# WandB Stops Logging Media Remotely After First Model Artifact is Saved

**URL:** <https://community.wandb.ai/t/wandb-stops-logging-media-remotely-after-first-model-artifact-is-saved/8922>\
**Category:** W&B Help\
**Tags:** artifacts\
**Created:** [February 3, 2025, 8:41am UTC](https://community.wandb.ai/t/wandb-stops-logging-media-remotely-after-first-model-artifact-is-saved/8922 "2025-02-03T08:41:20Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![adeeb-islam8](https://avatars.discourse-cdn.com/v4/letter/a/c5a1d2/32.png) [@adeeb-islam8](https://community.wandb.ai/u/adeeb-islam8)\
**Post date:** [February 3, 2025, 8:41am UTC](https://community.wandb.ai/t/wandb-stops-logging-media-remotely-after-first-model-artifact-is-saved/8922/1 "2025-02-03T08:41:20Z")

</div>

WandB correctly logs media (videos, images, etc.) both locally and remotely until the first model artifact is saved. After that, media files are still saved locally but do not appear in the remote media folder in WandB.

I initialize wandb logger using pytorch lightning like this.

Using wandb == 0.13.9. Issue persists even after upgrading wandb

```auto
      logger = WandbLogger(
          project="my_project", 
          name=f"{cfg.TAG}_{time.strftime('%d%B%Yat%H:%M:%S')}",  
          save_dir=save_dir,
          log_model='all'
      )
  
      callbacks = [
          pl.callbacks.ModelSummary(-1),
          pl.callbacks.LearningRateMonitor(),
          ModelCheckpoint(
              save_dir, 
              every_n_train_steps=int(cfg.VAL_CHECK_INTERVAL),
              filename="{epoch}-{step}",  
              save_top_k=-1 # Save all checkpoints
          ),
      ]
  
      trainer = pl.Trainer(
          gpus=cfg.GPUS,
          accelerator='gpu',
          strategy='ddp',
          precision=cfg.PRECISION,
          sync_batchnorm=True,
          max_epochs=None,
          max_steps=cfg.STEPS,
          callbacks=callbacks,
          logger=logger,
          log_every_n_steps=cfg.LOGGING_INTERVAL,
          val_check_interval=cfg.VAL_CHECK_INTERVAL,
          limit_val_batches=limit_val_batches,
          replace_sampler_ddp=replace_sampler_ddp,
          accumulate_grad_batches=cfg.OPTIMIZER.ACCUMULATE_GRAD_BATCHES,
          num_sanity_val_steps=0,
      )
  
      trainer.fit(model, datamodule=data)

```

I log metrics and video like this

```auto
              self.logger.experiment.log({
                  name: wandb.Video(video_np, fps=2, format="gif"),
                  "global_step": self.global_step
              })
              self.logger.experiment.log({
                  'val_acc': val_acc,
                  'val_prec': val_prec,
                  'val_rec': val_rec,
                  'val_f1': val_f1,
                  'val_auroc': val_auroc
              }, step=self.global_step)
              self.logger.experiment.log({'val_iou_' + key: value, 'global_step': self.global_step})
              self.logger.experiment.log({'val_mean_iou': torch.mean(scores), 'global_step': self.global_step})

```