cwchang / llama3-taide-lx-8b-chat-alpha1-32k

The Llama3-TAIDE model was developed and built by the National Applied Research Laboratories. This version was scaled to 32k using RoPE scaling.

8B

93 Pulls Updated 4 months ago

... /
params
2639fa516443 · 114B
{ "num_keep": 24, "stop": [ "<|start_header_id|>", "<|end_header_id|>", "\"<|eot_id|>\"" ] }