mirror of
https://github.com/huggingface/candle.git
synced 2025-06-15 02:16:37 +00:00
6056fd5c90
* 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>
candle-moondream
Moondream is a computer-vision model can answer real-world questions about images. It's tiny by today's models, with only 1.6B parameters. That enables it to run on a variety of devices, including mobile phones and edge devices.
Running some examples
First download an example image
$ wget https://raw.githubusercontent.com/vikhyat/moondream/main/assets/demo-1.jpg
Now you can run Moondream from the candle-examples crate:
$ cargo run --example moondream --release -- --prompt "What is the girl eating?" --image "./demo-1.jpg"
avavx: false, neon: true, simd128: false, f16c: false
temp: 0.00 repeat-penalty: 1.00 repeat-last-n: 64
retrieved the files in 3.395583ms
Running on CPU, to run on GPU(metal), build this example with `--features metal`
loaded the model in 5.485493792s
loaded and encoded the image Tensor[dims 3, 378, 378; f32] in 4.801396417s
starting the inference loop
The girl is eating a hamburger.<
9 tokens generated (0.68 token/s)