mosaicml / diffusion

mosaicml/diffusion Issues

Is this dataset still expected to be released?
Closed 2 months ago4
When is common canvas going to be released?
Closed 2 months ago23
ERROR:composer.cli.launcher:Rank 0 crashed with exit code -11.
Closed 3 months ago1
Implementing Mosaic Diffusion into Patch-Diffusion
Updated 5 months ago
Training a MosaicML
Updated 5 months ago
Example config for training VAE
Updated 6 months ago4
TypeError("'NoneType' object is not iterable")
Closed 6 months ago1
Multi-node training
Closed 7 months ago
How to do continue training when a job failed
Updated 7 months ago1
Convert Pt model to safetensor
Closed 7 months ago
Forward order differs across ranks: rank 0 is all-gathering 1 parameters while rank 4 is all-gathering -2142209024 parameters
Updated 7 months ago2
What is the benefit of multi phase training?
Closed 8 months ago2
Training with modest data
Closed 8 months ago1
FID from the mainline code is different from https://github.com/mosaicml/diffusion/tree/ejyuen-patch-1
Updated 8 months ago1
How can I solve this problem?
Closed a year ago
Error when training with local precomputed features
Updated 9 months ago3
[dynamo] `UnspecializedNNModuleVariable` does not implement object identity
Closed 9 months ago1
FID and CLIP score
Closed 9 months ago5
FID score changes a lot during the model training
Updated 9 months ago2
Bug report: ValueError: invalid literal for int() with base 10: '/tmp/mds-cache/mds-coco-2014-val-fid-clip-17'
Updated 9 months ago
upsample_nearest_nhwc only supports output tensors with less than int_max
Closed 9 months ago7
FID score is weird
Closed 9 months ago3
Any plan to support "Emu: Enhancing Image Generation Models Using Photogenic Needles in a Haystack"?
Closed 9 months ago2
train error during evaluation with 1 GPU and train with multi GPU
Updated 9 months ago2
An error occurred while running laion_cloudwriter.py
Closed 9 months ago
Bug Report of image_caption.py
Closed 9 months ago1
Compare latents from training code and latents from precompute_latents.py
Closed 9 months ago1
NaN loss and Watchdog timeout error caused by _accumulate_time_across_ranks
Closed 9 months ago1
How can I improve to solve this problem?
Closed 10 months ago10
Why do we load clip-vit-large-patch14?
Closed 10 months ago9
Grad become NaN
Updated a year ago3
can we use coco_captions dataset to train your model？？？
Closed a year ago2
fp32
Updated a year ago1
H100 support issue
Closed a year ago4
Are you planning releasing the trained model?
Closed a year ago3
Error doing composer run.py --config-path yamls/hydra-yamls --config-name SD-2-base-512.yaml
Updated a year ago4
ValueError('cannot mmap an empty file')
Updated a year ago1
Could this code be used for 3090ti
Closed a year ago1
leaked shared_memory
Updated a year ago1
save_overwrite conflicts with autoresume=true
Closed a year ago1
Training verbose logs
Updated a year ago3
Setting for used gpus
Closed a year ago1
Throughput is low
Closed a year ago4
MissingEnvironmentError: Torch distributed is initialized but environment variable NODE_RANK is not set.
Closed a year ago7
Missing/Unexpected Keys when load a checkpoint
Closed a year ago3
NaN loss after 6K steps with batch size 2048
Closed a year ago5
Why are you using COCO2014 to calculate FID and CLIP score
Closed a year ago
How long does it take to run laion_cloudwriter.py and precompute_latents.py for aesthetic-4.5
Closed a year ago1
A question regarding the lr scheduler
Closed a year ago8
No initialize_dist in precompute_latent.py
Closed a year ago2