Overwrite defaults + add visible note

README.md

@@ -4379,11 +4379,14 @@ configs:
 - config_name: volume_of_distribution_at_steady_state_lombardo_et_al-multimodal
   data_files:
   - split: test
-    path: volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/test-*
+    path: 
+      volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/test-*
   - split: train
-    path: volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/train-*
+    path: 
+      volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/train-*
   - split: validation
-    path: volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/validation-*
+    path: 
+      volume_of_distribution_at_steady_state_lombardo_et_al-multimodal/validation-*
 - config_name: zinc-multimodal
   data_files:
   - split: train
@@ -4432,276 +4435,191 @@ tags:
 - property-prediction
 dataset_version: 1.0.0
 dataset_release_date: '2025-05-18'
-dataset_citation: "@article{mirza2025chempile0,\n  title   = {ChemPile: A 250GB Diverse\
-  \ and Curated Dataset for Chemical Foundation Models},\n  author  = {Adrian Mirza\
-  \ and Nawaf Alampara and Martiño Ríos-García and others},\n  year    = {2025},\n\
-  \  journal = {arXiv preprint arXiv:2505.12534}\n}"
+dataset_citation: "@article{mirza2025chempile0,\n  title   = {ChemPile: A 250GB Diverse
+  and Curated Dataset for Chemical Foundation Models},\n  author  = {Adrian Mirza
+  and Nawaf Alampara and Martiño Ríos-García and others},\n  year    = {2025},\n \
+  \ journal = {arXiv preprint arXiv:2505.12534}\n}"
 ---
-# ChemPile-MLIFT
+# ChemPile-Reasoning
 
 <div align="center">
 
 ![ChemPile Logo](CHEMPILE_LOGO.png)
   
-[![Dataset](https://img.shields.io/badge/🤗%20Hugging%20Face-Dataset-yellow)](https://huggingface.co/datasets/jablonkagroup/chempile-mlift)
-[![License: CC BY-NC-SA 4.0](https://img.shields.io/badge/License-CC%20BY--NC--SA%204.0-blue.svg)](https://creativecommons.org/licenses/by-nc-sa/4.0/)
+[![Dataset](https://img.shields.io/badge/🤗%20Hugging%20Face-Dataset-yellow)](https://huggingface.co/datasets/jablonkagroup/chempile-reasoning)
+[![License: CC BY-SA 4.0](https://img.shields.io/badge/License-CC%20BY--SA%204.0-blue.svg)](https://creativecommons.org/licenses/by-sa/4.0/)
 [![Paper](https://img.shields.io/badge/📄-Paper-red)](https://arxiv.org/abs/2505.12534)
 [![Website](https://img.shields.io/badge/🌐-Website-green)](https://chempile.lamalab.org/)
 
-*A comprehensive multimodal dataset for chemistry property prediction using vision large language models*
+*A comprehensive collection of reasoning tasks for chemistry, spectral analysis, and scientific understanding*
 
 </div>
 
 ## 📋 Dataset Summary
 
-ChemPile-MLIFT is a dataset designed for multimodal chemistry property prediction tasks, specifically focusing on the prediction of chemical properties using vision large language models (VLLMs). It is part of the ChemPile project, which aims to create a comprehensive collection of chemistry-related data for training LLMs. The dataset includes a wide range of chemical properties and is structured to facilitate the training of models that can understand and predict chemical properties based on textual descriptions, molecular representations and the image of the corresponding molecule. Thus, each example in the dataset contains the image of the molecule involved.
+ChemPile-Reasoning is a dataset designed for reasoning tasks in the field of chemistry. It is part of the ChemPile project, which aims to create a comprehensive collection of chemistry-related data for training language models. This dataset includes a variety of reasoning tasks derived from scientific Stack Exchange platforms, as well as reasoning traces from state-of-the-art (SOTA) language models. The dataset is structured to facilitate the evaluation of reasoning capabilities in chemistry-related contexts.
 
