Datasets:

aiacontext
/

mini-enedina-dataset

@@ -92,7 +92,7 @@ configs:
         path: data/validation.parquet
       - split: test
         path: data/test.parquet
-license: apache-2.0
 task_categories:
   - text-generation
 language:
@@ -106,186 +106,199 @@ tags:
   - multimodal-text
   - code-generation
   - enedina
 size_categories:
   - 10K<n<100K
 ---
-# Mini-Enedina Dataset: Timoshenko Beam Analysis (60k)
-Dataset de treinamento para o modelo [Mini-Enedina 37.5M](https://huggingface.co/aiacontext/mini-enedina) -- um language model especializado em analise estrutural de eixos de transmissao usando Teoria de Vigas de Timoshenko.
-## Descricao
-60.000 conversacoes sinteticas em formato **Harmony-Enedina** (variante do ChatML), cobrindo 3 niveis de complexidade progressiva de analise de eixos:
-| Nivel | Analise | Amostras | Tokens/amostra (media) |
-|-------|---------|----------|------------------------|
-| **Bachelor** | Deflexao (V, M, w, theta) | 20,000 | 8,394 |
-| **Master** | + Tensao (Von Mises, sigma_adm) | 20,000 | 9,883 |
-| **Doctor** | + Fadiga (Marin, Goodman) | 20,000 | 12,775 |
-**Total:** 621M tokens (vocabulario de 32,768 tokens via BPE customizado).
-## Formato Harmony-Enedina
-Cada amostra contem uma conversacao completa em formato Harmony com 2 canais (reasoning + response) e metacognicao via headers:
 ```
 <|system|>
-Voce e Enedina, uma engenheira estrutural...
 <|end|>
 <|user|>
-Analise o eixo de transmissao com os seguintes parametros...
 <|end|>
 <|assistant|>
 <|ch1_start|>thinking<|ch1_end|>
-[Raciocinio tecnico detalhado]
 <|ch2_start|>response<|ch2_end|>
-[Resposta formatada com equacoes LaTeX e codigo Python]
 <|end|>
 <|return|>
 ```
-## Features (36 colunas)
-### Identificacao
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `id` | string | Hash SHA-256 unico do sample |
-| `problem_type` | string | Sempre `shaft_timoshenko` |
-| `problem_class` | string | Classe do problema |
-| `level` | string | `bachelor`, `master` ou `doctor` |
-| `curriculum_stage` | int32 | Estagio no curriculum (0, 1, 2) |
-| `analysis_depth` | string | `deflection`, `stress` ou `fatigue` |
-### Parametros do Eixo
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `D_ext_mm` | float64 | Diametro externo [mm] |
-| `D_int_mm` | float64 | Diametro interno [mm] (0 para macico) |
-| `L_mm` | float64 | Comprimento do eixo [mm] |
-| `section_type` | string | `macico` ou `vazado` |
-| `L_D_ratio` | float64 | Razao L/D (esbeltez) |
-| `material_name` | string | Nome do material (ex: AISI 1045) |
-| `E_GPa` | float64 | Modulo de Young [GPa] |
-| `S_y_MPa` | float64 | Tensao de escoamento [MPa] |
-| `S_ut_MPa` | float64 | Tensao ultima [MPa] |
-### Carregamento
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `n_loads` | int32 | Numero de cargas aplicadas |
-| `n_bearings` | int32 | Numero de mancais |
-| `total_load_kN` | float64 | Carga total [kN] |
-### Resultados da Analise
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `V_max_kN` | float64 | Cortante maximo [kN] |
-| `M_max_kNmm` | float64 | Momento fletor maximo [kN.mm] |
-| `w_max_mm` | float64 | Deflexao maxima [mm] |
-| `w_limit_mm` | float64 | Deflexao limite [mm] |
-### Validacao
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `validation_passed` | bool | Todas as validacoes OK |
-| `deflection_passed` | bool | Deflexao dentro do limite |
-| `stress_passed` | string | Resultado da analise de tensao |
-| `fatigue_passed` | string | Resultado da analise de fadiga |
-| `difficulty_score` | float64 | Score de dificuldade [0-1] |
-### Conversacao
