Welcome to your guide on utilizing the EZO model, an advanced text generation tool designed to understand and produce content across multiple languages, with a special emphasis on Japanese. With its foundation on Gemma-2-9B-it combined with various tuning techniques, this tutorial will help you harness the model’s capabilities effectively.

Model Information

The EZO model’s framework largely benefits from its robust data set drawn from sources like Japanese Wikipedia and FineWeb. Imagine this model like an experienced chef who has trained in various cuisines but excels in Japanese cooking. While the chef (the model) primarily specializes in one area, the diverse training helps it perform well in several other regions.

Benchmark Results

Terms of Use

Before getting started, it’s essential to understand the terms you’ll need to comply with. The EZO model follows the Gemma Terms of Use. For details, please refer to the official Gemma license page.

Usage Guidelines

• Commercial Use: If you plan to employ this model for commercial purposes, it’s advisable to contact via email at info@axcxept.com for collaboration opportunities.
• Attribution: Acknowledge the source by adding the following statement: “This project utilizes HODACHI/EZO-Common-9B-gemma-2-it, a model based on gemma-2 and fine-tuned by Axcxept co., ltd.”
• Feedback: You’re encouraged to share your experiences or improvements by emailing info@axcxept.com.

Quick Start with Code Snippets

To begin your journey with the EZO model, first, ensure you have the required library installed:

pip install -U transformers

Now, let’s move on to some example codes to help you get started:

Chat Template Example

from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model_id = "HODACHI/EZO-Common-9B-gemma-2-it"
dtype = torch.bfloat16
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="cuda",
    torch_dtype=dtype,
)

chat = [
    { "role": "user", "content": "How do different stages of life influence our understanding of time and death? Please provide examples." },
]

prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
inputs = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt")
outputs = model.generate(input_ids=inputs.to(model.device), max_new_tokens=150)
print(tokenizer.decode(outputs[0]))

Understanding the Code Example

Think of this code as setting up a conversation at a restaurant. You start by calling the waiter (loading the model), who brings you the menu (the tokenizer). When you order your favorite dish (the chat prompt), the kitchen prepares it based on the ingredients available (the model’s training data) and serves it back to you in a delightful presentation (the model’s output).

Common Issues and Troubleshooting

Here are some common stumbling blocks you may encounter while using the EZO model:

• If you face installation issues, ensure you have the latest version of the transformers library.
• In case of model loading errors, check your internet connection and that the model ID is correctly specified.
• If you need help or further insights on AI development, feel free to reach out. For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

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.