Commits

Branch selector

 main

User selector

RyanJDick

All time

Commit history

Commits on Nov 17, 2023

• Fix double LoRA patching of the UNet. This was presumably added by accident due to a previous merge conflict.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  d756c9b

  Copy full SHA for d756c9b

Commits on Nov 4, 2023

• Fix model cache gc.collect() condition. (#5036)

  Show description for 2bbba32
  RyanJDick

  authored
  2bbba32

  Copy full SHA for 2bbba32

• Fix model cache gc.collect() condition.

  RyanJDick

  committed
  aa02ebf

  Copy full SHA for aa02ebf

Commits on Nov 3, 2023

• Fix bug in model cache reference count checking.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  fb3d0c4

  Copy full SHA for fb3d0c4

• Reduce frequency that we call gc.collect() in the model cache.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  8488ab0

  Copy full SHA for 8488ab0

• Add reminder to clean up our model cache clearing logic.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  875231e

  Copy full SHA for 875231e

• Remove explicit gc.collect() after transferring models from device to CPU. I'm not sure why this was there in the first place, but it was taking a significant amount of time (up to ~1sec in my tests).

  

  RyanJDick

  authored and

  hipsterusername

  committed
  43b3004

  Copy full SHA for 43b3004

• Skip torch.nn.Embedding.reset_parameters(...) when loading a text encoder model.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  e391f3c

  Copy full SHA for e391f3c

• (minor) Fix static checks and typo.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  6e7a3f0

  Copy full SHA for 6e7a3f0

• Add a app config parameter to control the ModelCache logging behavior.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  4a683cc

  Copy full SHA for 4a683cc

• Add log_memory_usage param to ModelCache.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  3781e56

  Copy full SHA for 3781e56

• (minor) Fix int literal typing error.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  267e709

  Copy full SHA for 267e709

• Update get_pretty_snapshot_diff(...) to handle None-snapshots.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  8ff4910

  Copy full SHA for 8ff4910

• Replace deepcopy with a pickle roundtrip in apply_ti(...) to improve speed.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  bac2a75

  Copy full SHA for bac2a75

Commits on Nov 2, 2023

• Improve LoRA patching speed (#5017)

  Show description for a4a7b60
  RyanJDick

  authored
  a4a7b60

  Copy full SHA for a4a7b60

• Further tidying of LoRA patching. Revert some changes that didn't end up being important under the constraint that calculations are done on the same device as the model.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  fa7f6a6

  Copy full SHA for fa7f6a6

• Add minimal unit tests for ModelPatcher.apply_lora(...)

  

  RyanJDick

  authored and

  hipsterusername

  committed
  e92b849

  Copy full SHA for e92b849

• Add TODO note about improving _resolve_lora_key(...).

  

  RyanJDick

  authored and

  hipsterusername

  committed
  61b17c4

  Copy full SHA for 61b17c4

• Patch LoRA on device when model is already on device.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  379d68f

  Copy full SHA for 379d68f

• Remove device and dtype members from LoRAModelRaw, they can too easily get out-of-sync with the underlying layer states.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  545c811

  Copy full SHA for 545c811

• Remove unused _lora_forward_hook(...).

  

  RyanJDick

  authored and

  hipsterusername

  committed
  2ba5b44

  Copy full SHA for 2ba5b44

• auto-format lora.py

  

  RyanJDick

  authored and

  hipsterusername

  committed
  7f4ce51

  Copy full SHA for 7f4ce51

Commits on Oct 18, 2023

• Multi-Image IP-Adapter (#4882)

  Show description for 967a2da
  RyanJDick

  authored
  967a2da

  Copy full SHA for 967a2da

• Merge branch 'main' into ryan/multi-image-ip

  RyanJDick

  committed
  a078efc

  Copy full SHA for a078efc

Commits on Oct 16, 2023

• (minor) Make it more clear that shape annotations are just comments and not commented lines of code.

  RyanJDick

  committed
  3079c75

  Copy full SHA for 3079c75

Commits on Oct 14, 2023

• Bump invocation versions for the multi-image IP feature.

  RyanJDick

  committed
  35ebc9e

  Copy full SHA for 35ebc9e

• Update IP-Adapter unit test for multi-image.

  RyanJDick

  committed
  49279bb

  Copy full SHA for 49279bb

• Add support for multi-image IP-Adapter.

  RyanJDick

  committed
  8464450

  Copy full SHA for 8464450

Commits on Oct 12, 2023

• Add a comment to warn about a necessary action before bumping the diffusers version.

  

  RyanJDick

  authored and

  hipsterusername

  committed
  89db8c8

  Copy full SHA for 89db8c8

Commits on Oct 11, 2023

• Demote model cache logs from warning to debug based on the conversation here: https://discord.com/channels/1020123559063990373/1049495067846524939/1161647290189090816

  

  RyanJDick

  authored and

  hipsterusername

  committed
  40f9e49

  Copy full SHA for 40f9e49

• Bump safetensors to ~=0.4.0 (#4844)

  Show description for 98fa234
  RyanJDick

  authored
  98fa234

  Copy full SHA for 98fa234

Commits on Oct 10, 2023

• Bump safetensors to ~=0.4.0

  RyanJDick

  committed
  fe88923

  Copy full SHA for fe88923

• Improve model load times from disk: skip unnecessary weight init (#4840)

  Show description for 462c1d4
  RyanJDick

  authored
  462c1d4

  Copy full SHA for 462c1d4

• Merge branch 'main' into ryan/optimize-model-load

  RyanJDick

  authored
  0ed3615

  Copy full SHA for 0ed3615

• (minor) Fix Flake8.

  RyanJDick

  committed
  f3c138a

  Copy full SHA for f3c138a