Upload inference_RSICD_1.py with huggingface_hub

inference_RSICD_1.py

@@ -0,0 +1,56 @@
+import torch
+from diffusers import FluxPipeline
+import json
+from diffusers import DiffusionPipeline, FluxTransformer2DModel
+from torchao.quantization import quantize_, int8_weight_only
+from sd_embed.embedding_funcs import get_weighted_text_embeddings_flux1
+import os
+import random
+import gc
+
+for w in ["1", "2", "3", "4", "5"]:
+    os.makedirs(f"/data/xcl/sd-scripts/result/RSICD_test/{w}k", exist_ok=True)
+
+    transformer = FluxTransformer2DModel.from_pretrained(
+        "/data/xcl/flux/model/flux_dev/models--black-forest-labs--FLUX.1-dev/snapshots/0ef5fff789c832c5c7f4e127f94c8b54bbcced44"
+        , subfolder = "transformer"
+        , torch_dtype = torch.bfloat16
+    )
+    # quantize_(transformer, int8_weight_only())
+
+    pipe = FluxPipeline.from_pretrained("/data/xcl/flux/model/flux_dev/models--black-forest-labs--FLUX.1-dev/snapshots/0ef5fff789c832c5c7f4e127f94c8b54bbcced44", transformer = transformer, torch_dtype=torch.bfloat16)
+    pipe.enable_model_cpu_offload() #save some VRAM by offloading the model to CPU. Remove this if you have enough GPU power
+
+    pipe.load_lora_weights(
+        f"/data/xcl/sd-scripts/output/RSICD_test/rsicd_test-step000{w}0000.safetensors"
+    )
+
+    with open(r"/data/xcl/dataSet/RSICD_1/test_captions.json", "r", encoding="utf-8") as file:
+        data = json.load(file)
+
+    for key, value in data.items():
+
+        if os.path.exists(os.path.join(f"/data/xcl/sd-scripts/result/RSICD_test/{w}k", f"{key}.png")):
+            print(f"跳过 {key}")
+            continue
+
+        prompt = value
+
+        prompt_embeds, pooled_prompt_embeds = get_weighted_text_embeddings_flux1(
+            pipe        = pipe
+            , prompt    = prompt
+        )
+        image = pipe(
+            prompt_embeds               = prompt_embeds
+            , pooled_prompt_embeds      = pooled_prompt_embeds
+            , width                     = 512
+            , height                    = 512
+            , num_inference_steps       = 28
+            , guidance_scale            = 3.5
+            , generator                 = torch.Generator().manual_seed(random.randint(0, 2**32 - 1))
+        ).images[0]
+        image.save(os.path.join(f"/data/xcl/sd-scripts/result/RSICD_test/{w}k", f"{key}.png"))
+    del transformer
+    del pipe
+    gc.collect()
+    torch.cuda.empty_cache()