Arthur Samuel Galego Panucci FIgueiredo
commited on
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,65 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- pt
|
| 5 |
+
pipeline_tag: text-generation
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# MiniText-v1.0
|
| 9 |
+
|
| 10 |
+
MiniText-v1.0 is a minimal character-level language model trained from scratch
|
| 11 |
+
to learn basic Portuguese text patterns.
|
| 12 |
+
|
| 13 |
+
This project in the future will explore all the language modeling limits:
|
| 14 |
+
reasoning
|
| 15 |
+
math
|
| 16 |
+
code
|
| 17 |
+
(ALL WITH 10K PARAMETERS)
|
| 18 |
+
|
| 19 |
+
This project explores the lower limits of language modeling:
|
| 20 |
+
how small can a neural network be and still produce coherent text?
|
| 21 |
+
|
| 22 |
+
## Model details
|
| 23 |
+
|
| 24 |
+
- Architecture: custom MiniText (character-level)
|
| 25 |
+
- Parameters: 10k (educational scale)
|
| 26 |
+
- Training data: synthetic Portuguese dataset
|
| 27 |
+
- Training objective: next-character prediction
|
| 28 |
+
- Language: Portuguese (basic)
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
## What this model can do
|
| 32 |
+
|
| 33 |
+
- Generate simple Portuguese words and sentences
|
| 34 |
+
- Learn grammatical structure
|
| 35 |
+
- Mix domains (language + math) as a base model
|
| 36 |
+
|
| 37 |
+
## What this model is NOT
|
| 38 |
+
|
| 39 |
+
- Not a chatbot
|
| 40 |
+
- Not instruction-tuned
|
| 41 |
+
- Not reasoning-capable
|
| 42 |
+
- Not safe for production use
|
| 43 |
+
|
| 44 |
+
This is a **base model** intended for research, experimentation, and education.
|
| 45 |
+
|
| 46 |
+
## Example output
|
| 47 |
+
|
| 48 |
+
Input:
|
| 49 |
+
o gato é
|
| 50 |
+
|
| 51 |
+
Output (example):
|
| 52 |
+
o gato é um animal
|
| 53 |
+
|
| 54 |
+
## How to run inference
|
| 55 |
+
|
| 56 |
+
python infer.py
|
| 57 |
+
|
| 58 |
+
License
|
| 59 |
+
MIT
|
| 60 |
+
|
| 61 |
+
Training Environment
|
| 62 |
+
CPU - AMD Ryzen 5 5600G 32GB
|
| 63 |
+
Epochs - 12000
|
| 64 |
+
|
| 65 |
+
Made by: Arthur Samuel(loboGOAT)
|