LLM/docling-serve
1
0
Fork 0
mirror of https://github.com/docling-project/docling-serve.git synced 2025-11-29 08:33:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs: add docs for docling parameters like performance and debug (#424)

Browse Source 
Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>
This commit is contained in:
Michele Dolfi
2025-10-31 14:17:31 +01:00
committed by GitHub
parent 1ec44220f5
commit f3957aeb57
1 changed files with 13 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
docs/configuration.md
View File

@@ -57,6 +57,19 @@ THe following table describes the options to configure the Docling Serve app.
| | `DOCLING_SERVE_API_KEY` | | If specified, all the API requests must contain the header `X-Api-Key` with this value. |
| | `DOCLING_SERVE_ENG_KIND` | `local` | The compute engine to use for the async tasks. Possible values are `local`, `rq` and `kfp`. See below for more configurations of the engines. |
### Docling configuration
Some Docling settings, mostly about performance, are exposed as environment variable which can be used also when running Docling Serve.
| ENV | Default | Description |
| ----|---------|-------------|
| `DOCLING_NUM_THREADS` | `4` | Number of concurrent threads used for the `torch` CPU execution. |
| `DOCLING_DEVICE` | | Device used for the model execution. Valid values are `cpu`, `cude`, `mps`. When unset, the best device is chosen. For CUDA-enabled environments, you can choose which GPU using the syntax `cuda:0`, `cuda:1`, ... |
| `DOCLING_PERF_PAGE_BATCH_SIZE` | `4` | Number of pages processed in the same batch. |
| `DOCLING_PERF_ELEMENTS_BATCH_SIZE` | `8` | Number of document items/elements processed in the same batch during enrichment. |
| `DOCLING_DEBUG_PROFILE_PIPELINE_TIMINGS` | `false` | When enabled, Docling will provide detailed timings information. |
### Compute engine
Docling Serve can be deployed with several possible of compute engine.