Make the hf model compatible with transformers older versions

Hello!
Yesterday we merged the model compatible with hf CLIP class. It works fine on the latest transformers version.

However, I found out that it can't be loaded on the older versions of transformers (specifically, 4.20.1), so I converted the open_clip model on transformers==4.20.1 and it works like a charm on version >= 4.19.

Decided not to check on older versions because there are quite outdated.

config.json

@@ -1,5 +1,4 @@
 {
-  "_commit_hash": null,
   "architectures": [
     "CLIPModel"
   ],
@@ -13,13 +12,13 @@
     "architectures": null,
     "attention_dropout": 0.0,
     "bad_words_ids": null,
-    "begin_suppress_tokens": null,
     "bos_token_id": 0,
     "chunk_size_feed_forward": 0,
     "cross_attention_hidden_size": null,
     "decoder_start_token_id": null,
     "diversity_penalty": 0.0,
     "do_sample": false,
+    "dropout": 0.0,
     "early_stopping": false,
     "encoder_no_repeat_ngram_size": 0,
     "eos_token_id": 2,
@@ -60,17 +59,14 @@
     "pad_token_id": 1,
     "prefix": null,
     "problem_type": null,
-    "projection_dim": 512,
     "pruned_heads": {},
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
     "return_dict": true,
     "return_dict_in_generate": false,
     "sep_token_id": null,
-    "suppress_tokens": null,
     "task_specific_params": null,
     "temperature": 1.0,
-    "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
     "tie_word_embeddings": true,
     "tokenizer_class": null,
@@ -78,11 +74,17 @@
     "top_p": 1.0,
     "torch_dtype": null,
     "torchscript": false,
-    "transformers_version": "4.29.1",
+    "transformers_version": "4.20.1",
     "typical_p": 1.0,
     "use_bfloat16": false,
     "vocab_size": 49408
   },
+  "text_config_dict": {
+    "hidden_act": "gelu",
+    "hidden_size": 768,
+    "intermediate_size": 3072,
+    "num_attention_heads": 12
+  },
   "torch_dtype": "float32",
   "transformers_version": null,
   "vision_config": {
@@ -91,13 +93,13 @@
     "architectures": null,
     "attention_dropout": 0.0,
     "bad_words_ids": null,
-    "begin_suppress_tokens": null,
     "bos_token_id": null,
     "chunk_size_feed_forward": 0,
     "cross_attention_hidden_size": null,
     "decoder_start_token_id": null,
     "diversity_penalty": 0.0,
     "do_sample": false,
+    "dropout": 0.0,
     "early_stopping": false,
     "encoder_no_repeat_ngram_size": 0,
     "eos_token_id": null,
@@ -130,7 +132,6 @@
     "num_attention_heads": 16,
     "num_beam_groups": 1,
     "num_beams": 1,
-    "num_channels": 3,
     "num_hidden_layers": 24,
     "num_return_sequences": 1,
     "output_attentions": false,
@@ -140,17 +141,14 @@
     "patch_size": 14,
     "prefix": null,
     "problem_type": null,
-    "projection_dim": 512,
     "pruned_heads": {},
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
     "return_dict": true,
     "return_dict_in_generate": false,
     "sep_token_id": null,
-    "suppress_tokens": null,
     "task_specific_params": null,
     "temperature": 1.0,
-    "tf_legacy_loss": false,
     "tie_encoder_decoder": false,
     "tie_word_embeddings": true,
     "tokenizer_class": null,
@@ -158,8 +156,16 @@
     "top_p": 1.0,
     "torch_dtype": null,
     "torchscript": false,
-    "transformers_version": "4.29.1",
+    "transformers_version": "4.20.1",
     "typical_p": 1.0,
     "use_bfloat16": false
+  },
+  "vision_config_dict": {
+    "hidden_act": "gelu",
+    "hidden_size": 1024,
+    "intermediate_size": 4096,
+    "num_attention_heads": 16,
+    "num_hidden_layers": 24,
+    "patch_size": 14
   }
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cbae707ef72c3d2748f3c9482e9ca8645d8bbd915d59a86e4f11107c9b341068
-size 1710660257
+oid sha256:ff37869878b61d3a1f2a82dbbd093927ea658f7477e1a347c0ba384591238904
+size 1710664269