add

models/openFlamingo/openFlamingo.py

@@ -0,0 +1,16 @@
+from open_flamingo import create_model_and_transforms
+
+model, image_processor, tokenizer = create_model_and_transforms(
+    clip_vision_encoder_path="ViT-L-14",
+    clip_vision_encoder_pretrained="openai",
+    lang_encoder_path="<path to llama weights in HuggingFace format>",
+    tokenizer_path="<path to llama tokenizer in HuggingFace format>",
+    cross_attn_every_n_layers=4
+)
+
+# grab model checkpoint from huggingface hub
+from huggingface_hub import hf_hub_download
+import torch
+
+checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-9B", "checkpoint.pt")
+model.load_state_dict(torch.load(checkpoint_path), strict=False)