End of training

README.md

@@ -0,0 +1,72 @@
+---
+base_model: black-forest-labs/FLUX.1-dev
+library_name: diffusers
+license: other
+instance_prompt: a photo of sks poop emoji
+widget: []
+tags:
+- text-to-image
+- diffusers-training
+- diffusers
+- lora
+- flux
+- flux-diffusers
+- template:sd-lora
+---
+
+<!-- This model card has been generated automatically according to the information the training script had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+
+# Flux DreamBooth LoRA - cst7/poop_emoji_flux_lora_500_style
+
+<Gallery />
+
+## Model description
+
+These are cst7/poop_emoji_flux_lora_500_style DreamBooth LoRA weights for black-forest-labs/FLUX.1-dev.
+
+The weights were trained using [DreamBooth](https://dreambooth.github.io/) with the [Flux diffusers trainer](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README_flux.md).
+
+Was LoRA for the text encoder enabled? True.
+
+## Trigger words
+
+You should use `a photo of sks poop emoji` to trigger the image generation.
+
+## Download model
+
+[Download the *.safetensors LoRA](cst7/poop_emoji_flux_lora_500_style/tree/main) in the Files & versions tab.
+
+## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)
+
+```py
+from diffusers import AutoPipelineForText2Image
+import torch
+pipeline = AutoPipelineForText2Image.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16).to('cuda')
+pipeline.load_lora_weights('cst7/poop_emoji_flux_lora_500_style', weight_name='pytorch_lora_weights.safetensors')
+image = pipeline('a photo of sks poop emoji').images[0]
+```
+
+For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters)
+
+## License
+
+Please adhere to the licensing terms as described [here](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md).
+
+
+## Intended uses & limitations
+
+#### How to use
+
+```python
+# TODO: add an example code snippet for running this diffusion pipeline
+```
+
+#### Limitations and bias
+
+[TODO: provide examples of latent issues and potential remediations]
+
+## Training details
+
+[TODO: describe the data used to train the model]

checkpoint-500/optimizer.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2d46c72b8e557cee7159caf1864c840a9a53c9dd1400493e747933c0d892cc3
+size 81411010

checkpoint-500/pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:598529761a6136276738a8739fa8f020239ecde0f171285aad19cea9d6b9993d
+size 23107256

checkpoint-500/random_states_0.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aad16113d1bf0b6367a3fbe08890aa8557e5b1cae71e4a571208b701c4261148
+size 14344

checkpoint-500/scheduler.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72ef7361eb3ca738d82f6b0ce999ee77ee2bbe2c79a64c85a993f2a2eea063c5
+size 1064

poop_emoji_flux_lora_500_style.sh

@@ -0,0 +1,68 @@
+#!/bin/bash
+
+MODEL_NAME="black-forest-labs/FLUX.1-dev"
+DATASET_BASE="/disks/sata5/jiacheng/dataset/dreambooth_dataset"
+ACCELERATOR_ARGS="--num_processes=1 --num_machines=1 --gpu_ids=2"
+
+# 训练函数
+train() {
+  local dataset_name=$1
+
+  # 处理 instance_prompt：
+  # 1. 去掉数据集名称末尾的数字
+  processed_name=$(echo "$dataset_name" | sed 's/[0-9]*$//')
+  # 2. 替换下划线为空格
+  processed_name=$(echo "$processed_name" | sed 's/_/ /g')
+  instance_prompt="a photo of a sks ${processed_name}"  # 最终的 prompt
+
+  # 打印处理后的数据集名称
+  echo "原始 dataset_name: $dataset_name"
+  echo "处理后 processed_name: $processed_name"
+  echo "a photo of a sks ${processed_name}"
+
+  OUTPUT_NAME="${dataset_name}_flux_lora_500_style"
+  OUTPUT_DIR="output/dreambooth_lora_flux/${dataset_name}/$OUTPUT_NAME"
+
+  # 每次训练前，创建对应的文件夹并保存当前脚本
+  mkdir -p $OUTPUT_DIR
+  SCRIPT_BACKUP="$OUTPUT_DIR/$OUTPUT_NAME.sh"
+  cp "$0" "$SCRIPT_BACKUP"
+  echo "当前脚本已保存为配置文件: $SCRIPT_BACKUP"
+
+  # 保存 instance_prompt 到 txt 文件
+  PROMPT_FILE="$OUTPUT_DIR/prompt.txt"
+  echo "原始 dataset_name: $dataset_name" > "$PROMPT_FILE"
+  echo "处理后 processed_name: $processed_name" >> "$PROMPT_FILE"
+  echo "$instance_prompt" >> "$PROMPT_FILE"
+  echo "instance_prompt 已保存到: $PROMPT_FILE"
+  
+  export NCCL_P2P_DISABLE=1
+  export NCCL_IB_DISABLE=1
+
+  # 执行训练命令
+  accelerate launch $ACCELERATOR_ARGS examples/dreambooth/train_dreambooth_lora_flux.py \
+    --pretrained_model_name_or_path=$MODEL_NAME  \
+    --instance_data_dir="$DATASET_BASE/$dataset_name" \
+    --output_dir=$OUTPUT_DIR \
+    --mixed_precision="bf16" \
+    --train_text_encoder\
+    --instance_prompt="a photo of sks ${processed_name}" \
+    --resolution=512 \
+    --train_batch_size=1 \
+    --guidance_scale=1 \
+    --gradient_accumulation_steps=4 \
+    --optimizer="prodigy" \
+    --learning_rate=1. \
+    --report_to="wandb" \
+    --lr_scheduler="constant" \
+    --lr_warmup_steps=0 \
+    --max_train_steps=500 \
+    --seed="0" \
+    --push_to_hub
+}
+
+# 遍历 `dreambooth_dataset` 目录下的所有数据集
+for dataset in $(ls $DATASET_BASE); do
+  echo "开始训练数据集: $dataset"
+  train "$dataset" || true
+done

prompt.txt

@@ -0,0 +1,3 @@
+原始 dataset_name: poop_emoji
+处理后 processed_name: poop emoji
+a photo of a sks poop emoji

pytorch_lora_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1499be1226bfe53efaaffee9c546646edf84ae8e4d9dff1c62eafafdbdf38b20
+size 23697096