llama3-gradient:instruct

110.4K 1 year ago

This model extends LLama-3 8B's context length from 8k to over 1m tokens.

8b 70b
... /
params
577073ffcc6c · 110B
{
"num_keep": 24,
"stop": [
"<|start_header_id|>",
"<|end_header_id|>",
"<|eot_id|>"
]
}