asif00 commited on
Commit
92c39fe
1 Parent(s): 66dfc83

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -9
README.md CHANGED
@@ -11,12 +11,9 @@ base_model: unsloth/llama-3-8b-bnb-4bit
11
  library_name: transformers
12
  pipeline_tag: question-answering
13
  ---
14
-
15
- Model Description:
16
-
17
  Bangla LLaMA-4bit is a specialized model for context-based question answering and Bengali retrieval augment generation. It is derived from LLaMA 3 7B and trained on the iamshnoo/alpaca-cleaned-bengali dataset. This model is designed to provide accurate responses in Bengali with relevant contextual information. It is integrated with the transformers library, making it easy to use for context-based question answering and Bengali retrieval augment generation in projects.
18
 
19
- Model Details:
20
 
21
  - Model Family: Llama 3 7B
22
  - Language: Bengali
@@ -27,7 +24,8 @@ Model Details:
27
  - Batch Size: 80
28
  - Epoch: 1
29
 
30
- How to Use:
 
31
 
32
  You can use the model with a pipeline for a high-level helper or load the model directly. Here's how:
33
 
@@ -44,7 +42,7 @@ tokenizer = AutoTokenizer.from_pretrained("asif00/bangla-llama-4bit")
44
  model = AutoModelForCausalLM.from_pretrained("asif00/bangla-llama-4bit")
45
  ```
46
 
47
- General Prompt Structure:
48
 
49
  ```python
50
  prompt = """Below is an instruction in Bengali language that describes a task, paired with an input also in Bengali language that provides further context. Write a response in Bengali language that appropriately completes the request.
@@ -60,7 +58,7 @@ prompt = """Below is an instruction in Bengali language that describes a task, p
60
  """
61
  ```
62
 
63
- To get a cleaned up version of the response, you can use the `generate_response` function:
64
 
65
  ```python
66
  def generate_response(question, context):
@@ -72,7 +70,7 @@ def generate_response(question, context):
72
  return response
73
  ```
74
 
75
- Example Usage:
76
 
77
  ```python
78
  question = "ভারতীয় বাঙালি কথাসাহিত্যিক মহাশ্বেতা দেবীর মৃত্যু কবে হয় ?"
@@ -82,6 +80,11 @@ print(answer)
82
  ```
83
 
84
 
85
- Disclaimer:
86
 
87
  The Bangla LLaMA-4bit model has been trained on a limited dataset, and its responses may not always be perfect or accurate. The model's performance is dependent on the quality and quantity of the data it has been trained on. Given more resources, such as high-quality data and longer training time, the model's performance can be significantly improved.
 
 
 
 
 
 
11
  library_name: transformers
12
  pipeline_tag: question-answering
13
  ---
 
 
 
14
  Bangla LLaMA-4bit is a specialized model for context-based question answering and Bengali retrieval augment generation. It is derived from LLaMA 3 7B and trained on the iamshnoo/alpaca-cleaned-bengali dataset. This model is designed to provide accurate responses in Bengali with relevant contextual information. It is integrated with the transformers library, making it easy to use for context-based question answering and Bengali retrieval augment generation in projects.
15
 
16
+ # Model Details:
17
 
18
  - Model Family: Llama 3 7B
19
  - Language: Bengali
 
24
  - Batch Size: 80
25
  - Epoch: 1
26
 
27
+
28
+ # How to Use:
29
 
30
  You can use the model with a pipeline for a high-level helper or load the model directly. Here's how:
31
 
 
42
  model = AutoModelForCausalLM.from_pretrained("asif00/bangla-llama-4bit")
43
  ```
44
 
45
+ # General Prompt Structure:
46
 
47
  ```python
48
  prompt = """Below is an instruction in Bengali language that describes a task, paired with an input also in Bengali language that provides further context. Write a response in Bengali language that appropriately completes the request.
 
58
  """
59
  ```
60
 
61
+ # To get a cleaned up version of the response, you can use the `generate_response` function:
62
 
63
  ```python
64
  def generate_response(question, context):
 
70
  return response
71
  ```
72
 
73
+ # Example Usage:
74
 
75
  ```python
76
  question = "ভারতীয় বাঙালি কথাসাহিত্যিক মহাশ্বেতা দেবীর মৃত্যু কবে হয় ?"
 
80
  ```
81
 
82
 
83
+ # Disclaimer:
84
 
85
  The Bangla LLaMA-4bit model has been trained on a limited dataset, and its responses may not always be perfect or accurate. The model's performance is dependent on the quality and quantity of the data it has been trained on. Given more resources, such as high-quality data and longer training time, the model's performance can be significantly improved.
86
+
87
+
88
+ # Resources:
89
+ Work in progress...
90
+