Documentation Index
Fetch the complete documentation index at: https://togetherai-migration.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
See all commands with:
Quickstart
Before you can upload a dataset, you will need to format it. See the fine-tuning page for a guide on how to prepare your data.
To upload a new data file:
together files upload <FILENAME>
Hereβs a sample output:
$ together files upload example.jsonl
Uploading example.jsonl: 100%|ββββββββββββββββββββββββββββββ| 5.18M/5.18M [00:01<00:00, 4.20MB/s]
{
"filename": "example.jsonl",
"id": "file-d931200a-6b7f-476b-9ae2-8fddd5112308",
"object": "file"
}
The id field in the response will be the assigned file-id for this file object.
To list previously uploaded files:
Retrieve
To retrieve the metadata of a previously uploaded file,
together files retrieve <FILE-ID>
Hereβs a sample output:
$ together files retrieve file-d931200a-6b7f-476b-9ae2-8fddd5112308
{
"filename": "example.jsonl",
"bytes": 5433223,
"created_at": 1690432046,
"id": "file-d931200a-6b7f-476b-9ae2-8fddd5112308",
"purpose": "fine-tune",
"object": "file",
"LineCount": 0,
"Processed": true
}
Retrieve content
To download a previously uploaded file,
together files retrieve-content <FILE-ID>
Hereβs a sample output:
$ together files retrieve-content file-d931200a-6b7f-476b-9ae2-8fddd5112308
Downloading file-d931200a-6b7f-476b-9ae2-8fddd5112308.jsonl: 100%|ββββββββββ| 5.43M/5.43M [00:00<00:00, 10.0MiB/s]
file-d931200a-6b7f-476b-9ae2-8fddd5112308.jsonl
You can specify the output filename with --output FILENAME or -o FILENAME. By default, the dataset is saved to <FILE-ID>.jsonl.
To delete a previously uploaded file,
together files delete <FILE-ID>
Hereβs a sample output:
$ together files delete file-d931200a-6b7f-476b-9ae2-8fddd5112308
{
"id": "file-d931200a-6b7f-476b-9ae2-8fddd5112308",
"object": "file",
"deleted": "true"
}