-The origin of the dataset property data is from well-known chemistry datasets such as the QM9 dataset, which contains quantum mechanical properties of small organic molecules, and the RDKit dataset, which includes a wide range of chemical properties derived from molecular structures. Each of the subsets or Hugging Face configurations corresponds to a different source of chemical property data, allowing for diverse training scenarios:
+The dataset includes different subsets or Hugging Face configurations that correspond to different sources of scientific material:
 
-### 📊 Dataset Statistics
+- chemistry_stackexchange-completion_0
+- chemistry_stackexchange-completion_1
+- chemistry_stackexchange-instruction_0
+- chemistry_stackexchange-instruction_1
+- chemistry_stackexchange-instruction_2
+- chemistry_stackexchange-raw_data
+- claude-3.5-distilled-spectral-reasoning-default
+- mattermodeling_stackexchange-completion_0
+- mattermodeling_stackexchange-completion_1
+- mattermodeling_stackexchange-instruction_0
+- mattermodeling_stackexchange-instruction_1
+- mattermodeling_stackexchange-instruction_2
+- mattermodeling_stackexchange-raw_data
+- physics_stackexchange-completion_0
+- physics_stackexchange-completion_1
+- physics_stackexchange-instruction_0
+- physics_stackexchange-instruction_1
+- physics_stackexchange-instruction_2
+- physics_stackexchange-raw_data
+- spectra_reasoning_deepseek-default
+- spectra_reasoning_deepseek_mcq-default
 
-The resulting dataset contains **61.5M examples** with one image per example, making it a substantial resource for training and evaluating VLLMs in the field of chemistry.
+All the content is made open-source under the license cc-by-sa-4.0, allowing for free use and redistribution with proper attribution.
 
-## 🗂️ Dataset Configurations
+### 📊 Dataset Statistics
 
-The dataset includes diverse configurations covering various chemical property domains, organized by thematic groups:
+| Subset | Examples | Tokens | Description |
+|--------|----------|--------|-------------|
+| StackExchange | 71,658 | 21.3B | Reasoning tasks from scientific Stack Exchange platforms |
+| Spectra Reasoning | 1,070 | 2.16M | Spectral analysis reasoning traces from SOTA models |
+| **Total** | **~72.7K** | **~21.3B** | Scientific reasoning tasks and traces |
 
-### 💊 Drug Discovery and ADMET Properties
+## 🗂️ Dataset Configurations
 
-- BACE-multimodal
-- BBBP-multimodal
-- bioavailability_ma_et_al-multimodal
-- blood_brain_barrier_martins_et_al-multimodal
-- caco2_wang-multimodal
-- clearance_astrazeneca-multimodal
-- cyp2c9_substrate_carbonmangels-multimodal
-- cyp2d6_substrate_carbonmangels-multimodal
-- cyp3a4_substrate_carbonmangels-multimodal
-- cyp_p450_1a2_inhibition_veith_et_al-multimodal
-- cyp_p450_2c19_inhibition_veith_et_al-multimodal
-- cyp_p450_2c9_inhibition_veith_et_al-multimodal
-- cyp_p450_2d6_inhibition_veith_et_al-multimodal
-- cyp_p450_3a4_inhibition_veith_et_al-multimodal
-- drug_induced_liver_injury-multimodal
-- freesolv-multimodal
-- half_life_obach-multimodal
-- human_intestinal_absorption-multimodal
-- lipophilicity-multimodal
-- p_glycoprotein_inhibition_broccatelli_et_al-multimodal
-- pampa_ncats-multimodal
-- solubility_aqsoldb-multimodal
-- volume_of_distribution_at_steady_state_lombardo_et_al-multimodal
+### 🧪 Spectra Reasoning
 
-### ⚠️ Toxicology and Safety Assessment
+The Spectra Reasoning subsets of ChemPile-Reasoning contain reasoning tasks derived from spectral data, specifically focusing on the analysis and interpretation of spectral information. The dataset includes three configurations: two distilled for DeepSeek-R1 model reasoning about a series of spectra (proton and carbon NMR and IR) for one molecule, one open-ended and another for multiple-choice questions (MCQ) based on spectral data, and other configuration distilled from Claude-3.5-Sonnet for single-spectra reasoning (only proton NMR). The dataset is designed to evaluate the reasoning capabilities of language models in the context of spectral analysis.
 
