eriktonon/ glm4.7-flash-uncensored:30b

22 4 hours ago

Zhipu GLM-4.7-Flash 30B-A3B Uncensored MLA+MoE (:30b-a3b Q4_K_M 18.5GB & :iq3 IQ3_M 13.6GB for 16GB GPUs). 202.2 tok/s Decode (:iq3 196.2 tok/s), 50.0% Precision, 66.7% Hallucination, 100% Zero-Refusal. License: MIT.

30b
... /
params
6e63f787bd29 ยท 166B
{
"num_ctx": 16384,
"repeat_penalty": 1.05,
"stop": [
"<|user|>",
"<|endoftext|>",
"<|observation|>"
],
"temperature": 0.25,
"top_k": 40,
"top_p": 0.9
}