Added model and processor

config.json

@@ -13,7 +13,7 @@
     "architectures": null,
     "attention_dropout": 0.0,
     "bad_words_ids": null,
-    "bos_token_id": 49406,
+    "bos_token_id": 0,
     "chunk_size_feed_forward": 0,
     "cross_attention_hidden_size": null,
     "decoder_start_token_id": null,
@@ -22,7 +22,7 @@
     "dropout": 0.0,
     "early_stopping": false,
     "encoder_no_repeat_ngram_size": 0,
-    "eos_token_id": 49407,
+    "eos_token_id": 2,
     "exponential_decay_length_penalty": null,
     "finetuning_task": null,
     "forced_bos_token_id": null,
@@ -44,12 +44,12 @@
     },
     "layer_norm_eps": 1e-05,
     "length_penalty": 1.0,
-    "max_length": 20,
+    "max_length": 16,
     "max_position_embeddings": 16,
     "min_length": 0,
     "model_type": "owlvit_text_model",
     "no_repeat_ngram_size": 0,
-    "num_attention_heads": 12,
+    "num_attention_heads": 16,
     "num_beam_groups": 1,
     "num_beams": 1,
     "num_hidden_layers": 12,
@@ -57,7 +57,7 @@
     "output_attentions": false,
     "output_hidden_states": false,
     "output_scores": false,
-    "pad_token_id": 0,
+    "pad_token_id": 1,
     "prefix": null,
     "problem_type": null,
     "pruned_heads": {},
@@ -68,6 +68,7 @@
     "sep_token_id": null,
     "task_specific_params": null,
     "temperature": 1.0,
+    "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
     "tie_word_embeddings": true,
     "tokenizer_class": null,
@@ -145,6 +146,7 @@
     "sep_token_id": null,
     "task_specific_params": null,
     "temperature": 1.0,
+    "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
     "tie_word_embeddings": true,
     "tokenizer_class": null,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16c6725be23de3c3975d2358b1768da908f16e46afa5ec76d0a59a41cad27dec
-size 1735729705
+oid sha256:ce3096703702707ff2d868ec7fa9a8358d5e3bf647e07343fd73996855a27073
+size 1735758757

tokenizer_config.json

@@ -22,6 +22,7 @@
   "name_or_path": "openai/clip-vit-base-patch32",
   "pad_token": "!",
   "processor_class": "OwlViTProcessor",
+  "special_tokens_map_file": "/Users/adirik/.cache/huggingface/transformers/18a566598f286c9139f88160c99f84eec492a26bd22738fa9cb44d5b7e0a5c76.cce1206abbad28826f000510f22f354e53e66a97f7c23745a7dfe27609cc07f5",
   "tokenizer_class": "CLIPTokenizer",
   "unk_token": {
     "__type": "AddedToken",