TensorRT-LLMs/tests/unittest/api_stability/references/sampling_params.yaml

methods:
  __init__:
    name: __init__
    parameters:
      add_special_tokens:
        annotation: bool
        default: true
        name: add_special_tokens
      additional_model_outputs:
        annotation: Optional[List[tensorrt_llm.sampling_params.AdditionalModelOutput]]
        default: null
        name: additional_model_outputs
      apply_batched_logits_processor:
        annotation: bool
        default: false
        name: apply_batched_logits_processor
      bad:
        annotation: Union[List[str], str, NoneType]
        default: null
        name: bad
      bad_token_ids:
        annotation: Optional[List[int]]
        default: null
        name: bad_token_ids
      beam_search_diversity_rate:
        annotation: Optional[float]
        default: null
        name: beam_search_diversity_rate
      beam_width:
        annotation: int
        default: 1
        name: beam_width
      best_of:
        annotation: Optional[int]
        default: null
        name: best_of
      detokenize:
        annotation: bool
        default: true
        name: detokenize
      early_stopping:
        annotation: Optional[int]
        default: null
        name: early_stopping
      embedding_bias:
        annotation: Optional[torch.Tensor]
        default: null
        name: embedding_bias
      end_id:
        annotation: Optional[int]
        default: null
        name: end_id
      exclude_input_from_output:
        annotation: bool
        default: true
        name: exclude_input_from_output
      frequency_penalty:
        annotation: Optional[float]
        default: null
        name: frequency_penalty
      guided_decoding:
        annotation: Optional[tensorrt_llm.sampling_params.GuidedDecodingParams]
        default: null
        name: guided_decoding
      ignore_eos:
        annotation: bool
        default: false
        name: ignore_eos
      include_stop_str_in_output:
        annotation: bool
        default: false
        name: include_stop_str_in_output
      length_penalty:
        annotation: Optional[float]
        default: null
        name: length_penalty
      logits_processor:
        annotation: Optional[tensorrt_llm.sampling_params.LogitsProcessor]
        default: null
        name: logits_processor
      lookahead_config:
        annotation: Optional[tensorrt_llm.bindings.executor.LookaheadDecodingConfig]
        default: null
        name: lookahead_config
      max_new_tokens:
        annotation: Optional[int]
        default: null
        name: max_new_tokens
      max_tokens:
        annotation: int
        default: 32
        name: max_tokens
      min_length:
        annotation: Optional[int]
        default: null
        name: min_length
      min_p:
        annotation: Optional[float]
        default: null
        name: min_p
      min_tokens:
        annotation: Optional[int]
        default: null
        name: min_tokens
      n:
        annotation: int
        default: 1
        name: n
      no_repeat_ngram_size:
        annotation: Optional[int]
        default: null
        name: no_repeat_ngram_size
      num_return_sequences:
        annotation: Optional[int]
        default: null
        name: num_return_sequences
      pad_id:
        annotation: Optional[int]
        default: null
        name: pad_id
      presence_penalty:
        annotation: Optional[float]
        default: null
        name: presence_penalty
      random_seed:
        annotation: Optional[int]
        default: null
        name: random_seed
      repetition_penalty:
        annotation: Optional[float]
        default: null
        name: repetition_penalty
      return_context_logits:
        annotation: bool
        default: false
        name: return_context_logits
      return_encoder_output:
        annotation: bool
        default: false
        name: return_encoder_output
      return_generation_logits:
        annotation: bool
        default: false
        name: return_generation_logits
      return_log_probs:
        annotation: bool
        default: false
        name: return_log_probs
      return_perf_metrics:
        annotation: bool
        default: false
        name: return_perf_metrics
      seed:
        annotation: Optional[int]
        default: null
        name: seed
      skip_special_tokens:
        annotation: bool
        default: true
        name: skip_special_tokens
      spaces_between_special_tokens:
        annotation: bool
        default: true
        name: spaces_between_special_tokens
      stop:
        annotation: Union[List[str], str, NoneType]
        default: null
        name: stop
      stop_token_ids:
        annotation: Optional[List[int]]
        default: null
        name: stop_token_ids
      temperature:
        annotation: Optional[float]
        default: null
        name: temperature
      top_k:
        annotation: Optional[int]
        default: null
        name: top_k
      top_p:
        annotation: Optional[float]
        default: null
        name: top_p
      top_p_decay:
        annotation: Optional[float]
        default: null
        name: top_p_decay
      top_p_min:
        annotation: Optional[float]
        default: null
        name: top_p_min
      top_p_reset_ids:
        annotation: Optional[int]
        default: null
        name: top_p_reset_ids
      truncate_prompt_tokens:
        annotation: Optional[int]
        default: null
        name: truncate_prompt_tokens
      use_beam_search:
        annotation: bool
        default: false
        name: use_beam_search
    return_annotation: None
properties: {}