Merge pull request #22 from Sunwood-ai-labs/translate-readme-12026839076

docs/README.en.md

@@ -31,7 +31,7 @@ license: mit
 </p>
 
 <h2 align="center">
-  Llama Model Fine-tuning Experimentation Environment
+  ～ Llama Model Fine-tuning Experiment Environment ～
 </h2>
 
 <p align="center">
@@ -44,86 +44,93 @@ license: mit
 
 ## 🚀 Project Overview
 
-**Llama-finetune-sandbox** provides an experimental environment for learning and verifying the fine-tuning of Llama models.  You can try various fine-tuning methods, customize models, and evaluate performance.  It caters to a wide range of users, from beginners to researchers. Version 0.5.0 includes updated documentation and the addition of a context-aware reflexive QA generation system. This system generates high-quality Q&A datasets from Wikipedia data, leveraging LLMs to iteratively improve the quality of questions and answers, resulting in a more accurate dataset.
+**Llama-finetune-sandbox** provides an experimental environment for learning and validating fine-tuning of Llama models.  You can try various fine-tuning methods, customize models, and evaluate their performance.  It caters to a wide range of users, from beginners to researchers. Version 0.6.0 includes updated documentation and the implementation of an LLM evaluation system. This system automatically evaluates the quality of LLM responses and generates detailed evaluation reports.
 
 
-## ✨ Key Features
+## ✨ Main Features
 
-1. **Diverse Fine-tuning Methods:**
+1. **Diverse Fine-tuning Methods**:
    - LoRA (Low-Rank Adaptation)
    - QLoRA (Quantized LoRA)
 
-2. **Flexible Model Configuration:**
+2. **Flexible Model Settings**:
    - Customizable maximum sequence length
    - Various quantization options
    - Multiple attention mechanisms
 
-3. **Well-equipped Experimentation Environment:**
-   - Optimized memory usage
+3. **Experiment Environment Setup**:
+   - Memory usage optimization
    - Visualization of experimental results
 
-4. **Context-Aware Reflexive QA Generation System:**
+4. **Context-Aware Reflexive QA Generation System**:
     - Generates high-quality Q&A datasets from Wikipedia data.
-    - Uses LLMs to automatically generate context-aware questions and answers, evaluate quality, and iteratively improve them.
-    - Employs a reflexive approach, quantifying factuality, question quality, and answer completeness for iterative improvement.
+    - Utilizes LLMs to generate context-aware questions and answers, automatically evaluate quality, and iteratively improve them.
+    - Employs a reflexive approach, quantifying factuality, question quality, and answer completeness to enable incremental improvements.
     - Provides comprehensive code and explanations covering environment setup, model selection, data preprocessing, Q&A pair generation, quality evaluation, and the improvement process.
-    - Utilizes libraries such as `litellm`, `wikipedia`, and `transformers`.
+    - Uses libraries such as `litellm`, `wikipedia`, and `transformers`.
     - Generated Q&A pairs are saved in JSON format and can be easily uploaded to the Hugging Face Hub.
 
+5. **LLM Evaluation System**:
+    - Automatically evaluates the quality of LLM responses.
+    - Evaluates questions, model answers, and LLM responses on a 4-point scale, generating detailed evaluation reports.
+    - Features error handling, retry functionality, logging, customizable evaluation criteria, and report generation in CSV and HTML formats.
+    - Also includes functionality for uploading to the Hugging Face Hub.
 
-## 📚 Examples
 
-This repository includes the following examples:
+## 🔧 Usage
 
-### Fast Fine-tuning using Unsloth
- - Implementation of fast fine-tuning for Llama-3.2-1B/3B models.
+Refer to the notebooks in this repository.
+
+
+## 📦 Installation Instructions
+
+Refer to `requirements.txt` and install the necessary packages.
+
+
+## 📚 Implementation Examples
+
+This repository includes the following implementation examples:
+
+### High-Speed Fine-tuning using Unsloth
+ - High-speed fine-tuning implementation for Llama-3.2-1B/3B models
    - → See [`Llama_3_2_1B+3B_Conversational_+_2x_faster_finetuning_JP.md`](sandbox/Llama_3_2_1B+3B_Conversational_+_2x_faster_finetuning_JP.md) for details.
    - → [Use this to convert from markdown to notebook format](https://huggingface.co/spaces/MakiAi/JupytextWebUI)
  - [📒Notebook here](https://colab.research.google.com/drive/1AjtWF2vOEwzIoCMmlQfSTYCVgy4Y78Wi?usp=sharing)
 
