TensorRT-LLM Parameter Curation

Engine version: 1.2.1
Discovered at: 2026-07-13 00:00:00+00:00
Discovery method: TorchLlmArgs+TrtLlmArgs.model_json_schema() union with per-field backends + dataclasses.fields(SamplingParams)

Summary: 20/143 parameters curated (95 engine + 48 sampling discovered)

Engine Parameters

Field	Type	Default	Curated?
allreduce_strategy	string | None	AUTO	-
attention_dp_config	AttentionDpConfig | None	null	-
attn_backend	string	TRTLLM	-
backend	string | None	null	yes
batch_wait_max_tokens_ratio	number	0	-
batch_wait_timeout_iters	integer	0	-
batch_wait_timeout_ms	number	0	-
batched_logits_processor	Optional[tensorrt_llm.sampling_params.BatchedLogitsProcessor]	null	-
batching_type	BatchingType | None	null	-
build_config	BuildConfig | None	null	-
cache_transceiver_config	CacheTransceiverConfig | None	null	-
calib_config	CalibConfig | None	null	-
checkpoint_format	string | None	null	-
checkpoint_loader	Optional[tensorrt_llm._torch.models.checkpoints.BaseCheckpointLoader]	null	-
context_parallel_size	integer	1	-
cp_config	object | None	null	-
cuda_graph_config	CudaGraphConfig | None	null	-
custom_tokenizer	string | None	null	-
decoding_config	Optional[tensorrt_llm.llmapi.llm_args.DecodingConfig]	null	-
disable_flashinfer_sampling	boolean	False	-
disable_overlap_scheduler	boolean	False	-
dtype	string	auto	yes
embedding_parallel_mode	string	SHARDING_ALONG_VOCAB	-
enable_attention_dp	boolean	False	-
enable_autotuner	boolean	True	-
enable_build_cache	Union[tensorrt_llm.llmapi.build_cache.BuildCacheConfig, bool]	False	-
enable_chunked_prefill	boolean	False	-
enable_iter_perf_stats	boolean	False	-
enable_iter_req_stats	boolean	False	-
enable_layerwise_nvtx_marker	boolean	False	-
enable_lm_head_tp_in_adp	boolean	False	-
enable_lora	boolean	False	-
enable_min_latency	boolean	False	-
enable_prompt_adapter	boolean	False	-
enable_sleep	boolean	False	-
enable_tqdm	boolean	False	-
env_overrides	object | None	null	-
extended_runtime_perf_knob_config	ExtendedRuntimePerfKnobConfig | None	null	-
fail_fast_on_attention_window_too_large	boolean	False	-
fast_build	boolean	False	yes
force_dynamic_quantization	boolean	False	-
garbage_collection_gen0_threshold	integer	20000	-
gather_generation_logits	boolean	False	-
gpus_per_node	integer | None	null	-
guided_decoding_backend	string | None	null	-
iter_stats_max_iterations	integer | None	null	-
kv_cache_config	KvCacheConfig	null	yes
kv_connector_config	KvCacheConnectorConfig | None	null	-
load_format	Literal['auto', 'dummy']	auto	-
lora_config	LoraConfig | None	null	-
max_batch_size	integer | None	null	yes
max_beam_width	integer | None	null	-
max_input_len	integer | None	null	yes
max_num_tokens	integer | None	8192	yes
max_prompt_adapter_token	integer	0	-
max_seq_len	integer | None	null	yes
max_stats_len	integer	1000	-
mm_encoder_only	boolean	False	-
model	string	null	-
moe_cluster_parallel_size	integer | None	null	-
moe_config	MoeConfig	null	-
moe_expert_parallel_size	integer | None	null	-
moe_tensor_parallel_size	integer | None	null	-
normalize_log_probs	boolean	False	-
num_postprocess_workers	integer	0	-
nvfp4_gemm_config	Nvfp4GemmConfig	null	-
orchestrator_type	string | None	null	-
otlp_traces_endpoint	string | None	null	-
peft_cache_config	PeftCacheConfig | None	null	-
perf_metrics_max_requests	integer	0	-
pipeline_parallel_size	integer	1	yes
postprocess_tokenizer_dir	string | None	null	-
pp_partition	array | None	null	-
print_iter_log	boolean	False	-
quant_config	QuantConfig | None	null	yes
ray_placement_config	RayPlacementConfig | None	null	-
ray_worker_extension_cls	string | None	null	-
reasoning_parser	string | None	null	-
request_stats_max_iterations	integer | None	null	-
return_perf_metrics	boolean	False	-
revision	string | None	null	-
sampler_force_async_worker	boolean	False	-
sampler_type	string | SamplerType	auto	-
scheduler_config	SchedulerConfig	null	yes
skip_tokenizer_init	boolean	False	-
sparse_attention_config	RocketSparseAttentionConfig | DeepSeekSparseAttentionConfig | SkipSoftmaxAttentionConfig | None	null	-
speculative_config	DraftTargetDecodingConfig | EagleDecodingConfig | LookaheadDecodingConfig | MedusaDecodingConfig | MTPDecodingConfig | NGramDecodingConfig | UserProvidedDecodingConfig | SaveHiddenStatesDecodingConfig | AutoDecodingConfig | None	null	-
stream_interval	integer	1	-
tensor_parallel_size	integer	1	yes
tokenizer	string | None	null	-
tokenizer_mode	Literal['auto', 'slow']	auto	-
tokenizer_revision	string | None	null	-
torch_compile_config	TorchCompileConfig | None	null	-
trust_remote_code	boolean	False	-
workspace	string | None	null	-

Sampling Parameters

Field	Type	Default	Curated?
add_special_tokens	bool	True	-
additional_model_outputs	list[str] | None	null	-
apply_batched_logits_processor	bool	False	-
bad	str | list[str] | None	null	-
bad_token_ids	list[int] | None	null	-
beam_search_diversity_rate	float | None	null	-
beam_width_array	list[int] | None	null	-
best_of	int | None	null	-
detokenize	bool	True	-
early_stopping	int | None	null	-
embedding_bias	Tensor | None	null	-
end_id	int | None	null	-
exclude_input_from_output	bool	True	-
frequency_penalty	float | None	null	-
guided_decoding	GuidedDecodingParams | None	null	-
ignore_eos	bool	False	yes
include_stop_str_in_output	bool	False	-
length_penalty	float | None	null	-
logits_processor	LogitsProcessor | list[LogitsProcessor] | None	null	-
logprobs	int | None	null	-
lookahead_config	LookaheadDecodingConfig | None	null	-
max_tokens	int	32	-
min_p	float | None	null	yes
min_tokens	int | None	null	yes
n	int	1	yes
no_repeat_ngram_size	int | None	null	-
pad_id	int | None	null	-
presence_penalty	float | None	null	-
prompt_ignore_length	int | None	null	-
prompt_logprobs	int | None	null	-
repetition_penalty	float | None	null	yes
return_context_logits	bool	False	-
return_encoder_output	bool	False	-
return_generation_logits	bool	False	-
return_perf_metrics	bool	False	-
seed	int | None	null	-
skip_special_tokens	bool	True	-
spaces_between_special_tokens	bool	True	-
stop	str | list[str] | None	null	-
stop_token_ids	list[int] | None	null	-
temperature	float | None	null	yes
top_k	int | None	null	yes
top_p	float | None	null	yes
top_p_decay	float | None	null	-
top_p_min	float | None	null	-
top_p_reset_ids	int | None	null	-
truncate_prompt_tokens	int | None	null	-
use_beam_search	bool	False	-