Upload folder using huggingface_hub

.gitattributes

@@ -179,3 +179,149 @@ digit_upscaled/validation_images/step_900_unconditional_512x512.png filter=lfs d
 digit_upscaled/validation_images/step_900_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
 digit_upscaled/validation_images/step_950_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
 digit_upscaled/validation_images/step_950_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/benchmarks/base_model/unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/benchmarks/base_model/validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1000/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1000/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1250/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1250/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1500/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1500/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1750/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-1750/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2000/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2000/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2250/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2250/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-250/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-250/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2500/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2500/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2750/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-2750/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-500/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-500/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-750/assets/image_0_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/checkpoint-750/assets/image_1_0.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_0_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_0_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1000_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1000_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_100_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_100_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1050_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1050_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1100_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1100_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1150_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1150_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1200_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1200_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1250_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1250_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1300_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1300_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1350_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1350_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1400_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1400_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1450_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1450_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1500_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1500_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_150_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_150_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1550_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1550_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1600_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1600_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1650_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1650_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1700_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1700_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1750_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1750_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1800_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1800_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1850_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1850_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1900_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1900_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1950_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_1950_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2000_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2000_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_200_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_200_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2050_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2050_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2100_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2100_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2150_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2150_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2200_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2200_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2250_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2250_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2300_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2300_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2350_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2350_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2400_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2400_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2450_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2450_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2500_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2500_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_250_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_250_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2550_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2550_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2600_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2600_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2650_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2650_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2700_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2700_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2750_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2750_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2800_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2800_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2850_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2850_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2900_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2900_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2950_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_2950_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_300_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_300_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_350_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_350_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_400_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_400_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_450_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_450_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_500_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_500_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_50_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_50_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_550_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_550_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_600_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_600_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_650_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_650_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_700_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_700_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_750_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_750_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_800_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_800_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_850_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_850_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_900_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_900_validation_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_950_unconditional_512x512.png filter=lfs diff=lfs merge=lfs -text
+vlcs_upscaled/model/validation_images/step_950_validation_512x512.png filter=lfs diff=lfs merge=lfs -text

vlcs_upscaled/model/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 6
+- Training steps: 3000
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/all_text_cache_files_text-embeds.json

@@ -0,0 +1 @@
+{}

vlcs_upscaled/model/all_vae_cache_files_pacs.json

@@ -0,0 +1 @@
+{}

vlcs_upscaled/model/checkpoint-1000/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 2
+- Training steps: 1000
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/checkpoint-1000/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:889d68b8772e7897888a4b3ce89f269e6b5318c534500638a1238b6088889582
+size 349442426

vlcs_upscaled/model/checkpoint-1000/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0292ac0c32e7aec02b5fad98be33f643652f992f4d5b0c42d3307ccfafb79b3d
+size 116431016

vlcs_upscaled/model/checkpoint-1000/random_states_0.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:865bbff9ce0789ab10f17862fc6b85da55fb8fd0bc263602232a38ee64766d5c
+size 14408

vlcs_upscaled/model/checkpoint-1000/scheduler.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b40c7683c794eb2d2c5f51a0d043e67c0aac67ed037d02fc10ed51860ad3226
+size 1128

vlcs_upscaled/model/checkpoint-1000/training_state.json

@@ -0,0 +1 @@
+{"global_step": 1000, "epoch_step": 1000, "epoch": 3, "exhausted_backends": [], "repeats": {"pacs": 0}}

vlcs_upscaled/model/checkpoint-1250/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 2
+- Training steps: 1250
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/checkpoint-1250/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baefc5dcceeceb30e54adb2a5b86184251e23a68386a6d13a1cfc4ef6bc82470
+size 349442426

vlcs_upscaled/model/checkpoint-1250/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d48056e7a5d32122b81672db8c1777b8d9491ca88348e98072f25b5a1f5170dc
+size 116431016

vlcs_upscaled/model/checkpoint-1250/random_states_0.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b133853f0107ce4e5268fcb5ae0d264f4db8d4bd1931e6f9ae6a6471ce7e5513
+size 14344

vlcs_upscaled/model/checkpoint-1250/scheduler.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f0396e8354db804fd5f3b80b54b57a56511fa235d6226728e6557cfe331c1e6
+size 1128

vlcs_upscaled/model/checkpoint-1250/training_state.json

@@ -0,0 +1 @@
+{"global_step": 1250, "epoch_step": 1250, "epoch": 3, "exhausted_backends": [], "repeats": {"pacs": 0}}

vlcs_upscaled/model/checkpoint-1500/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 3
+- Training steps: 1500
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/checkpoint-1500/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a5203ef3b99bc48123d15f33d2a0757d980b0815b62c80d06398ae2a140894f
+size 349442426

vlcs_upscaled/model/checkpoint-1500/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1d9b454f598ceb4258ad2216ec64424f3d12ea8ba908c35775435edabd7f320
+size 116431016

vlcs_upscaled/model/checkpoint-1500/random_states_0.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ee22891055186d041691f54aa4b7126c693794c535050976985a128ba2678e2
+size 14344

vlcs_upscaled/model/checkpoint-1500/scheduler.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92c21aba5105638209ef13ea982d9ffda2517815d25498abf24fedadcdeec846
+size 1128

