---
license: gpl-3.0
language:
- vi
- en
pipeline_tag: text-generation
tags:
- chemistry
- biology
- code
datasets:
- open-r1/OpenR1-Math-220k
- open-thoughts/OpenThoughts-114k
base_model:
- meta-llama/Llama-3.3-70B-Instruct
- google/gemma-3-27b-it
metrics:
- chrf
library_name: fasttext
---
![Frame 34.png](https://cdn-uploads.huggingface.co/production/uploads/67bb241a51e34f525bd1b1fc/5A89_Ng_DDt57cahHJsxC.png)
# SOFIA X1 MODEL

<!-- Provide a quick summary of what the model is/does. -->

Trained aiming to be an AI-instructor.
Excels in coding, deep-learning and machine-learning.

## Model Details
- **Total Parameters:** 405 billion
- **Active Parameters:** 70 billion
- **Max input tokens:** 128K
- **Max output tokens:** 1M
- **FP8 Utilization**
- **Knowledge cut-off date:** 3/17/2025

### Model Description


- **Developed by:** ngkhoi, quren
- **Model type:** X1 v1.2
- **Language(s) (NLP):** Vietnamese, English
- **License:** GNU


## Risks & Limitation

- Due to the Llama 3.3 leftover cores, the model now still having issue with multilingual.
- Mediocre response speed.
- Requires high-end hardware to run.


## Demo

Discord bot access only.

### Training Data
### ! The model is designed to solve problems and assignments ONLY in Vietnam region !
- **Subjects**
- Mathematics 10
- Physics 10
- Chemistry 10 (confusing dataset with the online websites' results)
- **Programming languages**
- Python
- Lua
- HTML
- CSS
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
- **Integrated**
- Google Gemma (Image Analysis)
- Deep ML
[More Information Needed]

#### Hardware
- **CPU:** >= 8 cores
- **RAM:** > 32GB
- **GPU:** 3000 series and up
- **Storage:** More than 1TB
[More Information Needed]