Upload folder using huggingface_hub

COMPANY.md

@@ -0,0 +1,29 @@
+# RaxCore
+
+**A leading developer company in Africa and beyond**
+
+🌐 **Website**: [www.raxcore.dev](https://www.raxcore.dev/)  
+🤗 **Hugging Face**: [raxcore-dev](https://huggingface.co/raxcore-dev)
+
+RaxCore is at the forefront of AI and software development, creating innovative solutions that bridge technology gaps across Africa and the global market. 
+
+## About RaxCore
+
+RaxCore specializes in:
+- Advanced AI model development and fine-tuning
+- Conversational AI systems
+- Custom software solutions
+- Technology consulting and implementation
+
+## Our Mission
+
+To democratize access to cutting-edge AI technology while fostering innovation across Africa and beyond.
+
+## Rax 3.5 Chat
+
+Rax 3.5 Chat represents RaxCore's commitment to developing high-quality, accessible AI models that serve diverse communities and use cases.
+
+---
+
+**Contact RaxCore**  
+Visit [www.raxcore.dev](https://www.raxcore.dev/) for enterprise solutions, custom model development, and AI consulting services.

DEPLOYMENT.md

@@ -0,0 +1,78 @@
+# Rax 3.5 Chat - Deployment Guide
+
+## Uploading to Hugging Face
+
+### Prerequisites
+1. Install required packages:
+```bash
+pip install huggingface_hub transformers
+```
+
+2. Login to Hugging Face:
+```bash
+huggingface-cli login
+```
+
+### Upload Steps
+
+1. **Initialize Git LFS** (if not already done):
+```bash
+cd /home/ogega/Projects/models/rax-3.5-chat
+git lfs install
+```
+
+2. **Add all files**:
+```bash
+git add .
+git commit -m "Initial commit: Rax 3.5 Chat model"
+```
+
+3. **Create repository on Hugging Face**:
+   - Go to https://huggingface.co/new
+   - Create a new model repository named "rax-3.5-chat" under raxcore-dev
+   - Choose "Public" or "Private" as needed
+
+4. **Push to Hugging Face**:
+```bash
+git remote add origin https://huggingface.co/raxcore-dev/rax-3.5-chat
+git branch -M main
+git push -u origin main
+```
+
+### Alternative: Using huggingface_hub
+
+```python
+from huggingface_hub import HfApi
+
+api = HfApi()
+api.upload_folder(
+    folder_path="/home/ogega/Projects/models/rax-3.5-chat",
+    repo_id="raxcore-dev/rax-3.5-chat",
+    repo_type="model"
+)
+```
+
+## Model Testing
+
+Run the included test script:
+```bash
+cd /home/ogega/Projects/models/rax-3.5-chat
+python test_rax.py
+```
+
+## Files Included
+
+- `config.json` - Model configuration
+- `tokenizer_config.json` - Tokenizer configuration  
+- `model.safetensors` - Model weights
+- `tokenizer.json` - Tokenizer data
+- `tokenizer.model` - SentencePiece model
+- `generation_config.json` - Generation parameters
+- `README.md` - Comprehensive documentation
+- `model_card.md` - Hugging Face model card
+- `test_rax.py` - Test script
+- `.gitattributes` - Git LFS configuration
+
+## Ready for Release!
+
+Your Rax 3.5 Chat model is now fully rebranded and ready for upload to Hugging Face.

README.md

@@ -0,0 +1,140 @@
+# Rax 3.5 Chat
+
+**Developed by RaxCore - A leading developer company in Africa and beyond**
+
+Rax 3.5 Chat is a fine-tuned conversational AI model based on the Llama architecture. This model has been specifically optimized for chat interactions and dialogue generation.
+
+## Model Details
+
+- **Model Name**: Rax 3.5 Chat
+- **Architecture**: Llama (LlamaForCausalLM)
+- **Parameters**: ~1.1B
+- **Context Length**: 2048 tokens
+- **Precision**: bfloat16
+- **License**: Apache 2.0
+
+## Model Architecture
+
+- **Hidden Size**: 2048
+- **Intermediate Size**: 5632
+- **Attention Heads**: 32
+- **Key-Value Heads**: 4
+- **Hidden Layers**: 22
+- **Vocabulary Size**: 32,000
+
+## Usage
+
+### Quick Start
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+# Load model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained("rax-3.5-chat")
+model = AutoModelForCausalLM.from_pretrained(
+    "rax-3.5-chat",
+    torch_dtype=torch.bfloat16,
+    device_map="auto"
+)
+
+# Chat template
+messages = [
+    {"role": "system", "content": "You are Rax, a helpful AI assistant."},
+    {"role": "user", "content": "Hello! How are you?"}
+]
+
+# Apply chat template
+input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tokenizer(input_text, return_tensors="pt")
+
+# Generate response
+with torch.no_grad():
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=256,
+        temperature=0.7,
+        do_sample=True,
+        pad_token_id=tokenizer.eos_token_id
+    )
+
+response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
+print(response)
+```
+
+### Chat Format
+
+Rax 3.5 Chat uses the following conversation format:
+
+```
+<|system|>
+You are Rax, a helpful AI assistant.</s>
+<|user|>
+Hello! How are you?</s>
+<|assistant|>
+Hello! I'm doing well, thank you for asking. How can I help you today?</s>
+```
+
+## Training Details
+
+This model was fine-tuned from TinyLlama with:
+- Extended training over several days
+- Optimized for conversational interactions
+- Enhanced dialogue coherence and helpfulness
+
+## Intended Use
+
+Rax 3.5 Chat is designed for:
+- Conversational AI applications
+- Chatbots and virtual assistants
+- Educational and research purposes
+- Creative writing assistance
+
+## Limitations
+
+- Context window limited to 2048 tokens
+- May generate incorrect or biased information
+- Not suitable for production use without proper safety measures
+- Requires responsible deployment practices
+
+## Ethical Considerations
+
+Please use this model responsibly:
+- Implement appropriate content filtering
+- Monitor outputs for potential biases
+- Ensure compliance with applicable regulations
+- Consider the impact on users and society
+
+## Technical Specifications
+
+- **Framework**: Transformers 4.35.0+
+- **Hardware Requirements**: GPU with 4GB+ VRAM recommended
+- **Inference Speed**: Optimized for real-time chat applications
+
+## Citation
+
+If you use Rax 3.5 Chat in your research or applications, please cite:
+
+```bibtex
+@misc{rax35chat2024,
+  title={Rax 3.5 Chat: A Fine-tuned Conversational AI Model},
+  author={RaxCore},
+  year={2024},
+  note={Fine-tuned from TinyLlama architecture},
+  organization={RaxCore - Leading developer company in Africa and beyond}
+}
+```
+
+## Contact
+
+For questions, issues, or collaboration opportunities:
+- **Hugging Face**: https://huggingface.co/raxcore-dev
+- **Website**: https://www.raxcore.dev/
+- **Model Repository**: Contact RaxCore directly
+
+---
+
+**RaxCore** - A leading developer company in Africa and beyond  
+🌐 **Website**: [www.raxcore.dev](https://www.raxcore.dev/)  
+🤗 **Hugging Face**: [raxcore-dev](https://huggingface.co/raxcore-dev)  
+*Rax 3.5 Chat - Powering the next generation of conversational AI*

