Update README.md

README.md

@@ -4,15 +4,11 @@ base_model: meta-llama/Llama-2-7b-hf
 tags:
 - text-generation
 - conversational
-- assistant
-- safety
 - llama-2
-- autotrain
 - autotrain_compatible
+- function-calling
 language:
 - en
-datasets:
-- custom
 pipeline_tag: text-generation
 library_name: transformers
 model-index:
@@ -40,105 +36,84 @@ model-index:
       name: Win Rate %
   - task:
       type: text-generation
-      name: Coding
+      name: Code Generation
     dataset:
       name: HumanEval
       type: humaneval
     metrics:
-    - type: pass_at_1
+    - type: pass@1
       value: 42.3
-      name: Pass@1 Score
+      name: Pass@1
 widget:
-- text: "How do I learn Python programming?"
-  example_title: "Programming Help"
-- text: "Explain quantum computing in simple terms"
+- text: "Explain the difference between machine learning and deep learning"
   example_title: "Technical Explanation"
-- text: "Write a short story about a robot"
-  example_title: "Creative Writing"
+- text: "Write a Python function to calculate fibonacci numbers"
+  example_title: "Code Generation"
 ---
 
 <div align="center">
+
   <img src="https://imgur.com/aUIJXf7.png" alt="Helion-V1 Logo" width="100%"/>
+
 </div>
 
 ---
 
 # Helion-V1.5
 
-Helion-V1.5 is an improved conversational AI assistant fine-tuned with HuggingFace AutoTrain. Built on Llama-2-7B, it combines helpfulness, safety, and performance with enhanced training techniques.
+<div align="center">
+  <img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/powered-by-autotrain.svg" alt="Powered by AutoTrain"/>
+</div>
+
+**Helion-V1.5** is a 7B parameter conversational AI model fine-tuned from Llama-2 using QLoRA. It delivers improved performance over Helion-V1 with enhanced instruction following, code generation, and multi-turn dialogue capabilities.
 
 ## Model Details
 
-### Model Description
+**Architecture:** Llama-2-7B with LoRA adapters  
+**Parameters:** 7 billion (base) + 67M (LoRA)  
+**Context Length:** 4096 tokens  
+**Training:** QLoRA (4-bit) fine-tuning on high-quality instruction data  
+**License:** Apache 2.0
 
-- **Developed by:** DeepXR
-- **Model type:** Causal Language Model (Decoder-only Transformer)
-- **Base model:** meta-llama/Llama-2-7b-hf
-- **Language(s):** English
-- **License:** Apache 2.0
-- **Finetuned from:** Llama-2-7B using LoRA/QLoRA
-- **Training method:** HuggingFace AutoTrain
-- **Parameters:** 7 billion
-- **Context length:** 4096 tokens
+### Key Improvements over Helion-V1
 
-### Model Architecture
+| Feature | Helion-V1 | Helion-V1.5 | Improvement |
+|---------|-----------|-------------|-------------|
+| **MT-Bench Score** | 6.8 | 7.2 | +5.9% |
+| **AlpacaEval Win Rate** | 72.3% | 78.5% | +8.6% |
+| **HumanEval Pass@1** | 38.1% | 42.3% | +11.0% |
+| **Avg Response Time** | 2.3s | 1.8s | -21.7% |
+| **Function Calling** | ❌ | ✅ | New |
+| **Streaming Support** | Basic | Full | Enhanced |
 
-| Component | Specification |
-|-----------|--------------|
-| Architecture | Llama-2 (Transformer Decoder) |
-| Layers | 32 |
+### Technical Specifications
+
+| Component | Value |
+|-----------|-------|
 | Hidden Size | 4096 |
+| Layers | 32 |
 | Attention Heads | 32 |
-| Head Dimension | 128 |
 | Intermediate Size | 11008 |
-| Vocabulary Size | 32000 |
-| Position Embeddings | Rotary (RoPE) |
-| Normalization | RMSNorm |
-| Activation | SwiGLU |
+| Vocabulary | 32000 tokens |
+| Position Encoding | RoPE |
+| Precision | bfloat16 |
+
+**LoRA Configuration:**
+- Rank: 64
+- Alpha: 128  
+- Target Modules: All linear layers (q,k,v,o,gate,up,down)
+- Dropout: 0.05
 