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `conversation` | string | Conversacao completa em formato Harmony |
-| `user_tokens` | int32 | Tokens no prompt do usuario |
-| `assistant_tokens` | int32 | Tokens na resposta do assistente |
-| `total_tokens` | int32 | Total de tokens na conversacao |
 ### Metadata
-| Coluna | Tipo | Descricao |
-|--------|------|-----------|
-| `expected_route` | string | Rota esperada (bachelor/master/doctor) |
-| `split` | string | Split: `train`, `val` ou `test` |
-| `fold` | int32 | Fold para cross-validation |
-| `solve_time_ms` | float64 | Tempo de resolucao [ms] |
-| `template_variation` | int32 | Variacao do template utilizada |
-## Estatisticas de Tokens
-| Metrica | Valor |
-|---------|-------|
-| Total de tokens | 621,028,968 |
-| Media por amostra | 10,350 |
-| Minimo | 8,280 |
-| Maximo | 12,996 |
-| Mediana | 9,881 |
-| Tokens usuario (media) | 358 |
-| Tokens assistente (media) | 9,932 |
-## Uso
 ```python
 from datasets import load_dataset
 dataset = load_dataset("aiacontext/mini-enedina-dataset")
-# Acessar splits
-train = dataset["train"]       # 48,000 amostras
-val = dataset["validation"]    # 6,000 amostras
-test = dataset["test"]         # 6,000 amostras
-# Exemplo: filtrar por nivel
 bachelor = train.filter(lambda x: x["level"] == "bachelor")
 doctor = train.filter(lambda x: x["level"] == "doctor")
-# Acessar conversacao
 sample = train[0]
 print(sample["conversation"][:500])
 print(f"Tokens: {sample['total_tokens']}")
 print(f"Level: {sample['level']}")
 ```
-## Geracao dos Dados
-Os dados foram gerados pelo **Factorium SciML** -- um framework Julia/Python para geracao de dados sinteticos cientificos:
-1. **Solver numerico** (Python/NumPy): resolve cada problema de Timoshenko com parametros aleatorios
-2. **Templates variaveis**: 97k combinacoes de descricoes em PT-BR (46^3 variacoes)
-3. **Grounding numerico**: sistema de citacoes que garante consistencia entre parametros de entrada e valores na resposta
-4. **Validacao**: cada sample e verificado (deflexao, tensao, fadiga) antes de inclusao
-### Equacoes Governantes (Timoshenko com peso proprio)
 ```
-V(x)  = RA - q*x - sum(Fi * H(x - xi))           (Esforco cortante)
-M(x)  = RA*x - q*x^2/2 - sum(Fi*(x-xi)*H(x-xi)) (Momento fletor)
-dw/dx = theta + V/(kappa*G*A)                     (Deflexao com cisalhamento)
-q     = rho * g * A * 1e-9                        (Carga distribuida [N/mm])
 ```
-## Modelo Treinado
-O modelo treinado com este dataset esta disponivel em: [aiacontext/mini-enedina](https://huggingface.co/aiacontext/mini-enedina)
-- **Arquitetura:** Dense Transformer (RoPE, RMSNorm, SwiGLU)
-- **Parametros:** 37.5M
-- **Framework:** MLX (Apple Silicon)
-- **Test Loss:** 0.0787 | **Test PPL:** 1.08
-## Licenca
-Apache 2.0
-## Citacao
 ```bibtex
-@misc{mini-enedina-dataset-2026,
-  title={Mini-Enedina Dataset: Timoshenko Beam Analysis},
-  author={AIA Context},
-  year={2026},
-  publisher={Hugging Face},
-  url={https://huggingface.co/datasets/aiacontext/mini-enedina-dataset}
 }
 ```

         path: data/validation.parquet
       - split: test
         path: data/test.parquet
+license: cc-by-4.0
 task_categories:
   - text-generation
 language:
   - multimodal-text
   - code-generation
   - enedina
+  - validated-synthetic-data
+  - physics-informed-ai
 size_categories:
   - 10K<n<100K
 ---
+# Mini-Enedina Dataset: Physically Validated Timoshenko Shaft Analysis (60k)
+Training dataset for [Mini-Enedina 37.5M](https://huggingface.co/aiacontext/mini-enedina) -- a monotropic language model deliberately small and intensively specialized for structural shaft analysis according to Timoshenko beam theory.
