Update README.md

README.md

@@ -1,199 +1,231 @@
 ---
 library_name: transformers
-tags: []
+license: apache-2.0
+language:
+- pt
+- en
+base_model:
+- unsloth/Qwen3-4B-Base
+pipeline_tag: text-generation
 ---
 
-# Model Card for Model ID
+# 🧠 DogeAI-v2.0-4B-Reasoning
+# 📌 Model Details
+**Model Description**
 
-<!-- Provide a quick summary of what the model is/does. -->
+DogeAI-v2.0-4B-Reasoning é um modelo de linguagem focado em raciocínio, pensamento estruturado e respostas analíticas, criado a partir do merge de uma LoRA de reasoning sobre o modelo base Qwen3-4B-Base.
 
+O objetivo principal deste modelo é melhorar a coerência lógica, a capacidade de resolver problemas em múltiplos passos e a clareza explicativa, sem alterar drasticamente o comportamento geral do modelo base.
 
+Este modelo representa a versão merged e final, podendo ser utilizado sem dependência de LoRA externa.
 
-## Model Details
+Developed by: AxionLab-Co
 
-### Model Description
+Funded by: Independent / Community-driven
 
-<!-- Provide a longer summary of what this model is. -->
+Shared by: AxionLab-Co
 
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+Model type: Decoder-only Transformer (Causal Language Model)
 
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
+Language(s) (NLP): Primarily English
 
-### Model Sources [optional]
+License: Apache 2.0 (inherits from base model)
 
-<!-- Provide the basic links for the model. -->
+Finetuned from model: Qwen3-4B-Base
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+# 🔗 Model Sources
 
-## Uses
+Repository: Hugging Face – AxionLab-Co/DogeAI-v2.0-4B-Reasoning
 
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+Base Model: Qwen/Qwen3-4B-Base
 
-### Direct Use
+Training Platform: Kaggle
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+Frameworks: PyTorch, Transformers, PEFT
 
-[More Information Needed]
+# 🎯 Uses
+# Direct Use
 
-### Downstream Use [optional]
+Este modelo pode ser utilizado diretamente para:
 
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+Raciocínio lógico e analítico
 
-[More Information Needed]
+Resolução de problemas em múltiplos passos
 
-### Out-of-Scope Use
+Explicações detalhadas (“thinking-style responses”)
 
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+Pesquisa, experimentação e aprendizado em IA
 
-[More Information Needed]
+Downstream Use
 
-## Bias, Risks, and Limitations
+Conversational agents focados em reasoning
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+Fine-tuning adicional em domínios específicos
 
-[More Information Needed]
+Conversão para GGUF e uso em engines como llama.cpp
 
-### Recommendations
+Pesquisa acadêmica ou experimental
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+Out-of-Scope Use
 
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+Este modelo não é recomendado para:
 
-## How to Get Started with the Model
+Decisões médicas, legais ou financeiras
 
-Use the code below to get started with the model.
+Aplicações críticas de segurança
 
-[More Information Needed]
+Uso onde factualidade absoluta é obrigatória
 
-## Training Details
+# ⚠️ Bias, Risks, and Limitations
 
-### Training Data
+Pode gerar cadeias de raciocínio excessivas, mesmo quando não necessárias
 
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+Herdou possíveis vieses do modelo base e dos dados de treino
 
-[More Information Needed]
+Não passou por fine-tuning específico de alinhamento ou safety
 
-### Training Procedure
+Raciocínios gerados não são garantidamente corretos
 
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+Recommendations
 
-#### Preprocessing [optional]
+Usuários devem:
 
-[More Information Needed]
+Avaliar criticamente as respostas
 
+Utilizar camadas adicionais de segurança em produção
 
-#### Training Hyperparameters
+Evitar confiar cegamente em cadeias de raciocínio
 
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+# 🚀 How to Get Started with the Model
+'' from transformers import AutoModelForCausalLM, AutoTokenizer
 
-#### Speeds, Sizes, Times [optional]
 
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+model = AutoModelForCausalLM.from_pretrained(
+    "AxionLab-Co/DogeAI-v2.0-4B-Reasoning",
+    device_map="auto",
+    torch_dtype="auto"
+)
 
