Update README.md

README.md

@@ -1,183 +1,138 @@
-# Droplychee-40B
-
-<div align="center">
-
-# 🧠 Droplychee-40B
-
-**An open multilingual instruction model based on the Qwen family.**
-
-[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](#license)
-[![Status](https://img.shields.io/badge/Status-Under%20Development-orange)]()
-[![Language](https://img.shields.io/badge/Languages-Bangla%20%7C%20English-green)]()
-
-</div>
-
 ---
-
-## Overview
-
-Droplychee-40B is a multilingual large language model developed by **Droplychee Core**. The model is built on the **Qwen** family through **full fine-tuning** with the goal of improving multilingual instruction following, reasoning, coding, and Bangla language capabilities.
-
-This release is intended for research and early evaluation while the project continues to evolve.
-
+license: apache-2.0
+base_model:
+  - DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+  - text-generation
+  - chat
+  - multilingual
+  - reasoning
+  - coding
+  - qwen
+  - yarn
+  - long-context
+  - 1m-context
+  - bangla
+language:
+  - en
+  - bn
+  - multilingual
+license_name: apache-2.0
 ---
 
-## Highlights
-
-* 🌍 Multilingual (Bangla & English)
-* 🧠 Instruction Following
-* 💻 Coding Assistance
-* ✨ General Reasoning
-* 📚 Research Friendly
-* 🔓 Open Development
+# 🧠 Droplychee-2.0-40B
 
----
-
-## Model Details
+> A multilingual long-context language model developed by **Droplychee**, built upon the Qwen family and extended through full fine-tuning and architectural modifications.
 
-| Property          | Value                    |
-| ----------------- | ------------------------ |
-| Model Name        | Droplychee-40B           |
-| Base Model        | Qwen Family              |
-| Training Method   | Full Fine-Tuning         |
-| Architecture      | Decoder-only Transformer |
-| Primary Languages | Bangla, English          |
-| Status            | Under Development        |
-| Organization      | Droplychee Core          |
+## Overview
 
----
+Droplychee-2.0-40B is a decoder-only transformer model designed for multilingual reasoning, software engineering, long-context understanding, and instruction following. The model is derived from the Qwen family and further adapted through full fine-tuning using curated multilingual instruction datasets.
 
-## Intended Uses
+The project focuses on delivering strong performance in English and Bangla while maintaining support for over 40 languages.
 
-Droplychee-40B is designed for:
+## Key Features
 
-* Conversational AI
-* Coding assistance
-* Multilingual reasoning
-* Bangla NLP
-* Education
-* Research
-* AI Agents
-* Document understanding
+- 🌍 Supports 40+ languages
+- 📚 Maximum context length of 1,000,000 tokens
+- 🧵 YaRN-based context extension
+- 💻 Optimized for software engineering and coding tasks
+- 🧠 Advanced reasoning and instruction following
+- 📄 Long-document analysis and summarization
+- 🤖 Agent-oriented workflows and tool use
 
----
+## Model Details
 
-## Limitations
+| Property | Value |
+|----------|-------|
+| Model Name | Droplychee-2.0-40B |
+| Organization | Droplychee |
+| Base Model | DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking |
+| Architecture | Decoder-only Transformer |
+| Training Method | Full Fine-Tuning |
+| Context Length | 1,000,000 tokens |
+| Context Extension | YaRN |
+| Languages | 40+ |
+| License | Apache-2.0 |
 
-This model is currently under active development.
+## Architecture
 
-Users should expect:
+Droplychee-2.0-40B is derived from the Qwen architecture with additional modifications introduced during development.
 
