mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-12-17 17:18:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

Expose control over using flash_attention_2

Browse Source 
Signed-off-by: Christoph Auer <cau@zurich.ibm.com>
This commit is contained in:
Christoph Auer
2025-02-25 15:31:32 +01:00
parent 84c77c9fcb
commit 10f64a948c
3 changed files with 19 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
docs/examples/minimal_vlm_pipeline.py
View File

@@ -6,6 +6,7 @@ import yaml
from docling.datamodel.base_models import InputFormat
from docling.datamodel.pipeline_options import (
AcceleratorDevice,
VlmPipelineOptions,
granite_vision_vlm_conversion_options,
smoldocling_vlm_conversion_options,
@@ -24,9 +25,13 @@ pipeline_options.generate_page_images = True
# If force_backend_text = True, text from backend will be used instead of generated text
pipeline_options.force_backend_text = False
## Enable flash_attention_2 with CUDA:
# pipeline_options.accelerator_options.device = AcceleratorDevice.CUDA
# pipeline_options.accelerator_options.cuda_use_flash_attention2 = True
pipeline_options.vlm_options = smoldocling_vlm_conversion_options
# Choose alternative VLM models:
## Choose alternative VLM models:
# pipeline_options.vlm_options = granite_vision_vlm_conversion_options
from docling_core.types.doc import DocItemLabel, ImageRefMode