MCPcopy Index your code
hub / github.com/NVIDIA/TensorRT-LLM / SpecDecodingParams

Class SpecDecodingParams

tensorrt_llm/layers/attention.py:256–271  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

254
255
256class SpecDecodingParams:
257
258 def __init__(self,
259 spec_decoding_is_generation_length_variable: bool = False,
260 spec_decoding_max_generation_length: int = 1,
261 spec_decoding_generation_lengths: Tensor = None,
262 spec_decoding_position_offsets: Tensor = None,
263 spec_decoding_packed_mask: Tensor = None,
264 spec_decoding_use: Tensor = None):
265
266 self.spec_decoding_is_generation_length_variable = spec_decoding_is_generation_length_variable
267 self.spec_decoding_max_generation_length = spec_decoding_max_generation_length
268 self.spec_decoding_generation_lengths = spec_decoding_generation_lengths
269 self.spec_decoding_position_offsets = spec_decoding_position_offsets
270 self.spec_decoding_packed_mask = spec_decoding_packed_mask
271 self.spec_decoding_use = spec_decoding_use
272
273
274class MropeParams:

Callers 8

test_process_logitsMethod · 0.90
forwardMethod · 0.85
forwardMethod · 0.85
forwardMethod · 0.85
forwardMethod · 0.85
forwardMethod · 0.85
prepare_basic_inputsMethod · 0.85

Calls

no outgoing calls

Tested by 1

test_process_logitsMethod · 0.72