-* Possible factual errors
-* Hallucinations
-* Incomplete reasoning
-* Benchmark results may change
-* Future releases may significantly improve quality
+Key architectural characteristics include:
 
----
+- Decoder-only transformer architecture
+- Rotary Position Embeddings (RoPE)
+- YaRN-based long-context extension
+- Optimized KV-cache handling
+- Flash Attention support (where available in the serving stack)
+- Full supervised instruction fine-tuning
 
 ## Training
 
-Droplychee-40B was created through **full fine-tuning** of a model from the **Qwen family**.
-
-Additional training details will be documented in future releases.
-
-**TBD**
-
-* Dataset composition
-* Training tokens
-* Optimizer
-* Learning rate
-* Batch size
-* Epochs
-* Sequence length
+The model was fine-tuned using a curated multilingual instruction corpus emphasizing:
 
----
+- General reasoning
+- Coding and software engineering
+- Mathematics
+- Multilingual dialogue
+- Long-context comprehension
+- Agent-oriented tasks
 
-## Evaluation
+Approximate training statistics:
 
-Comprehensive benchmark results are not yet available.
+| Item | Value |
+|------|-------|
+| High-quality instruction pairs | ~500K |
+| Training tokens | ~100M–1B |
+| Training Platform | Unsloth Studio |
+| GPU Hardware | 3× NVIDIA RTX PRO 6000 Blackwell Server Edition |
 
-Future evaluations are planned for:
+## Context Extension
 
-* MMLU
-* GSM8K
-* HumanEval
-* MBPP
-* ARC
-* IFEval
-* BBH
-* Bangla-specific benchmarks
+Droplychee-2.0-40B supports a maximum context length of **1,000,000 tokens** through **YaRN (Yet another RoPE extensioN)**.
 
----
+YaRN extends the effective context window while preserving compatibility with Rotary Position Embeddings. Long-context performance may vary depending on the inference engine, hardware resources, and serving configuration.
 
-## Usage (Transformers)
+## Supported Languages
 
-```python
-from transformers import AutoTokenizer, AutoModelForCausalLM
+The model is optimized for multilingual instruction following and has been evaluated primarily on English and Bangla. It also supports more than 40 languages, including Hindi, Arabic, Chinese, Japanese, Korean, French, German, Spanish, Portuguese, and Russian.
 
-model_id = "droplychee/droplychee-40b"
+## Intended Use
 
-tokenizer = AutoTokenizer.from_pretrained(model_id)
+Recommended applications include:
 
-model = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    torch_dtype="auto",
-    device_map="auto"
-)
+- Conversational AI
+- Coding assistants
+- Research assistance
+- Educational tools
+- Long-document summarization
+- Retrieval-Augmented Generation (RAG)
+- AI agents
+- Software engineering workflows
 
-messages = [
-    {"role": "user", "content": "Explain recursion with a Python example."}
-]
+## Limitations
 
-text = tokenizer.apply_chat_template(
-    messages,
-    tokenize=False,
-    add_generation_prompt=True
-)
+- Outputs may contain factual inaccuracies or hallucinations.
+- Performance varies across languages and domains.
+- Independent third-party evaluation has not yet been completed.
+- Users should verify outputs before relying on them for high-impact decisions.
 
-inputs = tokenizer(text, return_tensors="pt").to(model.device)
+## Responsible Use
 
-outputs = model.generate(
-    **inputs,
-    max_new_tokens=512
-)
+This model is intended for research and general-purpose AI applications. Users are responsible for ensuring compliance with applicable laws, regulations, and ethical guidelines.
 
-print(tokenizer.decode(outputs[0], skip_special_tokens=True))
-```
+## Attribution
 
----
+This project is derived from:
 
-## Citation
+**DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking**
 
-```bibtex
-@misc{droplychee40b2026,
-  title={Droplychee-40B},
-  author={Droplychee Core},
-  year={2026},
-  publisher={Hugging Face},
-  howpublished={\url{https://huggingface.co/droplychee}}
-}
-```
-
----
+Please refer to the original model card and license for upstream attribution requirements.
 
 ## License
 
-**TBD**
-
-(Choose the appropriate license before the stable release.)
-
----
-
-## Acknowledgements
-
-Droplychee-40B builds upon the open-source AI ecosystem. We acknowledge the contributions of the Qwen team and the broader machine learning community.
-
----
-
-## Contact
+This model is distributed under the **Apache License 2.0**.
 
-* Organization: **Droplychee Core**
-* GitHub: https://github.com/DropLychee
-* Hugging Face: https://huggingface.co/droplychee
+Users must also comply with the license terms of the upstream base model where applicable.