-- ames_mutagenicity-multimodal
-- carcinogens-multimodal
-- clintox-multimodal
-- herg_blockers-multimodal
-- herg_central_at_10uM-multimodal
-- herg_central_at_1uM-multimodal
-- herg_central_inhib-multimodal
-- herg_karim_et_al-multimodal
-- ld50_catmos-multimodal
-- ld50_zhu-multimodal
-- nr_ahr_tox21-multimodal
-- nr_ar_lbd_tox21-multimodal
-- nr_ar_tox21-multimodal
-- nr_aromatase_tox21-multimodal
-- nr_er_lbd_tox21-multimodal
-- nr_er_tox21-multimodal
-- nr_ppar_gamma_tox21-multimodal
-- SIDER-multimodal
-- sigma_aldrich_safety_data-multimodal
-- skin_reaction-multimodal
-- sr_are_tox21-multimodal
-- sr_atad5_tox21-multimodal
-- sr_hse_tox21-multimodal
-- sr_mmp_tox21-multimodal
-- sr_p53_tox21-multimodal
+**DeepSeek Configurations Fields**:
+- `smiles`: The SMILES representation of the molecule associated with the spectral data
+- `reasoning`: The reasoning trace or explanation provided by the model for the spectral analysis
+- `response`: The model's response to the spectral reasoning task
+- `response_smiles`: The SMILES representation of the molecule parsed from the model's response
+- `correct`: If the model's response is correct or not, based on the spectral data
+- `question`: The question or task related to the spectral data that the model is addressing
+- `text`: The joined text of the question, reasoning, and response for the model's output
 
-### 🎯 Bioactivity and Target Interaction
+**Claude-3.5-Sonnet Configuration Fields**:
+- `prompt`: The prompt or question related to the spectral data
+- `extracted_reasoning`: The reasoning trace or explanation with the final answer provided by the model for the spectral analysis
+- `text`: The joined text of the prompt and extracted reasoning for the model's output
+- `index`: The index of the example in the dataset
 
-- cav3_t-type_calcium_channels_butkiewicz-multimodal
-- chembl_v29-multimodal
-- hiv-multimodal
-- m1_muscarinic_receptor_agonists_butkiewicz-multimodal
-- m1_muscarinic_receptor_antagonists_butkiewicz-multimodal
-- MUV_466-multimodal
-- MUV_548-multimodal
-- MUV_600-multimodal
-- MUV_644-multimodal
-- MUV_652-multimodal
-- MUV_689-multimodal
-- MUV_692-multimodal
-- MUV_712-multimodal
-- MUV_713-multimodal
-- MUV_733-multimodal
-- MUV_737-multimodal
-- MUV_810-multimodal
-- MUV_832-multimodal
-- MUV_846-multimodal
-- MUV_852-multimodal
-- MUV_858-multimodal
-- MUV_859-multimodal
-- orexin1_receptor_butkiewicz-multimodal
-- sarscov2_3clpro_diamond-multimodal
-- sarscov2_vitro_touret-multimodal
-- serine_threonine_kinase_33_butkiewicz-multimodal
-- uniprot_binding_single-multimodal
-- uniprot_binding_sites_multiple-multimodal
+**Statistics**: 1.07K examples with a total of over 2.16M tokens
 
-### 🔬 Computational Chemistry and Quantum Properties
+### 📚 StackExchange
 
-- flashpoint-multimodal
-- mol2svg-multimodal
-- opv-multimodal
-- qm8-multimodal
-- qm9-multimodal
-- rdkit_features-multimodal
-- rdkit_features-multimodal-chunk-1
-- rdkit_features-multimodal-chunk-2
-- rdkit_features-multimodal-chunk-3
-- smiles_to_3d-multimodal
-- thermosol-multimodal
+The StackExchange subsets of ChemPile-Reasoning contains reasoning tasks derived from scientific Stack Exchange platforms, specifically from the chemistry, matter modeling and physics domains. For each of the datasets, different configs are available: two in completion format and three in instruction format, as well as the raw data. For the different formats, different text templates are used to structure the data. The completion format is designed for tasks where the model needs to generate a response based on a given input, while the instruction format provides a more structured approach with specific instructions for the model to follow. The raw data config contains the original data without any modifications or formatting.
 