+## Dataset Description
+60,000 synthetic conversations in **Harmony-Enedina** format (a ChatML variant), covering three progressively complex levels of shaft analysis:
+| Level | Analysis Scope | Samples | Avg. Tokens/Sample |
+|-------|---------------|---------|---------------------|
+| **Bachelor** | Deflection (V, M, w, theta) | 20,000 | 8,394 |
+| **Master** | + Von Mises stress (sigma_VM, sigma_adm) | 20,000 | 9,883 |
+| **Doctor** | + Fatigue (Marin factors, Goodman criterion) | 20,000 | 12,775 |
+**Total:** 621M tokens (custom 8,012-token BPE vocabulary).
+Every sample is **physically validated** -- the numerical solver runs to completion, and deflection, stress, and fatigue checks are verified before inclusion in the dataset.
+## Conversation Format: Harmony-Enedina
+Each sample contains a complete conversation with two semantic channels (reasoning + response) and metacognition via headers:
 ```
 <|system|>
+You are Enedina, a structural engineer...
 <|end|>
 <|user|>
+Analyze the transmission shaft with the following parameters...
 <|end|>
 <|assistant|>
 <|ch1_start|>thinking<|ch1_end|>
+[Detailed technical reasoning]
 <|ch2_start|>response<|ch2_end|>
+[Formatted response with LaTeX equations and Python solver code]
 <|end|>
 <|return|>
 ```
+Domain-specific tokens (`<|shaft|>`, `<|python|>`, `<|numerical|>`, `<|latex|>`) demarcate semantic boundaries within the output. The training language is Brazilian Portuguese.
+## Features (36 columns)
+### Identification
+| Column | Type | Description |
+|--------|------|-------------|
+| `id` | string | Unique SHA-256 hash |
+| `problem_type` | string | Always `shaft_timoshenko` |
+| `problem_class` | string | Problem class |
+| `level` | string | `bachelor`, `master`, or `doctor` |
+| `curriculum_stage` | int32 | Curriculum stage (0, 1, 2) |
+| `analysis_depth` | string | `deflection`, `stress`, or `fatigue` |
+### Shaft Parameters
+| Column | Type | Description |
+|--------|------|-------------|
+| `D_ext_mm` | float64 | External diameter [mm] |
+| `D_int_mm` | float64 | Internal diameter [mm] (0 for solid) |
+| `L_mm` | float64 | Shaft length [mm] |
+| `section_type` | string | `macico` (solid) or `vazado` (hollow) |
+| `L_D_ratio` | float64 | L/D ratio (slenderness) |
+| `material_name` | string | Material name (e.g., AISI 1045) |
+| `E_GPa` | float64 | Young's modulus [GPa] |
+| `S_y_MPa` | float64 | Yield strength [MPa] |
+| `S_ut_MPa` | float64 | Ultimate tensile strength [MPa] |
+### Loading
+| Column | Type | Description |
+|--------|------|-------------|
+| `n_loads` | int32 | Number of applied loads |
+| `n_bearings` | int32 | Number of bearings |
+| `total_load_kN` | float64 | Total load [kN] |
+### Analysis Results
+| Column | Type | Description |
+|--------|------|-------------|
+| `V_max_kN` | float64 | Maximum shear force [kN] |
+| `M_max_kNmm` | float64 | Maximum bending moment [kN.mm] |
+| `w_max_mm` | float64 | Maximum deflection [mm] |
+| `w_limit_mm` | float64 | Deflection limit [mm] |
+### Validation
+| Column | Type | Description |
+|--------|------|-------------|
+| `validation_passed` | bool | All validations passed |
+| `deflection_passed` | bool | Deflection within limit |
+| `stress_passed` | string | Stress analysis result |
+| `fatigue_passed` | string | Fatigue analysis result |
+| `difficulty_score` | float64 | Difficulty score [0-1] |
+### Conversation
+| Column | Type | Description |
+|--------|------|-------------|
