README.md

---
base_model: Spestly/Athena-1-0.5B
tags:
- text-generation-inference
- transformers
- unsloth
- qwen2
- trl
- llama-cpp
- gguf-my-repo
license: apache-2.0
language:
- en
---

# Triangle104/Athena-1-0.5B-Q5_K_M-GGUF
This model was converted to GGUF format from [`Spestly/Athena-1-0.5B`](https://huggingface.co/Spestly/Athena-1-0.5B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
Refer to the [original model card](https://huggingface.co/Spestly/Athena-1-0.5B) for more details on the model.

---
Model details:
-
Athena-1 0.5B is a fine-tuned, instruction-following large language model derived from Qwen/Qwen2.5-0.5B-Instruct.
 Designed for ultra-lightweight applications, Athena-1 0.5B balances 
compactness with robust performance, making it suitable for tasks with 
limited computational resources.




	
		
	

		Key Features
	




	
		
	

		⚡ Ultra-Lightweight and Efficient
	



Compact Size: With just 500 million parameters, Athena-1 0.5B is ideal for edge devices and low-resource environments.
Instruction Following: Fine-tuned for reliable adherence to user instructions.
Coding and Mathematics: Capable of handling basic coding and mathematical tasks.



	
		
	

		📖 Contextual Understanding
	



Context Length: Supports up to 16,384 tokens, enabling processing of moderately sized conversations or documents.
Token Generation: Can generate up to 4K tokens of coherent output.



	
		
	

		🌍 Multilingual Support
	



Supports 20+ languages, including:
English, Chinese, French, Spanish, German, Italian, Russian
Japanese, Korean, Vietnamese, Thai, and more.





	
		
	

		📊 Structured Data & Outputs
	



Structured Data Interpretation: Handles formats like tables and JSON effectively.
Structured Output Generation: Produces well-formatted outputs for data-specific tasks.




	
		
	

		Model Details
	



Base Model: Qwen/Qwen2.5-0.5B-Instruct
Architecture: Transformers with RoPE, SwiGLU, RMSNorm, Attention QKV bias, and tied word embeddings.
Parameters: 500M total.
Layers: (Adjust if different from the base model)
Attention Heads: (Adjust if different from the base model)
Context Length: Up to 16,384 tokens.




	
		
	

		Applications
	



Athena-1 0.5B is optimized for:


Conversational AI: Power lightweight and responsive chatbots.
Code Assistance: Basic code generation, debugging, and explanations.
Mathematical Assistance: Solves fundamental math problems.
Document Processing: Summarizes and analyzes smaller documents effectively.
Multilingual Tasks: Supports global use cases with a compact model.
Structured Data: Reads and generates structured formats like JSON and tables.




	
		
	

		Quickstart
	



Here’s how you can use Athena-1 0.5B for quick text generation:


# Use a pipeline as a high-level helper
from transformers import pipeline

messages = [
    {"role": "user", "content": "What can you do?"},
]
pipe = pipeline("text-generation", model="Spestly/Athena-1-0.5B") # Update model name
print(pipe(messages))

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Spestly/Athena-1-0.5B") # Update model name
model = AutoModelForCausalLM.from_pretrained("Spestly/Athena-1-0.5B") # Update model name

---
## Use with llama.cpp
Install llama.cpp through brew (works on Mac and Linux)

```bash
brew install llama.cpp

```
Invoke the llama.cpp server or the CLI.

### CLI:
```bash
llama-cli --hf-repo Triangle104/Athena-1-0.5B-Q5_K_M-GGUF --hf-file athena-1-0.5b-q5_k_m.gguf -p "The meaning to life and the universe is"
```

### Server:
```bash
llama-server --hf-repo Triangle104/Athena-1-0.5B-Q5_K_M-GGUF --hf-file athena-1-0.5b-q5_k_m.gguf -c 2048
```

Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.
```
git clone https://github.com/ggerganov/llama.cpp
```

Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).
```
cd llama.cpp && LLAMA_CURL=1 make
```

Step 3: Run inference through the main binary.
```
./llama-cli --hf-repo Triangle104/Athena-1-0.5B-Q5_K_M-GGUF --hf-file athena-1-0.5b-q5_k_m.gguf -p "The meaning to life and the universe is"
```
or 
```
./llama-server --hf-repo Triangle104/Athena-1-0.5B-Q5_K_M-GGUF --hf-file athena-1-0.5b-q5_k_m.gguf -c 2048
```