model in it's 4_1 quantization is almost as good as guanaco-65b
#1
by
onealeph0cc
- opened
It performs very well in my tests, which is an execution of the chain:
(loop
(-> get-task-description
llm_generate_plan
llm_convert_plan_to_json
parse_command))
and what I measure is the number of steps to first command which leads to meaningful results.