List the download instructions at the very top
Browse files
    	
        README.md
    CHANGED
    
    | @@ -5,6 +5,17 @@ language: | |
| 5 | 
             
            - en
         | 
| 6 | 
             
            pretty_name: Red Pajama 1T
         | 
| 7 | 
             
            ---
         | 
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
| 8 | 
             
            ### Dataset Summary
         | 
| 9 |  | 
| 10 | 
             
            RedPajama is a clean-room, fully open-source implementation of the LLaMa dataset.
         | 
| @@ -20,15 +31,6 @@ RedPajama is a clean-room, fully open-source implementation of the LLaMa dataset | |
| 20 | 
             
            | StackExchange | 20 Billion         |
         | 
| 21 | 
             
            | Total         | 1.2 Trillion      |
         | 
| 22 |  | 
| 23 | 
            -
            The dataset consists of 2084 jsonl files. Download the entire dataset using the following command:
         | 
| 24 | 
            -
            ```
         | 
| 25 | 
            -
            wget -i https://data.together.xyz/redpajama-data-1T/v1.0.0/urls.txt
         | 
| 26 | 
            -
            ```
         | 
| 27 | 
            -
             | 
| 28 | 
            -
            A smaller 1B-token sample of the dataset can be found [here](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-1T-Sample).
         | 
| 29 | 
            -
             | 
| 30 | 
            -
            A full set of scripts to recreate the dataset from scratch can be found [here](https://github.com/togethercomputer/RedPajama-Data).
         | 
| 31 | 
            -
             | 
| 32 | 
             
            ### Languages
         | 
| 33 |  | 
| 34 | 
             
            Primarily English, though the Wikipedia slice contains multiple languages.
         | 
|  | |
| 5 | 
             
            - en
         | 
| 6 | 
             
            pretty_name: Red Pajama 1T
         | 
| 7 | 
             
            ---
         | 
| 8 | 
            +
            ### Getting Started
         | 
| 9 | 
            +
             | 
| 10 | 
            +
            The dataset consists of 2084 jsonl files. Download the entire dataset using the following command:
         | 
| 11 | 
            +
            ```
         | 
| 12 | 
            +
            wget -i https://data.together.xyz/redpajama-data-1T/v1.0.0/urls.txt
         | 
| 13 | 
            +
            ```
         | 
| 14 | 
            +
             | 
| 15 | 
            +
            A smaller 1B-token sample of the dataset can be found [here](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-1T-Sample).
         | 
| 16 | 
            +
             | 
| 17 | 
            +
            A full set of scripts to recreate the dataset from scratch can be found [here](https://github.com/togethercomputer/RedPajama-Data).
         | 
| 18 | 
            +
             | 
| 19 | 
             
            ### Dataset Summary
         | 
| 20 |  | 
| 21 | 
             
            RedPajama is a clean-room, fully open-source implementation of the LLaMa dataset.
         | 
|  | |
| 31 | 
             
            | StackExchange | 20 Billion         |
         | 
| 32 | 
             
            | Total         | 1.2 Trillion      |
         | 
| 33 |  | 
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
| 34 | 
             
            ### Languages
         | 
| 35 |  | 
| 36 | 
             
            Primarily English, though the Wikipedia slice contains multiple languages.
         | 

