komomike commited on
Commit
fe991bb
ยท
verified ยท
1 Parent(s): e84f805

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +57 -14
README.md CHANGED
@@ -1,37 +1,80 @@
1
- # MaringGPT
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
- **MaringGPT** is a fine-tuned language model based on Llama 3 (3B) designed for enhanced domain-specific instruction following and text generation.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
 
5
  ---
6
 
7
  ## ๐Ÿ“ธ Demo Preview
8
 
9
- ![MaringGPT Demo](./01.png)
10
- ![MaringGPT Demo](./02.png)
 
 
11
  ---
12
 
13
  ## โœจ Key Features
14
 
15
- - **Domain-Tailored Fine-Tuning:** Optimized to handle specialized conversational queries and context-aware responses.
16
- - **Lightweight & Efficient:** Powered by a 3B parameter base model, making it fast and easy to deploy on consumer hardware or local environments.
17
- - **Instruction-Tuned Chat:** Uses structured chat templates for interactive dialogue and task execution.
18
 
19
  ---
20
 
21
- ## โš™๏ธ Training Details
 
 
22
 
23
- - **Base Model:** `meta-llama/Llama-3.2-3B-Instruct`
24
- - **Fine-Tuning Method:** Parameter-Efficient Fine-Tuning (PEFT / QLoRA 4-bit quantization) via Unsloth.
25
- - **Precision:** `bfloat16` / `float16`
26
- - **Dataset:** Custom instruction-response dataset formatted with standard chat templates.
 
 
27
 
28
  ---
29
 
30
  ## ๐Ÿš€ Quickstart
31
 
32
- Run **MaringGPT** locally using Hugging Face `transformers`:
33
 
34
- ### Installation
35
 
36
  ```bash
37
  pip install --upgrade transformers torch accelerate
 
 
1
+ ---
2
+ license: llama3.2
3
+ base_model: meta-llama/Llama-3.2-3B-Instruct
4
+ language:
5
+ - en
6
+ - nng
7
+ pipeline_tag: text-generation
8
+ tags:
9
+ - llama-3
10
+ - llama-3.2
11
+ - fine-tuned
12
+ - maringgpt
13
+ - maring-language
14
+ - nlp
15
+ ---
16
+
17
+ <div align="center">
18
+
19
+ # ๐ŸŒŒ MaringGPT
20
 
21
+ **A modern, lightweight AI model built to preserve, translate, and advance digital support for the Maring language.**
22
+
23
+ [![Base Model](https://img.shields.io/badge/Base%20Model-Llama--3.2--3B--Instruct-blue?style=for-the-badge&logo=meta)](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct)
24
+ [![License](https://img.shields.io/badge/License-Llama%203.2-green?style=for-the-badge)](https://llama.meta.com/license)
25
+ [![Framework](https://img.shields.io/badge/Framework-Unsloth%20%2F%20TRL-orange?style=for-the-badge)](https://github.com/unslothai/unsloth)
26
+
27
+ </div>
28
+
29
+ ---
30
+
31
+ ## ๐Ÿ“œ About the Maring Language & History
32
+
33
+ > *"Meiringba" / "Maringa" โ€” People who keep the unquenched fire alive.*
34
+
35
+ **Maring** (ISO 639-3: `nng`) is a Sino-Tibetan language spoken primarily by the Maring community in the Tengnoupal and Chandel districts of Manipur, Northeast India.
36
+
37
+ * **Ethno-Linguistic Roots:** Linguistically classified within the Tibeto-Burman branch, Maring shares unique connections with both Tangkhulic and Kuki-Chin languages.
38
+ * **Cultural Identity:** Traditionally an oral-rich language, its speakers are deeply connected to the hills of Southeast Manipur.
39
+ * **The Digital Gap:** Like many indigenous languages of Northeast India, Maring remains under-represented in modern computational linguistics and digital datasets. **MaringGPT** is created to help bridge this gap by bringing native context and modern AI capabilities into a fine-tuned LLM.
40
 
41
  ---
42
 
43
  ## ๐Ÿ“ธ Demo Preview
44
 
45
+ | Preview 01 | Preview 02 |
46
+ | :---: | :---: |
47
+ | ![MaringGPT Demo 1](./01.png) | ![MaringGPT Demo 2](./02.png) |
48
+
49
  ---
50
 
51
  ## โœจ Key Features
52
 
53
+ * ๐Ÿ—ฃ๏ธ **Language Awareness:** Fine-tuned to understand context and structured text associated with Maring language instruction.
54
+ * โšก **Ultra-Efficient (3B Parameters):** Designed on Meta's Llama 3.2 3B architecture, enabling low latency and local execution on consumer GPUs or mobile platforms.
55
+ * ๐ŸŽฏ **Instruction-Tuned:** Structured dialogue handling via ChatML / Llama-3 system prompts for crisp, natural conversations.
56
 
57
  ---
58
 
59
+ ## โš™๏ธ Training Setup
60
+
61
+ The model was fine-tuned using parameter-efficient optimization techniques:
62
 
63
+ | Parameter | Specification |
64
+ | :--- | :--- |
65
+ | **Base Model** | `meta-llama/Llama-3.2-3B-Instruct` |
66
+ | **Method** | QLoRA (4-bit Quantization) via Unsloth |
67
+ | **Precision** | `bfloat16` |
68
+ | **Data Format** | Custom conversational instruction-response pairs |
69
 
70
  ---
71
 
72
  ## ๐Ÿš€ Quickstart
73
 
74
+ Run **MaringGPT** locally using Python and the `transformers` library:
75
 
76
+ ### 1. Installation
77
 
78
  ```bash
79
  pip install --upgrade transformers torch accelerate
80
+