+| `conversation` | string | Full conversation in Harmony format |
+| `user_tokens` | int32 | User prompt token count |
+| `assistant_tokens` | int32 | Assistant response token count |
+| `total_tokens` | int32 | Total conversation token count |
 ### Metadata
+| Column | Type | Description |
+|--------|------|-------------|
+| `expected_route` | string | Expected route (bachelor/master/doctor) |
+| `split` | string | Split: `train`, `val`, or `test` |
+| `fold` | int32 | Cross-validation fold |
+| `solve_time_ms` | float64 | Solver execution time [ms] |
+| `template_variation` | int32 | Template variation used |
+## Token Statistics
+| Metric | Value |
+|--------|-------|
+| Total tokens | 621,028,968 |
+| Mean per sample | 10,350 |
+| Minimum | 8,280 |
+| Maximum | 12,996 |
+| Median | 9,881 |
+| User tokens (mean) | 358 |
+| Assistant tokens (mean) | 9,932 |
+## Usage
 ```python
 from datasets import load_dataset
 dataset = load_dataset("aiacontext/mini-enedina-dataset")
+# Access splits
+train = dataset["train"]       # 48,000 samples
+val = dataset["validation"]    # 6,000 samples
+test = dataset["test"]         # 6,000 samples
+# Filter by level
 bachelor = train.filter(lambda x: x["level"] == "bachelor")
 doctor = train.filter(lambda x: x["level"] == "doctor")
+# Access a conversation
 sample = train[0]
 print(sample["conversation"][:500])
 print(f"Tokens: {sample['total_tokens']}")
 print(f"Level: {sample['level']}")
 ```
+## Data Generation
+The data was generated by **Factorium SciML** -- a Julia/Python framework for scientific synthetic data generation:
+1. **Numerical solver** (Python/NumPy): solves each Timoshenko problem with randomized parameters
+2. **Variable templates**: 97k description combinations in PT-BR (46^3 variations)
+3. **Numerical grounding**: a citation system that ensures consistency between input parameters and values in the response
+4. **Validation**: every sample is verified (deflection, stress, fatigue) before inclusion
+### Governing Equations (Timoshenko with self-weight)
 ```
+V(x)  = RA - q*x - sum(Fi * H(x - xi))           (Shear force)
+M(x)  = RA*x - q*x^2/2 - sum(Fi*(x-xi)*H(x-xi)) (Bending moment)
+dw/dx = theta + V/(kappa*G*A)                     (Deflection with shear)
+q     = rho * g * A * 1e-9                        (Distributed self-weight [N/mm])
 ```
+## Trained Model
+The model trained on this dataset is available at: [aiacontext/mini-enedina](https://huggingface.co/aiacontext/mini-enedina)
+| Parameter | Value |
+|-----------|-------|
+| **Architecture** | Dense Transformer (RoPE, RMSNorm, SwiGLU) |
+| **Parameters** | 37.5M |
+| **Framework** | MLX (Apple Silicon) |
+| **Test Loss** | 0.0787 |
+| **Test Perplexity** | 1.08 |
+## Citation
+If you use this dataset, please cite:
 ```bibtex
+@article{leitaofilho2026minienedina,
+  title={Mini-Enedina: A Domain-Specialized Small Language Model for Structural Shaft Analysis Using Timoshenko Beam Theory},
+  author={Leit{\~a}o Filho, Antonio de Sousa and Barros Filho, Allan Kardec Duailibe and Lima, Fabr{\'i}cio Saul and Santos, Selby Mykael Lima dos and Sousa, Rejani Bandeira Vieira},
+  year={2026}
 }
 ```
+## Acknowledgments
+This work was supported by Aia Context Ltda. and by FINEP -- Funding Authority for Studies and Projects, a Brazilian government agency for science, technology, and innovation linked to the Ministry of Science, Technology and Innovation (MCTI), under Contract No. 03.25.0080.00.
+## License
+CC-BY-4.0