Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
ShreyanGondaliya
/
s5-reduced
:latest
63
Downloads
Updated
2 months ago
reduced s5 default context to 4096 tokens to allow better local inference thinks in chinese somtimes??
reduced s5 default context to 4096 tokens to allow better local inference thinks in chinese somtimes??
Cancel
s5-reduced:latest
...
/
params
9744aa0ee273 · 202B
{
"num_ctx": 4096,
"num_gpu": 1,
"num_thread": 8,
"repeat_penalty": 1.2,
"stop": [
"<|user|>",
"<|assistant|>",
"<|system|>",
"<|observation|>"
],
"temperature": 0.5,
"top_p": 0.8
}