For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python function

bind_decoder_scratch

bind_decoder_scratch()

max.pipelines.lib.module_registration.bind_decoder_scratch(decoder_cls, model_config, batch_processor, *, max_batch_size, devices)

source

Allocates a registered decoder’s workspace and gives it to the batcher.

Parameters:

• decoder_cls (type[Any]) – The registered DECODER class.
• model_config (Any) – The finalized config the decoder is built from.
• batch_processor (object) – The pipeline model’s batch processor.
• max_batch_size (int) – The most requests a step holds.
• devices (Sequence[Device]) – The devices the decoder runs on.

Returns:

The workspace, or None when the decoder declares none.

Raises:

TypeError – If the decoder declares workspace and the batch processor isn’t a DecoderBatchProcessor, which passes it.

Return type:

DecoderScratch | None