sammcj/
deepseek-r1-distilled-llama-70b

tools

142 2 months ago

1 Tag
... /
params
fa89fdfaa98c · 90B
{
"min_p": 0.9,
"num_batch": 1024,
"num_ctx": 65536,
"num_keep": 256,
"temperature": 0.2,
"top_p": 1
}