directly:
+
+```python
+from datasets import load_dataset
+
+smiley = load_dataset("cadd-instruct", "smileyllama")
+link   = load_dataset("cadd-instruct", "linkllama")
+```
+
+Stream instead of downloading everything up front:
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("cadd-instruct", split="train", streaming=True)
+for example in ds:
+    print(example["input"], "->", example["output"])
+    break
+```
+
+Format a row into a prompt for supervised fine-tuning:
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("cadd-instruct", split="train")
+
+def to_prompt(ex):
+    return {"text": f"{ex['instruction']}\n\n{ex['input']}\n{ex['output']}"}
+
+ds = ds.map(to_prompt)
+print(ds[0]["text"])
+```
+
+## Sources
+
+- **smileyllama**: SFT data for SmileyLlama — https://figshare.com/articles/dataset/SFT_Data_for_SmileyLlama/30854573
+- **linkllama**: https://huggingface.co/datasets/THGLab/LinkLlama-cap50-train

all/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ab731616a4a6f9084260b180c3668b04c5e512b7a6fa97df8bc7f8b95c1339c
+size 305431882

linkllama/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c12244729a62b3dce0b1f52e3a95470e6a998a9ddc5f02bf4ec0b26d3d1ed8a
+size 136829760

smileyllama/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e93399c684d4f13fccb7bf4c6b5aa89c87c7be7921489ef2007efd03c13d9ba9
+size 168228366