-
### Idea
Use int4 as the compression technique to fit larger models onto Navi machines or possibly MI series machines. Weights would be compressed using encoding scheme that would pack two 4 bits n…
-
请问训练的这个权重是什么呀,是检测的权重还是识别的权重呢
-
Traceback (most recent call last):
File "scripts/inference.py", line 100, in
run()
File "scripts/inference.py", line 39, in run
net = HyperInverter(opts)
File "./models/hyper_inver…
-
Congratulations on your paper.
Are you planning on releasing the trained model weights?
-
Hi, whether adjust UNET structure for 4 depth, instead of 3 depth, and training show an error, for weight names,
ValueError Traceback (most recent call last)
in ()
16 start = time.time()
17 # St…
-
Nice works!!! Looking forward to releasing CLIP model weights !
-
Your work is excellent. Do you have plans to release the weights of the first stage of training?
-
### Summary
If a user specifies `in_chans` and `weights`, and `weights.meta['in_chans']` differs from `in_chans`, the user-specified argument should take precedence and weights should be repeated, si…
-
Would you happen to have the model file and weights?
-
Hi, congratulations on your excellent work on this project!
I have a couple of questions:
- How can I run inference on custom images?
- When will the pretrained model weights be released?
Th…