-### 📚 Chemical Knowledge Bases and Databases
+**Completion and Instruction Format Fields**:
+- `text`: The original text from the Stack Exchange post
+- `input`: The input text for the model, which may include the question or context
+- `output`: The expected output or answer to the question
+- `answer_choices`: A list of possible answer choices for the question
+- `correct_output_index`: The index of the correct answer in the answer_choices list
 
-- aminoacids-multimodal
-- chebi_20-multimodal
-- drugchat_liang_zhang_et_al-multimodal
-- mona-multimodal
-- moses-multimodal
-- RedDB-multimodal
-- zinc-multimodal
+**Raw Data Configuration Fields**:
+- `title`: The title of the Stack Exchange post
+- `q`: The question text from the Stack Exchange post
+- `a`: The answer text from the Stack Exchange post
+- `split`: The split of the dataset (train, test, or validation)
+- `index`: The index of the post in the dataset
+- `text`: The joined text of the title, question, and answer for the post
 
-## 📜 License
+**Statistics**: 71,658 examples with a total of over 21.3B tokens
 
-All content is made open-source under the **CC BY-NC-SA 4.0** license, allowing for:
+## � License
 
-- ✅ Non-commercial use and sharing with attribution
+All content is released under the **CC BY-SA 4.0** license, which allows for:
+- ✅ Free use and distribution
+- ✅ Commercial use
 - ✅ Modification and derivatives
 - ⚠️ Attribution required
-- ⚠️ Non-commercial use only
-
-## 📖 Data Fields
-
-The dataset contains the following fields for all the configurations allowing for a consistent structure across different chemical property datasets:
-
-- **`IMAGE`**: The image representation of the molecule, typically in PNG format. This image is crucial for vision-based tasks and is derived from the molecular structure.
-- **`template`**: The filled template with the SMILES of the molecule and the corresponding chemical property.
-- **`template_original`**: The template or schema used for the chemical property prediction task. This field provides a structured format for the input and output data, with different templates used for ensure diversity in the tasks.
-- **`SMILES`**: The SMILES (Simplified Molecular Input Line Entry System) representation of the molecule.
-- **`SELFIES`**: The SELFIES (SELF-referencIng Embedded Strings) representation of the molecule, which is a more robust alternative to SMILES for representing chemical structures.
-- **`InChI`**: The International Chemical Identifier (InChI) representation of the molecule, providing a unique identifier for chemical substances.
-- **`IUPAC`**: The IUPAC (International Union of Pure and Applied Chemistry) name of the molecule, which is a systematic way to name chemical compounds.
-- **`split`**: The split of the dataset, indicating whether the example is part of the training, validation, or test set. This field helps in organizing the dataset for model training and evaluation.
-- **property or properties**: The specific chemical property or properties being predicted. This field varies depending on the dataset configuration and can include properties such as solubility, bioavailability, toxicity, etc. Each configuration focuses on a different set of chemical properties.
-
-## 🔬 Dataset Groups Detailed Description
-
-### 💊 Drug Discovery and ADMET Properties
-
-The Drug Discovery and ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) group encompasses datasets focused on pharmaceutical compound development and safety assessment. This collection includes bioavailability prediction (bioavailability_ma_et_al-multimodal), blood-brain barrier permeability (BBBP-multimodal, blood_brain_barrier_martins_et_al-multimodal), intestinal permeability (caco2_wang-multimodal, pampa_ncats-multimodal), hepatic clearance (clearance_astrazeneca-multimodal), drug-induced liver injury assessment (drug_induced_liver_injury-multimodal), and various CYP450 enzyme interactions (cyp2c9_substrate_carbonmangels-multimodal, cyp2d6_substrate_carbonmangels-multimodal, cyp3a4_substrate_carbonmangels-multimodal, cyp_p450_1a2_inhibition_veith_et_al-multimodal, cyp_p450_2c19_inhibition_veith_et_al-multimodal, cyp_p450_2c9_inhibition_veith_et_al-multimodal, cyp_p450_2d6_inhibition_veith_et_al-multimodal, cyp_p450_3a4_inhibition_veith_et_al-multimodal). The group also covers crucial pharmacokinetic properties such as half-life prediction (half_life_obach-multimodal), human intestinal absorption (human_intestinal_absorption-multimodal), lipophilicity (lipophilicity-multimodal), volume of distribution (volume_of_distribution_at_steady_state_lombardo_et_al-multimodal), P-glycoprotein inhibition (p_glycoprotein_inhibition_broccatelli_et_al-multimodal), and solubility (freesolv-multimodal). These datasets are essential for early-stage drug development, helping predict whether compounds will have favorable drug-like properties before expensive clinical trials.
+- ⚠️ Share-alike requirements
 
