Update README.md

README.md

@@ -1,3 +1,66 @@
----
-license: mit
----
+---
+license: mit
+datasets:
+- Open-Orca/OpenOrca
+- microsoft/orca-math-word-problems-200k
+- meta-math/MetaMathQA
+language:
+- en
+tags:
+- turbo
+- conversational
+- chicka
+---
+# TurboLM by Chickaboo AI
+
+Welcome to TurboLM, the state-of-the-art language model developed by Chickaboo AI. TurboLM is designed to deliver a high-speed, low computing, and high-quality reasoning conversational experience.
+
+## Table of Contents
+- **Technical Details**
+- **Training Details**
+- **Benchmarks**
+- **Usage**
+- **License**
+
+## Model Details
+TurboLM utilizes a transformer-based architecture with the state of the art [Xenova/gpt-4o](https://huggingface.co/Xenova/gpt-4o) Tokenizer. The model has 150M parameters, making it  high-speed and extremely efficient. This efficiency allows it to run on low-end devices while still delivering industry-best performance.
+
+## Training Details
+TurboLM was trained on these datasets with the presetage of the model they make up to the side datasets: [Open-Orca/OpenOrca](https://huggingface.co/datasets/Open-Orca/OpenOrca) 75%, [meta-math/MetaMathQA](https://huggingface.co/datasets/meta-math/MetaMathQA?row=4) 15%, [microsoft/orca-math-word-problems-200k](microsoft/orca-math-word-problems-200k) 10% using this [Training Script]() in [Google Cloud](https://cloud.google.com/) with a T4 GPU for 2 days.
+
+## OpenLLM Learderboards
+
+| **Benchmark**    | **TurboLM**  | **Mistral-7B-Instruct-v0.2** | **Meta-Llama-3-8B** |
+|--------------|----------------------|--------------------------|-----------------|
+| **Average**      | **69.19**                |  60.97                   | 62.55           |
+| **ARC**          | **64.08**                |  59.98                   | 59.47           |
+| **Hellaswag**    | **83.96**                |  83.31                   | 82.09           |
+| **MMLU**         | 64.87                |  64.16                   | **66.67**           |
+| **TruthfulQA**   | **50.51**                |  42.15                   | 43.95           |
+| **Winogrande**   | **81.06**                |  78.37                   | 77.35           |
+| **GSM8K**        | **70.66**                |  37.83                   | 45.79           |
+
+## Usage
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+device = "cuda" # the device to load the model onto
+
+model = AutoModelForCausalLM.from_pretrained("Chickaboo/TurboLM")
+tokenizer = AutoTokenizer.from_pretrained("Chickaboo/TurboLM")
+
+messages = [
+    {"role": "user", "content": "What is your favourite condiment?"},
+    {"role": "assistant", "content": "Well, I'm quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever I'm cooking up in the kitchen!"},
+    {"role": "user", "content": "Do you have mayonnaise recipes?"}
+]
+
+encodeds = tokenizer.apply_chat_template(messages, return_tensors="pt")
+
+model_inputs = encodeds.to(device)
+model.to(device)
+
+generated_ids = model.generate(model_inputs, max_new_tokens=1000, do_sample=True)
+decoded = tokenizer.batch_decode(generated_ids)
+print(decoded[0])