davanstrien HF Staff commited on
Commit
cbe9423
·
verified ·
1 Parent(s): daa5aea

Sync from GitHub via hub-sync

Browse files
lighton-ocr2-saturate.py CHANGED
@@ -4,6 +4,11 @@
4
  # "saturate[hf]>=0.1.1",
5
  # "pillow>=10",
6
  # ]
 
 
 
 
 
7
  # ///
8
  """
9
  Convert document images to markdown using LightOnOCR-2 via saturate.
@@ -23,11 +28,11 @@ buys over the -server recipe:
23
  an `[OCR ERROR]` string in the text column; `--retry-errors` re-admits only
24
  those rows on a later run.
25
 
26
- Run on HF Jobs (the script starts `vllm serve` itself; the --image flag
27
- provides the `vllm` binary):
 
28
 
29
- hf jobs uv run --detach --flavor a10g-small -s HF_TOKEN --timeout 4h \\
30
- --image vllm/vllm-openai:latest \\
31
  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/lighton-ocr2-saturate.py \\
32
  <input-dataset> <output-dataset>
33
 
 
4
  # "saturate[hf]>=0.1.1",
5
  # "pillow>=10",
6
  # ]
7
+ #
8
+ # [tool.hf-jobs]
9
+ # image = "vllm/vllm-openai:latest"
10
+ # flavor = "a10g-small"
11
+ # secrets = ["HF_TOKEN"]
12
  # ///
13
  """
14
  Convert document images to markdown using LightOnOCR-2 via saturate.
 
28
  an `[OCR ERROR]` string in the text column; `--retry-errors` re-admits only
29
  those rows on a later run.
30
 
31
+ Run on HF Jobs. The script starts `vllm serve` itself; its [tool.hf-jobs] header
32
+ declares the image that provides the `vllm` binary, the hardware and the HF_TOKEN
33
+ secret (`hf` CLI 1.32+). Pass --timeout for a long run; flags override the header:
34
 
35
+ hf jobs uv run --detach --timeout 4h \\
 
36
  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/lighton-ocr2-saturate.py \\
37
  <input-dataset> <output-dataset>
38
 
ovis-ocr2-saturate.py CHANGED
@@ -4,6 +4,11 @@
4
  # "saturate[hf]>=0.1.1",
5
  # "pillow>=10",
6
  # ]
 
 
 
 
 
7
  # ///
8
  """
9
  Convert document images to markdown using OvisOCR2 via saturate.
@@ -22,11 +27,11 @@ What that buys over the -server recipe:
22
  an `[OCR ERROR]` string in the text column; `--retry-errors` re-admits only
23
  those rows on a later run.
24
 
25
- Run on HF Jobs (the script starts `vllm serve` itself; the --image flag
26
- provides the `vllm` binary):
 
27
 
28
- hf jobs uv run --detach --flavor a10g-small -s HF_TOKEN --timeout 4h \\
29
- --image vllm/vllm-openai:latest \\
30
  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/ovis-ocr2-saturate.py \\
31
  <input-dataset> <output-dataset>
32
 
 
4
  # "saturate[hf]>=0.1.1",
5
  # "pillow>=10",
6
  # ]
7
+ #
8
+ # [tool.hf-jobs]
9
+ # image = "vllm/vllm-openai:latest"
10
+ # flavor = "a10g-small"
11
+ # secrets = ["HF_TOKEN"]
12
  # ///
13
  """
14
  Convert document images to markdown using OvisOCR2 via saturate.
 
27
  an `[OCR ERROR]` string in the text column; `--retry-errors` re-admits only
28
  those rows on a later run.
29
 
30
+ Run on HF Jobs. The script starts `vllm serve` itself; its [tool.hf-jobs] header
31
+ declares the image that provides the `vllm` binary, the hardware and the HF_TOKEN
32
+ secret (`hf` CLI 1.32+). Pass --timeout for a long run; flags override the header:
33
 
34
+ hf jobs uv run --detach --timeout 4h \\
 
35
  https://huggingface.co/datasets/uv-scripts/ocr/raw/main/ovis-ocr2-saturate.py \\
36
  <input-dataset> <output-dataset>
37