vLLM v0.23.0 config scheduler.py

Type: source_file
Publisher: vllm-project
Relationship: first_party
Version: vLLM v0.23.0
Retrieved: Jul 9
Local snapshot: corpus/domains/vllm/core-v0/evidence/snapshots/vllm-code-v0230-config-scheduler.py
SHA-256: 36bdfcf50a8f30d9…

No claims cite this source yet.

Name	Description	Locator
max_model_len	Maximum sequence length, including prompt and output tokens, used by vLLM configuration and scheduler validation.	vllm/config/scheduler.py:29-33
is_encoder_decoder	SchedulerConfig InitVar marking encoder-decoder models; stored in ModelConfig and used to disable chunked prefill and prefix caching.	vllm/config/scheduler.py:35-40
runner_type	SchedulerConfig field selecting the runner type to launch for the model: generate, pooling, or draft.	vllm/config/scheduler.py:46-47
max_num_batched_tokens	Maximum number of tokens to process in a scheduler iteration.	vllm/config/scheduler.py:49-54
max_num_scheduled_tokens	SchedulerConfig field capping the tokens the scheduler may issue in a single iteration; defaults to max_num_batched_tokens.	vllm/config/scheduler.py:56-61
max_num_seqs	Maximum number of sequences to process in a scheduler iteration.	vllm/config/scheduler.py:63-68
max_num_partial_prefills	SchedulerConfig field for the maximum number of sequences that can be partially prefilled concurrently under chunked prefill.	vllm/config/scheduler.py:70-72