vlcs_upscaled/model/checkpoint-1500/training_state.json

@@ -0,0 +1 @@
+{"global_step": 1500, "epoch_step": 1500, "epoch": 4, "exhausted_backends": [], "repeats": {"pacs": 0}}

vlcs_upscaled/model/checkpoint-1750/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 3
+- Training steps: 1750
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/checkpoint-1750/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29680a012e5a8ed1a79a0ad8ae6b2a9e25131a337dc55b298c65bd308204d15d
+size 349442426

vlcs_upscaled/model/checkpoint-1750/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a326e89559656ae8b8689c1f2d4cd0f76c892fe840bd6d8381776c28cf59aa78
+size 116431016

vlcs_upscaled/model/checkpoint-1750/random_states_0.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62cdc2b3120ca6f1f466356bccfb50819cb5137e0c2b90b1d4aa259593342f5d
+size 14344

vlcs_upscaled/model/checkpoint-1750/scheduler.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2922907515a7eda3d4adff43a1e668fe04b9659067a29ebea923dbbcf57043c
+size 1128

vlcs_upscaled/model/checkpoint-1750/training_state.json

@@ -0,0 +1 @@
+{"global_step": 1750, "epoch_step": 1750, "epoch": 4, "exhausted_backends": [], "repeats": {"pacs": 0}}

vlcs_upscaled/model/checkpoint-2000/README.md

@@ -0,0 +1,132 @@
+---
+license: other
+base_model: "sd3/unknown-model"
+tags:
+  - sd3
+  - sd3-diffusers
+  - text-to-image
+  - diffusers
+  - simpletuner
+  - not-for-all-audiences
+  - lora
+  - template:sd-lora
+  - standard
+inference: true
+widget:
+- text: 'unconditional (blank prompt)'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_0_0.png
+- text: 'A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant''s posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.'
+  parameters:
+    negative_prompt: 'blurry, cropped, ugly'
+  output:
+    url: ./assets/image_1_0.png
+---
+
+# simpletuner-lora
+
+This is a standard PEFT LoRA derived from [sd3/unknown-model](https://huggingface.co/sd3/unknown-model).
+
+
+The main validation prompt used during training was:
+```
+A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant.
+```
+
+
+## Validation settings
+- CFG: `7.5`
+- CFG Rescale: `0.0`
+- Steps: `35`
+- Sampler: `FlowMatchEulerDiscreteScheduler`
+- Seed: `42`
+- Resolution: `512x512`
+- Skip-layer guidance: 
+
+Note: The validation settings are not necessarily the same as the [training settings](#training-settings).
+
+You can find some example images in the following gallery:
+
+
+<Gallery />
+
+The text encoder **was not** trained.
+You may reuse the base model text encoder for inference.
+
+
+## Training settings
+
+- Training epochs: 4
+- Training steps: 2000
+- Learning rate: 0.0001
+  - Learning rate schedule: cosine
+  - Warmup steps: 100
+- Max grad norm: 2.0
+- Effective batch size: 16
+  - Micro-batch size: 4
+  - Gradient accumulation steps: 4
+  - Number of GPUs: 1
+- Gradient checkpointing: True
+- Prediction type: flow-matching (extra parameters=['shift=3'])
+- Optimizer: adamw_bf16
+- Trainable parameter precision: Pure BF16
+- Caption dropout probability: 10.0%
+
+
+- LoRA Rank: 128
+- LoRA Alpha: None
+- LoRA Dropout: 0.1
+- LoRA initialisation style: default
+    
+
+## Datasets
+
+### pacs
+- Repeats: 0
+- Total number of images: 7680
+- Total number of aspect buckets: 1
+- Resolution: 1.0 megapixels
+- Cropped: False
+- Crop style: None
+- Crop aspect: None
+- Used for regularisation data: No
+
+
+## Inference
+
+
+```python
+import torch
+from diffusers import DiffusionPipeline
+
+model_id = '/ephemeral/shashmi/llava_lets_go/chimaa_finetuner/stable-diffusion-3.5-medium'
+adapter_id = 'Sarim-Hash/simpletuner-lora'
+pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
+pipeline.load_lora_weights(adapter_id)
+
+prompt = "A simplistic, hand-drawn illustration of an elephant. the elephant is depicted in a walking pose, with its trunk raised slightly. the drawing is done in black ink on a white background. the elephant's posture and the positioning of its legs suggest movement. the style is minimalistic, with clean lines and a lack of intricate details. the lighting appears to be coming from the top left, casting a shadow on the right side of the elephant."
+negative_prompt = 'blurry, cropped, ugly'
+
+## Optional: quantise the model to save on vram.
+## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
+from optimum.quanto import quantize, freeze, qint8
+quantize(pipeline.transformer, weights=qint8)
+freeze(pipeline.transformer)
+    
+pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
+image = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_inference_steps=35,
+    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
+    width=512,
+    height=512,
+    guidance_scale=7.5,
+).images[0]
+image.save("output.png", format="PNG")
+```
+
+
+

vlcs_upscaled/model/checkpoint-2000/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:752bd81543f096a119a6ce8b5aa79abc98ea4cc4270ff7fb733097a7880a4995
+size 349442426

vlcs_upscaled/model/checkpoint-2000/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e853f34851f4a239f7f910f96fedaa04fb78e5ec02e9ef63ab8f61bc12ffe04
+size 116431016