update readme
Browse files
README.md
CHANGED
|
@@ -51,7 +51,22 @@ The model, while powerful, has limitations inherent to AI, including biases pres
|
|
| 51 |
|
| 52 |
### Training Data
|
| 53 |
|
| 54 |
-
The dataset used was [instruct-legal-refugiados-es](https://huggingface.co/datasets/somosnlp/instruct-legal-refugiados-es)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
|
| 56 |
### Training Procedure
|
| 57 |
The training was done using RTX 4090 from Vast.ai with PeRF and Lora
|
|
|
|
| 51 |
|
| 52 |
### Training Data
|
| 53 |
|
| 54 |
+
The dataset used was [instruct-legal-refugiados-es](https://huggingface.co/datasets/somosnlp/instruct-legal-refugiados-es)
|
| 55 |
+
We wanted to make a conversation model so we investigated the base model prompt in order to make conversational base on [chatml format](https://github.com/MicrosoftDocs/azure-docs/blob/main/articles/ai-services/openai/includes/chat-markup-language.md#working-with-chat-markup-language-chatml)
|
| 56 |
+
|
| 57 |
+
we identified the special tokens so the model could understand the different roles in the conversation
|
| 58 |
+
|
| 59 |
+
Example
|
| 60 |
+
```
|
| 61 |
+
<bos><|im_start|>system
|
| 62 |
+
You are Gemma.<|im_end|>
|
| 63 |
+
<|im_start|>user
|
| 64 |
+
Hello, how are you?<|im_end|>
|
| 65 |
+
<|im_start|>assistant
|
| 66 |
+
I'm doing great. How can I help you today?<|im_end|>\n<eos>
|
| 67 |
+
```
|
| 68 |
+
|
| 69 |
+
so we used [Phil Schmid's gemma chatml tokenizer](https://huggingface.co/philschmid/gemma-tokenizer-chatml) to adapt our dataset for training
|
| 70 |
|
| 71 |
### Training Procedure
|
| 72 |
The training was done using RTX 4090 from Vast.ai with PeRF and Lora
|