Update README.md

README.md

@@ -1,199 +1,121 @@
 ---
 library_name: transformers
-tags: []
+base_model:
+- google/gemma-2-9b-it
+language:
+- ko
+pipeline_tag: text-generation
+tags:
+- mental-health
+- cbt
+- counseling
+- psychology
+- wellness
+- qlora
+license: gemma
 ---
 
-# Model Card for Model ID
+# 🧠 Doran-i (CBT Counseling Specialist)
 
-<!-- Provide a quick summary of what the model is/does. -->
+![Model Architecture](https://img.shields.io/badge/Model-Gemma_2_9B-blue) ![Task](https://img.shields.io/badge/Task-CBT_Counseling-green) ![Technique](https://img.shields.io/badge/Technique-QLoRA_Fine_Tuning-orange) ![Status](https://img.shields.io/badge/Status-Stable_Release-brightgreen)
 
-
-
-## Model Details
+## 📖 Model Details
 
 ### Model Description
 
-<!-- Provide a longer summary of what this model is. -->
-
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-
-- **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]
-
-### Model Sources [optional]
-
-<!-- Provide the basic links for the model. -->
-
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
-
-## Uses
-
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+**Doran-i(도란이)**는 구글의 고성능 오픈 모델인 `gemma-2-9b-it`를 기반으로, **CBT(인지행동치료) 기법**을 수행할 수 있도록 정교하게 미세조정(Fine-tuning)된 한국어 심리 상담 AI 모델입니다.
 
-### Direct Use
+기존 `Gemma 3`의 실험적 아키텍처 대신, 검증된 성능과 안정성을 자랑하는 **Gemma 2 9B**를 채택하여 한국어 뉘앙스 파악 능력과 상담의 깊이를 대폭 강화했습니다.
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+단순한 위로를 넘어, 내담자의 말 속에 숨겨진 **12가지 인지 왜곡(Cognitive Distortion)**을 탐지하고, **소크라테스식 질문(Socratic Questioning)**을 통해 내담자가 스스로 부정적 사고의 고리를 끊도록 돕습니다.
 
-[More Information Needed]
+- **Developed by:** Kong Yoonseo (0xMori) @ Safori
+- **Model type:** Causal Language Model (QLoRA Fine-tuned)
+- **Language(s):** Korean (한국어)
+- **License:** Gemma Terms of Use
+- **Base Model:** `google/gemma-2-9b-it`
+- **Hardware:** Trained on NVIDIA T4, Merged on TPU v5e-8
 
-### Downstream Use [optional]
+### Model Sources
 
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+- **Repository:** [https://huggingface.co/0xMori/gemma-2-9b-safori-cbt-merged](https://huggingface.co/0xMori/gemma-2-9b-safori-cbt-merged)
+- **Service Github:** [[Team Safori](https://github.com/safori-team)]
 
-[More Information Needed]
+## 🎯 Uses
 
-### Out-of-Scope Use
+### Direct Use (JSON Output)
+이 모델은 구조화된 JSON 형식으로 상담 결과를 출력하도록 훈련되었습니다:
+```json
+{
+  "emotion": "sad",
+  "empathy": "내담자의 감정에 대한 깊은 공감 멘트",
+  "detected_distortion": "흑백논리",
+  "analysis": "내담자가 왜 그런 생각을 하게 되었는지에 대한 분석",
+  "socratic_question": "반박 증거를 찾도록 유도하는 질문",
+  "alternative_thought": "긍정적 대안 사고 예시"
+}
+```
 
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+### Out-of-Scope Use (사용 제한)
+- **의료적 진단:** 이 모델은 의사가 아니며, 정신질환을 진단하거나 약물을 처방할 수 없습니다.
+- **즉각적인 위기 개입:** 자살/자해 등 응급 상황 발생 시 핫라인 안내가 필요합니다.
 