-[More Information Needed]
 
-## Evaluation
+tokenizer = AutoTokenizer.from_pretrained(
+    "AxionLab-Co/DogeAI-v2.0-4B-Reasoning"
+)
 
-<!-- This section describes the evaluation protocols and provides the results. -->
 
-### Testing Data, Factors & Metrics
+inputs = tokenizer("Solve this step by step:", return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_new_tokens=256)
 
-#### Testing Data
 
-<!-- This should link to a Dataset Card if possible. -->
+print(tokenizer.decode(outputs[0], skip_special_tokens=True)) ''
+# 🏋️ Training Details
+Training Data
 
-[More Information Needed]
+O modelo foi ajustado utilizando datasets focados em reasoning e chain-of-thought, contendo:
 
-#### Factors
+Resolução passo a passo de problemas
 
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+Respostas explicativas estruturadas
 
-[More Information Needed]
+Prompts analíticos sintéticos e curados
 
-#### Metrics
+Os dados foram pré-processados manualmente para melhorar qualidade e consistência.
 
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+Training Procedure
+Preprocessing
 
-[More Information Needed]
+Tokenização com tokenizer original do Qwen
 
-### Results
+Filtragem de exemplos inconsistentes ou de baixa qualidade
 
-[More Information Needed]
+Training Hyperparameters
 
-#### Summary
+Training regime: fp16 mixed precision
 
+Fine-tuning method: LoRA (PEFT)
 
+Optimizer: AdamW
 
-## Model Examination [optional]
+Framework: Transformers + PEFT
 
-<!-- Relevant interpretability work for the model goes here -->
+Speeds, Sizes, Times
 
-[More Information Needed]
+Treinamento realizado em GPU do Kaggle
 
-## Environmental Impact
+LoRA mantida propositalmente leve
 
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+Merge final realizado via PEFT (merge_and_unload)
 
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+# 📊 Evaluation
+Testing Data, Factors & Metrics
+Testing Data
 
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
+Prompts manuais de reasoning
 
-## Technical Specifications [optional]
+Comparação direta com o modelo base
 
-### Model Architecture and Objective
+Factors
 
-[More Information Needed]
+Clareza do raciocínio
 
-### Compute Infrastructure
+Coerência lógica
 
-[More Information Needed]
+Tendência a alucinação
 
-#### Hardware
+Metrics
 
-[More Information Needed]
+Avaliação qualitativa humana
 
-#### Software
+Comparação subjetiva de respostas
 
-[More Information Needed]
+Results
 
-## Citation [optional]
+O modelo demonstra melhor organização lógica e explicações mais consistentes em comparação direta com o Qwen3-4B-Base.
 
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+Summary
 
-**BibTeX:**
+DogeAI-v2.0-4B-Reasoning prioriza qualidade de pensamento, não apenas fluência textual.
 
-[More Information Needed]
+# 🌱 Environmental Impact
 
-**APA:**
+Hardware Type: NVIDIA GPU (Kaggle)
 
-[More Information Needed]
+Hours used: Few hours (single-session fine-tuning + merge)
 
-## Glossary [optional]
+Cloud Provider: Kaggle
 
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+Compute Region: Unknown
 
-[More Information Needed]
+Carbon Emitted: Not measured
 
-## More Information [optional]
+# ⚙️ Technical Specifications
+# Model Architecture and Objective
 
-[More Information Needed]
+Decoder-only Transformer
 
-## Model Card Authors [optional]
+Objetivo: melhorar raciocínio via fine-tuning eficiente
 
-[More Information Needed]
+Compute Infrastructure
+Hardware
 
-## Model Card Contact
+NVIDIA GPU (Kaggle environment)
 
-[More Information Needed]
+Software
+
+PyTorch
+
+Transformers
+
+PEFT 0.18.1
+
+# 📚 Citation
+
+Se você utilizar este modelo em pesquisas ou projetos derivados, cite o modelo base e este repositório.
+
+# 👥 Model Card Authors
+
+AxionLab-Co
+
+# 📬 Model Card Contact
+
+Para dúvidas, feedback ou colaboração:
+AxionLab-Co – Hugging Face