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "rax-3.5-chat",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 5632,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 22,
+  "num_key_value_heads": 4,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.35.0",
+  "use_cache": true,
+  "vocab_size": 32000
+}

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 3.0,
+    "eval_logits/chosen": -2.707406759262085,
+    "eval_logits/rejected": -2.656524419784546,
+    "eval_logps/chosen": -370.1297607421875,
+    "eval_logps/rejected": -296.0738525390625,
+    "eval_loss": 0.513750433921814,
+    "eval_rewards/accuracies": 0.738095223903656,
+    "eval_rewards/chosen": -0.02744222804903984,
+    "eval_rewards/margins": 1.0087225437164307,
+    "eval_rewards/rejected": -1.03616464138031,
+    "eval_runtime": 93.5908,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 21.37,
+    "eval_steps_per_second": 0.673
+}

generation_config.json

@@ -0,0 +1,8 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "max_length": 2048,
+  "pad_token_id": 2,
+  "transformers_version": "4.35.0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e6001da2106d4757498752a021df6c2bdc332c650aae4bae6b0c004dcf14933
+size 2200119864

model_card.md

@@ -0,0 +1,65 @@
+---
+license: apache-2.0
+language:
+- en
+pipeline_tag: text-generation
+tags:
+- chat
+- conversational
+- llama
+- fine-tuned
+- rax
+- raxcore
+model_type: llama
+---
+
+# Rax 3.5 Chat
+
+**Developed by RaxCore - A leading developer company in Africa and beyond**
+
+## Model Description
+
+Rax 3.5 Chat is a fine-tuned conversational AI model based on the Llama architecture, specifically optimized for chat interactions and dialogue generation. This model represents several days of careful fine-tuning to enhance conversational capabilities.
+
+## Quick Start
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("rax-3.5-chat")
+model = AutoModelForCausalLM.from_pretrained("rax-3.5-chat")
+
+messages = [
+    {"role": "system", "content": "You are Rax, a helpful AI assistant."},
+    {"role": "user", "content": "Hello!"}
+]
+
+input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+inputs = tokenizer(input_text, return_tensors="pt")
+outputs = model.generate(**inputs, max_new_tokens=256)
+```
+
+## Model Details
+
+- **Architecture**: Llama (1.1B parameters)
+- **Context Length**: 2048 tokens
+- **Training**: Fine-tuned for conversational AI
+- **License**: Apache 2.0
+
+## Intended Use
+
+- Conversational AI applications
+- Research and educational purposes
+- Creative writing assistance
+- Chatbot development
+
+## Limitations
+
+- 2048 token context limit
+- May generate biased or incorrect information
+- Requires responsible deployment practices
+
+## Links
+
+- **RaxCore Website**: [www.raxcore.dev](https://www.raxcore.dev/)
+- **Hugging Face Profile**: [raxcore-dev](https://huggingface.co/raxcore-dev)

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

test_rax.py

@@ -0,0 +1,48 @@
+#!/usr/bin/env python3
+"""
+Test script for Rax 3.5 Chat model
+"""
+
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+def test_rax_chat():
+    print("Loading Rax 3.5 Chat model...")
+    
+    # Load model and tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(".")
+    model = AutoModelForCausalLM.from_pretrained(
+        ".",
+        torch_dtype=torch.bfloat16,
+        device_map="auto"
+    )
+    
+    print("Model loaded successfully!")
+    
+    # Test conversation
+    messages = [
+        {"role": "system", "content": "You are Rax, a helpful AI assistant."},
+        {"role": "user", "content": "Hello! Can you tell me about yourself?"}
+    ]
+    
+    # Apply chat template
+    input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    print(f"Input: {input_text}")
+    
+    inputs = tokenizer(input_text, return_tensors="pt")
+    
+    # Generate response
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=128,
+            temperature=0.7,
+            do_sample=True,
+            pad_token_id=tokenizer.eos_token_id
+        )
+    
+    response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
+    print(f"Rax: {response}")
+
+if __name__ == "__main__":
+    test_rax_chat()

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -0,0 +1,41 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n'  + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 2048,
+  "name_or_path": "rax-3.5-chat",
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}