Welcome to the exciting world of SambaLingo-Hungarian-Chat, a cutting-edge human-aligned chat model that flits gracefully between Hungarian and English, opening up new conversational avenues. This article will guide you through loading the model, interacting with it, and troubleshooting any potential hiccups along the way. Let’s dive in!

What is SambaLingo-Hungarian-Chat?

SambaLingo-Hungarian-Chat is a human-aligned chat model crafted to engage users in both Hungarian and English. Built on the robust foundation of the SambaLingo-Hungarian-Base, it utilizes direct preference optimization to enhance user interactions. This model is adapted from the notable Llama-2-7b by training on a hefty dataset of 59 billion tokens sourced from the Hungarian portion of the Cultura-X, enriching its language capabilities.

How to Load and Interact with the Model

Ready to get started? Follow the steps below:

Loading the Model with Hugging Face

Ensure that you set the use_fast=False parameter when loading the tokenizer. Here’s how to do it:

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("sambanovasystems/SambaLingo-Hungarian-Chat", use_fast=False)
model = AutoModelForCausalLM.from_pretrained("sambanovasystems/SambaLingo-Hungarian-Chat", device_map="auto", torch_dtype="auto")

Interacting with the Model Pipeline

Next, set up your pipeline for text generation:

from transformers import pipeline

pipe = pipeline("text-generation", model="sambanovasystems/SambaLingo-Hungarian-Chat", device_map="auto", use_fast=False)

messages = [{"role": "user", "content": "YOUR_QUESTION"}]
prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
outputs = pipe(prompt)[0]
outputs = outputs["generated_text"]

Suggested Inference Parameters

• Temperature: 0.8
• Repetition penalty: 1.0
• Top-p: 0.9

Prompting Guidelines

To effectively prompt this model, use the following chat template:

Example Prompts
Here’s how to get the conversation rolling:

    
User: Mi a jelentőssége a magyar szürkemarhának?
    
Response: A magyar szürkemarha jelentős kulturális és gazdasági jelentőséggel bír Magyarország számára...


Training Details
The training of SambaLingo followed a structured approach combining supervised fine-tuning (SFT) and Direct Performance Optimization (DPO). This dual-phase methodology is akin to training for a marathon; you first build your endurance and then refine your strategy for peak performance!

Troubleshooting and Tips
If you encounter any issues, here are some strategies to consider:

    
Ensure your dependencies are up-to-date, especially the transformers library.
    
Check your internet connection—loading large models may require a stable and fast connection.
    
Review the configuration settings; incorrect parameters may lead to unexpected behavior.
    
For further assistance and collaboration on AI projects, explore resources at fxis.ai.


Conclusion
At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.

Limitations and Acknowledgments
Like all large language models, SambaLingo comes with inherent limitations like hallucination, repetition, and code-switching. Be sure to use the model appropriately, especially in sensitive contexts.
We extend our heartfelt gratitude to the open-source AI community and our collaborators, including providers of essential datasets, for making this endeavor possible.

Try This Model!
Curious to see it in action? Give SambaLingo-Hungarian-Chat a spin by visiting the SambaLingo-chat-space!