mirror of https://github.com/huggingface/candle.git synced 2025-06-14 01:48:08 +00:00

Files

shua 6056fd5c90 onnx: fix pad, unsqueeze (#2317 )

* onnx: fix pad, unsqueeze

both implementations have off-by-one errors:
- Pad 'reflect' cycle for eg `dim==3` is `[0,1,2,1]` which has length of
  4 (or `dim*2 - 2`) not 5 (current code `dim*2 - 1`)
- Unsqueeze(-1) for tensor with `dim==3` should be 3 (ie `dim+index+1`)
  not 2 (ie currently `dim+index`)

in addition, Pad is incorrectly calculating the starting padding.
If we want to pad out 2 elements to the start, and we have this cycle
of indices of length 6, then we should skip 4 elements, but currently
we skip 2. A more visual representation of what's going on is below:

```
pad_start: 2
data:      [a,b,c,d]
indices:   [0, 1, 2, 3, 2, 1, 0, 1, 2, 3, 2, 1, 0, ..] // zigzag between 0..4
actual:    skip [ c  d| c  b  a  b]
expected:  ~  skip  ~ [ c  b| a  b  c  d]
```

The values between `[` and `|` are padding and the values between
`|` and `]` in the example should match the original data being padded.

* Fix clippy lints.

---------

Co-authored-by: Laurent <laurent.mazare@gmail.com>

2024-07-23 23:10:57 +02:00

assets

docs: add trocr examples (#1692 )

2024-02-10 16:14:50 +01:00

image_processor.rs

onnx: fix pad, unsqueeze (#2317 )

2024-07-23 23:10:57 +02:00

main.rs

Move image tensor to device in trocr example (#2063 )

2024-04-15 06:50:32 +02:00

readme.md

docs: add trocr examples (#1692 )

2024-02-10 16:14:50 +01:00

readme.md

candle-trocr

TrOCR is a transformer OCR Model. In this example it is used to transcribe image text. See the associated model card for details on the model itself.

Supported models include:

--which base: small handwritten OCR model.
--which large: large handwritten OCR model.
--which base-printed: small printed OCR model.
--which large-printed: large printed OCR model.

Running an example

cargo run --example trocr --release -- --image candle-examples/examples/trocr/assets/trocr.png
cargo run --example trocr --release -- --which large --image candle-examples/examples/trocr/assets/trocr.png
cargo run --example trocr --release -- --which base-printed --image candle-examples/examples/trocr/assets/noto.png
cargo run --example trocr --release -- --which large-printed --image candle-examples/examples/trocr/assets/noto.png

Outputs

industry , Mr. Brown commented icily . " Let us have a
industry , " Mr. Brown commented icily . " Let us have a
THE QUICK BROWN FOR JUMPS OVER THE LAY DOG
THE QUICK BROWN FOX JUMPS OVER THE LAZY DOG