Update README.md
Browse files
README.md
CHANGED
@@ -21,18 +21,49 @@ This model is a fine-tuned version of [unsloth/llama-3-8b-Instruct-bnb-4bit](htt
|
|
21 |
|
22 |
## Model description
|
23 |
|
24 |
-
|
25 |
|
26 |
-
|
|
|
|
|
27 |
|
28 |
-
|
|
|
|
|
29 |
|
30 |
-
|
|
|
|
|
31 |
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
## Training procedure
|
35 |
|
|
|
|
|
36 |
### Training hyperparameters
|
37 |
|
38 |
The following hyperparameters were used during training:
|
|
|
21 |
|
22 |
## Model description
|
23 |
|
24 |
+
The notebook is structured to guide the user through the fine-tuning process with the following components:
|
25 |
|
26 |
+
1. **Setup and Configuration**:
|
27 |
+
- Imports necessary libraries and sets up the environment.
|
28 |
+
- Configures GPU settings and initializes the Jupyter Widgets.
|
29 |
|
30 |
+
2. **Data Preparation**:
|
31 |
+
- Loads and preprocesses the dataset.
|
32 |
+
- Splits the data into training and validation sets.
|
33 |
|
34 |
+
3. **Model Initialization**:
|
35 |
+
- Loads the pre-trained model.
|
36 |
+
- Configures the model for fine-tuning.
|
37 |
|
38 |
+
4. **Training Loop**:
|
39 |
+
- Implements the training loop with real-time progress updates.
|
40 |
+
- Displays training metrics and updates the progress bar widget.
|
41 |
+
|
42 |
+
## How to use
|
43 |
+
|
44 |
+
1. **Install Required Libraries**
|
45 |
+
|
46 |
+
```python
|
47 |
+
import torch
|
48 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
49 |
+
from unsloth import FastLanguageModel
|
50 |
+
from unsloth.chat_templates import get_chat_template
|
51 |
+
from peft import PeftModel, PeftConfig
|
52 |
+
```
|
53 |
+
|
54 |
+
|
55 |
+
2. **Load the Model and Tokenizer**
|
56 |
+
|
57 |
+
3. **Prepare Inputs**
|
58 |
+
|
59 |
+
4. **Run Inference**
|
60 |
+
|
61 |
+
5.
|
62 |
|
63 |
## Training procedure
|
64 |
|
65 |
+
|
66 |
+
|
67 |
### Training hyperparameters
|
68 |
|
69 |
The following hyperparameters were used during training:
|