Hugging Face's logo

OdiaGenAI-LLM
/
Llama3_8B_Odia_Unsloth

PEFT
Safetensors
llama
Model card Files Community
Llama3_8B_Odia_Unsloth
File size: 3,267 Bytes 
e328a5b
 
 
 
 
 
 
 
 
 
 
4c2bb6b
e328a5b
4c2bb6b
 
 
e328a5b
 
 
 
 
 
 
4c2bb6b
 
 
 
0fc529e
4c2bb6b
3d13e2f
 
 
 
 
 
 
 
 
 
 
 
 
e80cbbf
3d13e2f
0fc529e
e80cbbf
4c2bb6b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e80cbbf
0fc529e
e80cbbf
4c2bb6b
 
 
 
3d13e2f
4c2bb6b
 
 
 
 
 
 
 
 
e328a5b
 
 
 
 
 
 
476cbce
70de0d5
e328a5b
 
 
 
 
 
 
 
 
4c2bb6b
df2ed89
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
 
---
license: llama3
library_name: peft
base_model: unsloth/llama-3-8b-bnb-4bit
model-index:
- name: Llama3_8B_Odia_Unsloth
  results: []
---

# Llama3_8B_Odia_Unsloth

Llama3_8B_Odia_Unsloth is a fine-tuned Odia large language model with 8 billion parameters, and it is based on Llama3. The model is fine-tuned on a comprehensive [171k Odia instruction set](https://huggingface.co/datasets/OdiaGenAI/all_combined_odia_171k), encompassing domain-specific and cultural nuances.

The fine-tuning process leverages Unsloth, expediting the training process for optimal efficiency.

For more details about the model, data, training procedure, and evaluations, go through the blog [post](https://www.odiagenai.org/blog/odiagenai-releases-llama3-fine-tuned-model-for-the-odia-language).

## Model Description
* Model type: A 8B fine-tuned model
* Primary Language(s): Odia and English 
* License: Llama3


## Inference

Sample inference script.

### Installation
```
#Install Unsloth
%%capture
import torch
major_version, minor_version = torch.cuda.get_device_capability()
# Must install separately since Colab has torch 2.2.1, which breaks packages
!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
if major_version >= 8:
    # Use this for new GPUs like Ampere, Hopper GPUs (RTX 30xx, RTX 40xx, A100, H100, L40)
    !pip install --no-deps packaging ninja einops flash-attn xformers trl peft accelerate bitsandbytes
else:
    # Use this for older GPUs (V100, Tesla T4, RTX 20xx)
    !pip install --no-deps xformers trl peft accelerate bitsandbytes
pass
```

### Model loading
```
from unsloth import FastLanguageModel
import torch
max_seq_length = 2048 
dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "OdiaGenAI-LLM/Llama3_8B_Odia_Unsloth",
    max_seq_length = max_seq_length,
    dtype = dtype,
    load_in_4bit = load_in_4bit,
)

alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

### Instruction:
{}

### Input:
{}

### Response:
{}"""
```
### Inference
```
FastLanguageModel.for_inference(model) 
inputs = tokenizer(
[
    alpaca_prompt.format(
        "କୋଭିଡ୍ 19 ର ଲକ୍ଷଣଗୁଡ଼ିକ କ’ଣ?", # instruction
        "", # input
        "", # output - leave this blank for generation!
    )
], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 512, use_cache = True)
tokenizer.batch_decode(outputs)

```

### Citation Information

If you find this model useful, please consider giving 👏 and citing:

```
@misc{Llama3_8B_Odia_Unsloth,
  author = {Shantipriya Parida and Sambit Sekhar and Debasish Dhal and Shakshi Panwar},
  title = {OdiaGenAI Releases Llama3 Fine-tuned Model for the Odia Language},
  year = {2024},
  publisher = {Hugging Face},
  journal = {Hugging Face repository},
  howpublished = {\url{https://huggingface.co/OdiaGenAI}},
}
```

### Contributions

- Dr.Shantipriya Parida
- Sambit Sekhar
- Debasish Dhal
- Shakshi Panwar