A 4-bit, 128g, act_order=True GPTQ quantisation of JackFram/llama-68m, a 68M Llama model created on request for software testing.
Not for normal usage!
