Update README.md

README.md

@@ -1,22 +1,49 @@
 ---
+license: apache-2.0
 base_model:
 - mistralai/Mistral-7B-v0.1
 - HuggingFaceH4/zephyr-7b-beta
 tags:
 - merge
 - mergekit
-- lazymergekit
-- mistralai/Mistral-7B-v0.1
-- HuggingFaceH4/zephyr-7b-beta
+- mistral
+- zephyr
+- slerp
+- cosmetic
+- business
 ---
 
-# CosmeticVenture
+# CosmetiGuide-7B
+
+This is a specialized model merge created using MergeKit, combining the French language expertise of Mistral-7B with the business acumen of Zephyr-7B through an advanced SLERP fusion technique with variable interpolation values, optimized for the cosmetic industry.
+
+## About Me
+
+I'm David Soeiro-Vuong, a third-year Computer Science student at STS1 working on specialized language models. Passionate about artificial intelligence and its business applications, I focus on creating domain-specific model merges that enhance performance for targeted industry use cases.
+
+🔗 [Connect with me on LinkedIn](https://www.linkedin.com/in/david-soeiro-vuong-a28b582ba/)
+
+## Project Overview
+
+CosmetiGuide-7B is designed as a specialized LLM for the cosmetic industry, capable of providing expert business guidance for entrepreneurs and professionals in this sector. The model combines Mistral's excellent French language capabilities with Zephyr's business understanding to create a powerful advisor for cosmetic business development.
+
+## Merge Details
+
+### Merge Method
 
-CosmeticVenture is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
-* [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
-* [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta)
+This model uses SLERP (Spherical Linear Interpolation) with carefully calibrated parameters:
 
-## 🧩 Configuration
+- **Attention Layers**: Variable interpolation values [0, 0.5, 0.3, 0.7, 1] leveraging Zephyr's instruction-following and business capabilities
+- **MLP Layers**: Variable interpolation values [1, 0.5, 0.7, 0.3, 0] maintaining Mistral's French language expertise and reasoning
+- **Other Parameters**: 0.5 interpolation value creating a balanced fusion
+- **Format**: bfloat16 precision for efficient memory usage
+
+### Models Merged
+
+* [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) - A powerful French-origin base model with excellent multilingual capabilities
+* [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) - An instruction-tuned model with strong business reasoning capabilities
+
+### Configuration
 
 ```yaml
 slices:
@@ -35,29 +62,43 @@ parameters:
       value: [1, 0.5, 0.7, 0.3, 0]
     - value: 0.5
 dtype: bfloat16
+MODEL_NAME: "CosmetiGuide-7B"
 ```
 
-## 💻 Usage
+## Model Capabilities
+
+This specialized merge combines:
+
+- Mistral's excellent French language understanding and multilingual capabilities
+- Zephyr's instruction-following and business reasoning abilities
+- Domain adaptation for the cosmetic industry through strategic parameter fusion
+
+The resulting model is optimized for tasks in the cosmetic business sector, such as:
+
+- Regulatory guidance for cosmetic products in various markets
+- Business planning and market analysis for cosmetic startups
+- Ingredient formulation and technical documentation assistance
+- Marketing strategy and brand positioning advice
+- Trend analysis and innovation forecasting for beauty products
+
+## Future Development
+
+This interim model will be further enhanced once access to Meta-Llama-3-8B is granted, providing even stronger business reasoning capabilities. The model will also be fine-tuned on a specialized dataset consisting of:
 
-```python
-!pip install -qU transformers accelerate
+- Regulatory documentation (EU regulations, BPF, ANSM guidelines)
+- Market research and industry analyses (FEBEA reports, segment studies)
+- Technical documentation (ingredient specifications, manufacturing processes)
+- Business resources (business plans, pricing strategies, case studies)
+- Industry trends (clean beauty, solid cosmetics, circular economy)
+- Marketing strategies (brand positioning, digital strategies)
 
-from transformers import AutoTokenizer
-import transformers
-import torch
+## Limitations
 
-model = "Davidsv/CosmeticVenture"
-messages = [{"role": "user", "content": "What is a large language model?"}]
+- Limited domain-specific training beyond parameter merging
+- May require additional fine-tuning for highly specialized cosmetic industry tasks
+- General limitations inherited from the base 7B parameter models
+- Pending enhancement with Meta-Llama-3-8B once access is granted
 
-tokenizer = AutoTokenizer.from_pretrained(model)
-prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-pipeline = transformers.pipeline(
-    "text-generation",
-    model=model,
-    torch_dtype=torch.float16,
-    device_map="auto",
-)
+## License
 
-outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
-print(outputs[0]["generated_text"])
-```
+This model is released under the Apache 2.0 license, consistent with the underlying models' licenses.