-### ⚠️ Toxicology and Safety Assessment
-
-The Toxicology and Safety Assessment group focuses on predicting harmful effects of chemical compounds across various biological systems. This includes mutagenicity prediction (ames_mutagenicity-multimodal), carcinogenicity assessment (carcinogens-multimodal), acute toxicity (ld50_catmos-multimodal, ld50_zhu-multimodal), and comprehensive toxicity screening through the Tox21 initiative datasets (nr_ahr_tox21-multimodal, nr_ar_lbd_tox21-multimodal, nr_ar_tox21-multimodal, nr_aromatase_tox21-multimodal, nr_er_lbd_tox21-multimodal, nr_er_tox21-multimodal, nr_ppar_gamma_tox21-multimodal, sr_are_tox21-multimodal, sr_atad5_tox21-multimodal, sr_hse_tox21-multimodal, sr_mmp_tox21-multimodal, sr_p53_tox21-multimodal). The group also includes specialized toxicity assessments such as hERG channel blocking potential (herg_blockers-multimodal, herg_central_at_10uM-multimodal, herg_central_at_1uM-multimodal, herg_central_inhib-multimodal, herg_karim_et_al-multimodal), clinical toxicity prediction (clintox-multimodal), adverse drug reactions (SIDER-multimodal), skin reactions (skin_reaction-multimodal), and safety data assessment (sigma_aldrich_safety_data-multimodal). These datasets are crucial for environmental safety assessment and pharmaceutical safety profiling.
-
-### 🎯 Bioactivity and Target Interaction
-
-The Bioactivity and Target Interaction group contains datasets focused on molecular interactions with specific biological targets. This includes enzyme inhibition studies (BACE-multimodal for β-secretase), receptor binding and modulation data from the Butkiewicz collection (cav3_t-type_calcium_channels_butkiewicz-multimodal, m1_muscarinic_receptor_agonists_butkiewicz-multimodal, m1_muscarinic_receptor_antagonists_butkiewicz-multimodal, orexin1_receptor_butkiewicz-multimodal, serine_threonine_kinase_33_butkiewicz-multimodal), and comprehensive screening datasets like MUV (Maximum Unbiased Validation) series covering various protein targets (MUV_466-multimodal, MUV_548-multimodal, MUV_600-multimodal, MUV_644-multimodal, MUV_652-multimodal, MUV_689-multimodal, MUV_692-multimodal, MUV_712-multimodal, MUV_713-multimodal, MUV_733-multimodal, MUV_737-multimodal, MUV_810-multimodal, MUV_832-multimodal, MUV_846-multimodal, MUV_852-multimodal, MUV_858-multimodal, MUV_859-multimodal). The group also includes antiviral activity data (hiv-multimodal, sarscov2_3clpro_diamond-multimodal, sarscov2_vitro_touret-multimodal), protein binding studies (uniprot_binding_single-multimodal, uniprot_binding_sites_multiple-multimodal), and comprehensive chemical bioactivity databases (chembl_v29-multimodal). These datasets enable the development of target-specific therapeutics and understanding of molecular mechanisms of action.
-
-### 🔬 Computational Chemistry and Quantum Properties
-
-The Computational Chemistry and Quantum Properties group encompasses datasets for predicting fundamental quantum mechanical and computational chemistry properties. This includes quantum mechanical property prediction (qm8-multimodal, qm9-multimodal), molecular descriptor calculations (rdkit_features-multimodal, rdkit_features-multimodal-chunk-1, rdkit_features-multimodal-chunk-2, rdkit_features-multimodal-chunk-3), thermodynamic properties (flashpoint-multimodal, thermosol-multimodal), and molecular visualization and structure generation (mol2svg-multimodal, smiles_to_3d-multimodal). These datasets are fundamental for computational chemistry research, enabling the prediction of molecular properties from first principles and supporting the development of new theoretical models and computational methods.
