-
### 🐛 Describe the bug
Exported model for both pt2 and so. pt2 file is 2x larger:
llama31_1bit.pt2 filesize: 3.09GB
llama31_1bit.so filesize: 1.55GB
pt2 command:
```
OMP_NUM_THREA…
-
Is is possible to use RGBtoHDMI for a somewhat simpler but probably unusual case.
The monitor for an old computer is gone and a replacement is not available anymore.
I would like to interface the ol…
-
Any plans on supporting bit depths other than 8 and 16? I would have use for a 1bit greyscale implementation. Also not averse to contributing but would like some hints why other depths weren't impleme…
-
The currently supported model appears rather scarce. Would it be considered to support a broader range of models?
-
After running the selftest file, why is the.res file empty? After running selftest, it shows:
truerand_1bit.res: truerand_1bit.res: No corresponding value for Literal Most Common Value Estimate: p_u
…
-
hi. can you help me print file image? please.
-
With erasmus people getting a 1BIT role or something right now so that they have permissions for subject channels, we should probably have the permissions under the erasmus role so it's easier to reco…
-
### Describe the feature request
Support for quantizing and running quantized models in 4bit, 2bit and 1bit. Also saving and loading these models in onnx format for lower file sizes.
The GPU doesn…
-
When Inkplate is set to 1bit mode, no output is displayed during display refresh. This started happening since branch v0.9.8.
I was able to reproduce it with esp-idf v4.4 and Native_IDF/Hello_World…
-
Noticed that linear barcodes in white on black background (a number of manufacturers use those making their black packaging more attractive) are not recognized: black stripes instead of white are stil…