MCPcopy Index your code
hub / github.com/InternLM/lmdeploy / add_spec_group

Method add_spec_group

lmdeploy/cli/utils.py:756–774  ·  view source on GitHub ↗
(parser)

Source from the content-addressed store, hash-verified

754
755 @staticmethod
756 def add_spec_group(parser):
757 spec_group = parser.add_argument_group('Speculative decoding arguments')
758 spec_group.add_argument('--speculative-algorithm',
759 type=str,
760 default=None,
761 choices=['eagle', 'eagle3', 'deepseek_mtp', 'qwen3_5_mtp'],
762 help='The speculative algorithm to use. `None` means speculative decoding is disabled')
763
764 spec_group.add_argument('--speculative-draft-model',
765 type=str,
766 default=None,
767 help='The path to speculative draft model')
768
769 spec_group.add_argument('--speculative-num-draft-tokens',
770 type=int,
771 default=1,
772 help='The number of speculative tokens to generate per step')
773
774 return spec_group
775
776 @staticmethod
777 def distributed_executor_backend(parser):

Callers 5

add_parser_chatMethod · 0.80
add_parser_api_serverMethod · 0.80
parse_argsFunction · 0.80
parse_argsFunction · 0.80
parse_argsFunction · 0.80

Calls

no outgoing calls

Tested by

no test coverage detected