Update README.md
Browse files
    	
        README.md
    CHANGED
    
    | @@ -1,15 +1,33 @@ | |
| 1 | 
             
            ---
         | 
| 2 | 
            -
            license:  | 
| 3 | 
             
            datasets:
         | 
| 4 | 
             
            - facebook/belebele
         | 
| 5 | 
             
            ---
         | 
| 6 | 
            -
            Pretrained toy  | 
| 7 |  | 
| 8 | 
            -
             | 
|  | |
| 9 | 
             
            * batch_size = 64
         | 
| 10 | 
             
            * block_size = 256
         | 
| 11 | 
             
            * n_layer = 8
         | 
| 12 | 
             
            * n_head = 8
         | 
| 13 | 
             
            * n_embd = 768
         | 
|  | |
| 14 |  | 
| 15 | 
            -
             | 
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | 
|  | |
| 1 | 
             
            ---
         | 
| 2 | 
            +
            license: other
         | 
| 3 | 
             
            datasets:
         | 
| 4 | 
             
            - facebook/belebele
         | 
| 5 | 
             
            ---
         | 
| 6 | 
            +
            Pretrained toy models. Made with Andrej Karpathy's NanoGPT.
         | 
| 7 |  | 
| 8 | 
            +
            # nano_35m
         | 
| 9 | 
            +
            * Trained late 2023 on part of Tagalog portion of Belebele.
         | 
| 10 | 
             
            * batch_size = 64
         | 
| 11 | 
             
            * block_size = 256
         | 
| 12 | 
             
            * n_layer = 8
         | 
| 13 | 
             
            * n_head = 8
         | 
| 14 | 
             
            * n_embd = 768
         | 
| 15 | 
            +
            * Everything else is left as is.
         | 
| 16 |  | 
| 17 | 
            +
            # nano_76m
         | 
| 18 | 
            +
            * Trained January 2024 on part of Tagalog portion of Belebele.
         | 
| 19 | 
            +
            * batch_size = 64
         | 
| 20 | 
            +
            * block_size = 256
         | 
| 21 | 
            +
            * n_layer = 11
         | 
| 22 | 
            +
            * n_head = 16
         | 
| 23 | 
            +
            * n_embd = 768
         | 
| 24 | 
            +
            * Everything else is left as is.
         | 
| 25 | 
            +
             | 
| 26 | 
            +
            # nano-ito_35m
         | 
| 27 | 
            +
            * Trained March 2024 on part of PALITO Tagalog dataset.
         | 
| 28 | 
            +
            * batch_size = 64
         | 
| 29 | 
            +
            * block_size = 256
         | 
| 30 | 
            +
            * n_layer = 11
         | 
| 31 | 
            +
            * n_head = 16
         | 
| 32 | 
            +
            * n_embd = 512
         | 
| 33 | 
            +
            * Everything else is left as is.
         | 