-### Efficient Model Deployment using Ollama and LiteLLM
- - Setup and deployment guide on Google Colab.
+### Efficient Model Operation using Ollama and LiteLLM
+ - Setup and operation guide for Google Colab
  - → See [`efficient-ollama-colab-setup-with-litellm-guide.md`](sandbox/efficient-ollama-colab-setup-with-litellm-guide.md) for details.
  - [📒Notebook here](https://colab.research.google.com/drive/1buTPds1Go1NbZOLlpG94VG22GyK-F4GW?usp=sharing)
 
 ### Q&A Dataset Generation from Wikipedia Data (Sentence Pool QA Method)
-- High-quality Q&A dataset generation using the sentence pool QA method.
-  - → A new dataset creation method that generates Q&A pairs while preserving context by pooling sentence segments delimited by punctuation.
-  - → Chunk size is flexibly adjustable (default 200 characters) to generate Q&A pairs with an optimal context range depending on the application.
+- High-quality Q&A dataset generation using the sentence pool QA method
+  - → A new dataset creation method that retains context by pooling sentences separated by punctuation marks.
+  - → Chunk size is flexibly adjustable (default 200 characters) to generate Q&A pairs with optimal context ranges for various applications.
   - → See [`wikipedia-qa-dataset-generator.md`](sandbox/wikipedia-qa-dataset-generator.md) for details.
 - [📒Notebook here](https://colab.research.google.com/drive/1mmK5vxUzjk3lI6OnEPrQqyjSzqsEoXpk?usp=sharing)
 
 ### Context-Aware Reflexive QA Generation System
-- Q&A dataset generation with reflexive quality improvement.
-  - → Automatically evaluates the quality of generated Q&A pairs and iteratively improves them.
+- Q&A dataset generation with reflexive quality improvement
+  - → A new method that automatically evaluates the quality of generated Q&A pairs and iteratively improves them.
   - → Quantifies factuality, question quality, and answer completeness for evaluation.
-  - → Generates high-precision questions and performs consistency checks on answers using contextual information.
+  - → Uses contextual information for high-precision question generation and answer consistency checks.
   - → See [`context_aware_Reflexive_qa_generator_V2.md`](sandbox/context_aware_Reflexive_qa_generator_V2.md) for details.
 - [📒Notebook here](https://colab.research.google.com/drive/1OYdgAuXHbl-0LUJgkLl_VqknaAEmAm0S?usp=sharing)
 
 ### LLM Evaluation System (LLMs as a Judge)
-- Advanced quality evaluation system using LLMs as evaluators.
-  - → Automatically evaluates questions, model answers, and LLM responses on a four-level scale.
-  - → Robust design with error handling and retry functions.
+- Advanced quality evaluation system utilizing LLMs as evaluators
+  - → Automatically evaluates questions, model answers, and LLM responses on a 4-point scale.
+  - → Robust design with error handling and retry functionality.
   - → Generates detailed evaluation reports in CSV and HTML formats.
   - → See [`LLMs_as_a_Judge_TOHO_V2.md`](sandbox/LLMs_as_a_Judge_TOHO_V2.md) for details.
 - [📒Notebook here](https://colab.research.google.com/drive/1Zjw3sOMa2v5RFD8dFfxMZ4NDGFoQOL7s?usp=sharing)
 
 
-## 🛠️ Setup
+## 🆕 What's New (v0.6.0)
 
-1. Clone the repository:
-```bash
-git clone https://github.com/Sunwood-ai-labs/Llama-finetune-sandbox.git
-cd Llama-finetune-sandbox
-```
+- **Implementation of the LLM Evaluation System**: Added a system to automatically evaluate the quality of LLM responses. Questions, model answers, and LLM answers are compared and evaluated on a 4-point scale.  Features error handling, retry functionality, logging, customizable evaluation criteria, and report generation in CSV and HTML formats.
+- Added information about the LLM evaluation system to README.md
 
-## 📝 Adding Examples
-
-1. Add new implementations to the `sandbox/` directory.
-2. Add necessary settings and utilities to `utils/` (This section was removed as `utils/` directory appears not to exist).
-3. Update documentation and tests (This section was removed as there's no mention of existing tests).
-4. Create a pull request.
 
 ## 🤝 Contributions
 
@@ -132,19 +139,6 @@ cd Llama-finetune-sandbox
 - Documentation improvements
 - Addition of usage examples
 
-## 📚 References
-
-- [HuggingFace PEFT documentation](https://huggingface.co/docs/peft)
-- [About Llama models](https://github.com/facebookresearch/llama)
-- [Fine-tuning best practices](https://github.com/Sunwood-ai-labs/Llama-finetune-sandbox/wiki) (This section was removed as the wiki page appears not to exist).
-
 ## 📄 License
 
-This project is licensed under the MIT License.
-
-## v0.5.0 Updates
-
-**🆕 What's New:**
-
-- Implementation of the context-aware reflexive QA generation system.
-- Addition of relevant information to README.md.
+This project is licensed under the MIT License.