-[More Information Needed]
+## 💻 How to Get Started
 
-## Bias, Risks, and Limitations
+**Hugging Face Transformers** 라이브러리를 사용하여 바로 실행할 수 있습니다.
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+```python
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
 
-[More Information Needed]
+# Merged Model (어댑터 병합 완료)
+model_id = "0xMori/gemma-2-9b-safori-cbt-merged"
 
-### Recommendations
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    device_map="auto",
+    torch_dtype=torch.float16
+)
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+# Alpaca 포맷 (학습 시 사용한 포맷 준수 권장)
+prompt = """당신은 전문적인 인��행동치료(CBT) AI 상담사 '도란이'입니다.
+### 사용자 입력:
+사람들이 다 나를 싫어하는 것 같아서 모임에 나가기가 두려워.
+### 응답 (JSON):
+""
 
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+input_ids = tokenizer(prompt, return_tensors="pt").to("cuda")
 
-## How to Get Started with the Model
+outputs = model.generate(
+    **input_ids,
+    max_new_tokens=512,
+    do_sample=True,
+    temperature=0.7,
+    top_p=0.9
+)
 
-Use the code below to get started with the model.
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
 
-[More Information Needed]
-
-## Training Details
+## ⚙️ Training Details
 
 ### Training Data
-
-<!-- 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. -->
-
-[More Information Needed]
+- **Custom Dataset (Safori CBT):** 인지행동치료 이론을 바탕으로 자체 구축한 고품질 한국어 상담 대화셋 (약 400건).
+- **Preprocessing:**
+    - `Input(내담자 발화)` - `Output(JSON 구조)` 형태의 Alpaca 프롬프트 포맷 적용.
+    - 다양한 인지 왜곡 유형(흑백사고, 과잉일반화 등)을 골고루 분포시킴.
 
 ### Training Procedure
-
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-
-#### Preprocessing [optional]
-
-[More Information Needed]
-
-
-#### Training Hyperparameters
-
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
-
-<!-- This section describes the evaluation protocols and provides the results. -->
-
-### Testing Data, Factors & Metrics
-
-#### Testing Data
-
-<!-- This should link to a Dataset Card if possible. -->
-
-[More Information Needed]
-
-#### Factors
-
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-
-[More Information Needed]
-
-#### Metrics
-
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-
-[More Information Needed]
-
-### Results
-
-[More Information Needed]
-
-#### Summary
-
-
-
-## Model Examination [optional]
-
-<!-- Relevant interpretability work for the model goes here -->
-
-[More Information Needed]
-
-## Environmental Impact
-
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-
-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).
-
-- **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]
-
-## Technical Specifications [optional]
-
-### Model Architecture and Objective
-
-[More Information Needed]
-
-### Compute Infrastructure
-
-[More Information Needed]
-
-#### Hardware
-
-[More Information Needed]
-
-#### Software
-
-[More Information Needed]
-
-## Citation [optional]
-
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
-
-**BibTeX:**
-
-[More Information Needed]
-
-**APA:**
-
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-
-[More Information Needed]
-
-## More Information [optional]
-
-[More Information Needed]
-
-## Model Card Authors [optional]
-
-[More Information Needed]
-
-## Model Card Contact
-
-[More Information Needed]
+- **Technique:** QLoRA (Quantized Low-Rank Adaptation)
+- **Optimization Strategy:**
+    - **Early Stopping:** Training Loss 0.65 도달 시 과적합 방지를 위해 조기 종료.
+    - **Stable Environment:** 호환성 이슈가 있는 Unsloth 대신 **Pure Hugging Face (TRL 0.8.6)** 라이브러리 사용.
+
+- **Hyperparameters:**
+    - Learning Rate: 2e-4
+    - Batch Size: 1 (Gradient Accumulation: 8) -> Effective Batch Size 8
+    - Optimizer: paged_adamw_8bit
+    - Quantization: 4-bit (NF4)
+    - LoRA Rank (r): 16, Alpha: 16
+    - Max Sequence Length: 2048