-
-### 📚 Chemical Knowledge Bases and Databases
-
-The Chemical Knowledge Bases and Databases group contains datasets derived from established chemical and biological databases. This includes comprehensive chemical databases (chebi_20-multimodal, zinc-multimodal), mass spectrometry databases (mona-multimodal), molecular generation and chemical space exploration (moses-multimodal), conversational chemistry data (drugchat_liang_zhang_et_al-multimodal), and amino acid properties (aminoacids-multimodal). These datasets provide broad coverage of chemical space and enable the development of AI systems that can navigate and reason about diverse chemical information from established scientific databases and knowledge repositories.
-
-## 🚀 Usage
+## �🚀 Quick Start
 
 ```python
 from datasets import load_dataset, get_dataset_config_names
 
 # Print available configs for the dataset
-configs = get_dataset_config_names("jablonkagroup/chempile-mlift")
+configs = get_dataset_config_names("jablonkagroup/chempile-reasoning")
 print(f"Available configs: {configs}")
-# Available configs: ['BACE-multimodal', 'BBBP-multimodal', 'MUV_466-multimodal', ...
+# Available configs: ['chemistry_stackexchange-completion_0', 'chemistry_stackexchang...
 
-dataset = load_dataset("jablonkagroup/chempile-mlift", name=configs[0])
-# Loading config: BACE-multimodal
+dataset = load_dataset("jablonkagroup/chempile-reasoning", name=configs[0])
+# Loading config: chemistry_stackexchange-completion_0
 
 print(dataset)
 # DatasetDict({
-    # train: Dataset({
-        # features: ['IMAGE', 'template', 'template_original', 'SMILES', ... 
-        # num_rows: 5440
-    # })
-    # validation: Dataset({
-        # features: ['IMAGE', 'template', 'template_original', 'SMILES', ...
-        # num_rows: 480
-    # })
-    # test: Dataset({
-        # features: ['IMAGE', 'template', 'template_original', 'SMILES', ...
-    # })
+#     train: Dataset({
+#         features: ['text', 'input', 'output', 'answer_choices', 'correct_output_index'],
+#         num_rows: 3207
+#     })
+#     test: Dataset({
+#         features: ['text', 'input', 'output', 'answer_choices', 'correct_output_index'],
+#         num_rows: 687
+#     })
+#     val: Dataset({
+#         features: ['text', 'input', 'output', 'answer_choices', 'correct_output_index'],
+#         num_rows: 687
+#     })
 # })
 
 split_name = list(dataset.keys())[0]
 sample = dataset[split_name][0]
 print(sample)
 # {
-#     'IMAGE': <PIL.PngImagePlugin.PngImageFile ..., 
-#     'template': 'The chemical with the SMILES ..., 
-#     'template_original': 'The {#compound|chemical!} ...', 
-#     'SMILES': 'Cc1ccccc1-c1ccc2nc(N)c(C[C@@H](C)C(=O)...', 
-#     'SMILES_ORIGINAL': 'Cc1ccccc1-c1ccc2nc(N)c(C[C@@H]...', 
-#     'SELFIES': '[C][C][=C][C][=C][C][=C][Ring1]...', 
-#     'InChI': 'InChI=1S/C27H33N3O2/c1-17-7-5-6-8-...', 
-#     'IUPAC': '(2R)-3-[2-amino-6-(2-methylphenyl)...', 
-#     'split': 'train', 
-#     'pIC50': 9.1549015, 
-#     'BACE_inhibition': 1
+#     'text': 'The answer to the query "We know that the...
+#     'input': 'The answer to the query "We know that the...
+#     'output': '', 
+#     'answer_choices': [], 
+#     'correct_output_index': None
 # }
 ```
 