-### Training Configuration
+## Performance Benchmarks
 
-**LoRA Parameters:**
-- Rank (r): 64
-- Alpha: 128
-- Dropout: 0.05
-- Target Modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
-
-**Training Hyperparameters:**
-- Learning Rate: 2e-5
-- Batch Size: 4 per device
-- Gradient Accumulation: 8 steps
-- Epochs: 3
-- Warmup Steps: 100
-- Max Sequence Length: 4096
-- Optimizer: AdamW
-- Scheduler: Cosine with warmup
-- Mixed Precision: bfloat16
-
-**Hardware:**
-- Training: 1x NVIDIA A100 (40GB)
-- Training Time: ~6 hours
-- Total Steps: ~5,000
-
-## Intended Use
-
-### Primary Use Cases
-
-✅ **General Conversation** - Natural, helpful dialogue  
-✅ **Question Answering** - Accurate information retrieval  
-✅ **Code Assistance** - Programming help and debugging  
-✅ **Writing Support** - Content creation and editing  
-✅ **Education** - Explanations and tutoring  
-✅ **Problem Solving** - Logical reasoning and analysis  
-
-### Out-of-Scope Use
-
-❌ **Medical Advice** - Not qualified for medical diagnosis/treatment  
-❌ **Legal Advice** - Not a substitute for legal counsel  
-❌ **Financial Advice** - Not for investment decisions  
-❌ **Harmful Content** - Will refuse to generate dangerous content  
-❌ **Impersonation** - Not for pretending to be real people  
-❌ **Misinformation** - Not for spreading false information  
+| Benchmark | Score | Category |
+|-----------|-------|----------|
+| MT-Bench | 7.2/10 | Multi-turn conversation |
+| AlpacaEval | 78.5% | Instruction following |
+| HumanEval | 42.3% | Code generation |
+| GSM8K | 35.7% | Mathematical reasoning |
+| TruthfulQA | 51.2% | Factual accuracy |
+| MMLU | 48.9% | Knowledge |  
 
 ## How to Use
 
@@ -289,51 +264,33 @@ The model may exhibit biases present in the training data. We've implemented:
 - User feedback integration
 - Ongoing bias mitigation efforts
 
-## Ethical Considerations
-
-### Responsible Use
+## Responsible Use
 
 Users should:
-- ✅ Verify important information from authoritative sources
-- ✅ Monitor outputs for accuracy in production
-- ✅ Provide proper attribution for AI-generated content
-- ✅ Implement appropriate safeguards for your use case
-- ✅ Follow applicable laws and regulations
-
-### Environmental Impact
-
-- **Training CO2 Emissions:** ~15 kg CO2eq (estimated)
-- **Training Energy:** ~30 kWh
-- **Compute Used:** 1x A100 GPU for 6 hours
+- Verify critical information from authoritative sources
+- Implement appropriate safeguards for production use
+- Monitor outputs for accuracy and appropriateness
+- Comply with applicable laws and regulations
+- Provide proper attribution for AI-generated content
 
 ## Citation
 
 ```bibtex
-@misc{helion-v1.5,
+@misc{helion-v1.5-2024,
   author = {DeepXR},
-  title = {Helion-V1.5: An Enhanced Conversational AI Assistant},
+  title = {Helion-V1.5: Enhanced Conversational AI},
   year = {2024},
   publisher = {HuggingFace},
-  howpublished = {\url{https://huggingface.co/DeepXR/Helion-V1.5}},
-  note = {Trained with HuggingFace AutoTrain}
+  url = {https://huggingface.co/DeepXR/Helion-V1.5}
 }
 ```
 
-## Model Card Authors
-
-DeepXR Team
-
-
-## Acknowledgments
+## Contact
 
-- Built on Meta's Llama-2 foundation
-- Trained using HuggingFace AutoTrain
-- Community feedback and testing
-- Open-source ecosystem support
+- **Repository:** https://huggingface.co/DeepXR/Helion-V1.5
+- **Issues:** https://huggingface.co/DeepXR/Helion-V1.5/discussions
+- **Email:** contact@deepxr.ai
 
 ---
 
-**Version:** 1.5.0  
-**Release Date:** November 2024  
-**Status:** Production Ready  
-**AutoTrain Compatible:** Yes 
+**Model Version:** 1.5.0 | **Release:** November 2024 | **Status:** Production Ready