 ## 🎯 Use Cases
 
-- **🖼️ Multimodal Chemical Property Prediction**: Training vision-language models to predict molecular properties using both molecular images and textual descriptions
-- **💊 Drug Discovery**: Building systems for pharmaceutical compound screening using visual molecular representations
-- **⚠️ Safety Assessment**: Developing multimodal models for toxicity and environmental impact prediction
-- **🔬 Materials Design**: Creating AI tools that leverage both visual and textual molecular information for materials science
-- **📖 Scientific Multimodal Understanding**: Training models to understand and reason about chemical information across multiple modalities
+- **� Scientific Reasoning**: Training models for complex chemical and physical reasoning tasks
+- **📊 Spectral Analysis**: Building systems for automated spectral interpretation and structure elucidation
+- **🔬 Educational AI**: Developing tutoring systems for chemistry and materials science education
+- **� Question Answering**: Advanced scientific question-answering systems for research support
+- **🤖 Research Assistance**: Automated analysis and interpretation of scientific problems
 
 ## ⚠️ Limitations & Considerations
 
-- **Scope**: Focused on chemistry and materials science; domain-specific terminology and concepts
-- **Quality**: Variable quality across sources; expert curation applied but some noise may remain
-- **Bias**: Reflects biases present in chemical databases and scientific literature
-- **License**: Non-commercial use only under CC BY-NC-SA 4.0
-- **Language**: Primarily English content
-- **Completeness**: Some datasets may have missing values or incomplete property annotations
-- **Image Quality**: Molecular images are automatically generated and may vary in visual quality
+- **Language**: Primarily English content (monolingual dataset)
+- **Scope**: Focused on chemistry, physics, and materials science; specialized domain knowledge required
+- **Quality**: Variable quality across sources; some reasoning traces may contain errors or inconsistencies
+- **Bias**: Reflects biases present in Stack Exchange communities and model-generated content
+- **Complexity**: Contains advanced scientific concepts that may require domain expertise to validate
 
 ## 🛠️ Data Processing Pipeline
 
-1. **Collection**: Automated extraction from well-known chemistry datasets (QM9, RDKit, Tox21, etc.)
-2. **Standardization**: Consistent formatting and SMILES representation across all configurations
-3. **Image Generation**: Creation of molecular structure images from SMILES representations
-4. **Template Generation**: Conversion of structured data into natural language templates
-5. **Quality Control**: Expert curation and validation of chemical property representations
-6. **Deduplication**: Removal of duplicate entries and data cleaning
+1. **Collection**: Automated extraction from Stack Exchange platforms and model reasoning traces
+2. **Filtering**: Domain-specific filtering for chemistry, physics, and materials science relevance
+3. **Format Conversion**: Multiple formatting approaches (completion, instruction, raw data)
+4. **Quality Control**: Expert validation and automated filtering
+5. **Reasoning Extraction**: Parsing and structuring of model reasoning traces
+6. **Standardization**: Consistent formatting and metadata extraction
 7. **Validation**: Train/validation/test splits and quality checks
 
 ## 🏗️ ChemPile Collection
 
-This dataset is part of the **ChemPile** collection, a comprehensive open dataset containing over 75 billion tokens of curated chemical data for training and evaluating general-purpose models in the chemical sciences.
+This dataset is part of the **ChemPile** collection, a comprehensive open dataset containing over 75 billion tokens of curated chemical data for training and evaluating general-purpose models in the chemical sciences. 
 
 ### Collection Overview
-
 - **📊 Scale**: 75+ billion tokens across multiple modalities
-- **🧬 Modalities**: Structured representations (SMILES, SELFIES, IUPAC, InChI), scientific text, executable code, and molecular images
+- **🧬 Modalities**: Structured representations (SMILES, SELFIES, IUPAC, InChI), scientific text, executable code, reasoning traces, and molecular images
 - **🎯 Design**: Integrates foundational educational knowledge with specialized scientific literature
 - **🔬 Curation**: Extensive expert curation and validation
 - **📈 Benchmarking**: Standardized train/validation/test splits for robust evaluation
@@ -4724,7 +4642,7 @@ If you use this dataset in your research, please cite:
 
 - **Paper**: [arXiv:2505.12534](https://arxiv.org/abs/2505.12534)
 - **Website**: [ChemPile Project](https://chempile.lamalab.org/)
-- **Dataset**: [Hugging Face](https://huggingface.co/datasets/jablonkagroup/chempile-mlift)
+- **Dataset**: [Hugging Face](https://huggingface.co/datasets/jablonkagroup/chempile-reasoning)
 - **Issues**: Please report data issues or questions via the Hugging Face dataset page
 
 ---