Training in progress, epoch 1

config.json

@@ -1,6 +1,5 @@
 {
-  "_commit_hash": "a959cf33c20e09215873e338299c900f57047c61",
-  "_name_or_path": "naver-clova-ix/donut-base",
+  "_commit_hash": null,
   "architectures": [
     "VisionEncoderDecoderModel"
   ],
@@ -91,27 +90,26 @@
   },
   "decoder_start_token_id": 0,
   "encoder": {
-    "_name_or_path": "",
+    "_name_or_path": "microsoft/dit-large",
     "add_cross_attention": false,
-    "architectures": null,
+    "architectures": [
+      "BeitForMaskedImageModeling"
+    ],
     "attention_probs_dropout_prob": 0.0,
+    "auxiliary_channels": 256,
+    "auxiliary_concat_input": false,
+    "auxiliary_loss_weight": 0.4,
+    "auxiliary_num_convs": 1,
     "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,
-    "depths": [
-      2,
-      2,
-      14,
-      2
-    ],
     "diversity_penalty": 0.0,
     "do_sample": false,
     "drop_path_rate": 0.1,
     "early_stopping": false,
-    "embed_dim": 128,
     "encoder_no_repeat_ngram_size": 0,
     "eos_token_id": null,
     "exponential_decay_length_penalty": null,
@@ -126,48 +124,55 @@
       "1": "LABEL_1"
     },
     "image_size": [
-      960,
-      720
+      224,
+      224
     ],
     "initializer_range": 0.02,
+    "intermediate_size": 4096,
     "is_decoder": false,
     "is_encoder_decoder": false,
     "label2id": {
       "LABEL_0": 0,
       "LABEL_1": 1
     },
-    "layer_norm_eps": 1e-05,
+    "layer_norm_eps": 1e-12,
+    "layer_scale_init_value": 0.1,
     "length_penalty": 1.0,
     "max_length": 20,
     "min_length": 0,
-    "mlp_ratio": 4.0,
-    "model_type": "donut-swin",
+    "model_type": "beit",
     "no_repeat_ngram_size": 0,
+    "num_attention_heads": 16,
     "num_beam_groups": 1,
     "num_beams": 1,
     "num_channels": 3,
-    "num_heads": [
-      4,
-      8,
-      16,
-      32
-    ],
-    "num_layers": 4,
+    "num_hidden_layers": 24,
     "num_return_sequences": 1,
+    "out_indices": [
+      3,
+      5,
+      7,
+      11
+    ],
     "output_attentions": false,
     "output_hidden_states": false,
     "output_scores": false,
     "pad_token_id": null,
-    "patch_size": 4,
-    "path_norm": true,
+    "patch_size": 16,
+    "pool_scales": [
+      1,
+      2,
+      3,
+      6
+    ],
     "prefix": null,
     "problem_type": null,
     "pruned_heads": {},
-    "qkv_bias": true,
     "remove_invalid_values": false,
     "repetition_penalty": 1.0,
     "return_dict": true,
     "return_dict_in_generate": false,
+    "semantic_loss_ignore_index": 255,
     "sep_token_id": null,
     "suppress_tokens": null,
     "task_specific_params": null,
@@ -178,13 +183,18 @@
     "tokenizer_class": null,
     "top_k": 50,
     "top_p": 1.0,
-    "torch_dtype": null,
+    "torch_dtype": "float32",
     "torchscript": false,
     "transformers_version": "4.25.0.dev0",
     "typical_p": 1.0,
-    "use_absolute_embeddings": false,
+    "use_absolute_position_embeddings": true,
+    "use_auxiliary_head": true,
     "use_bfloat16": false,
-    "window_size": 10
+    "use_mask_token": true,
+    "use_mean_pooling": true,
+    "use_relative_position_bias": false,
+    "use_shared_relative_position_bias": false,
+    "vocab_size": 8192
   },
   "is_encoder_decoder": true,
   "model_type": "vision-encoder-decoder",

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a12ffd3f922819c10c8ccac60968c5c0f678297dfb55185ffe2c6ea589bc613
-size 809181691
+oid sha256:719c994e0d5d21a92add2f2793cb4ac4188ca7b32910b27213d5f70997b78a85
+size 1724178702

runs/Dec05_07-07-43_vidani1/1670224272.8914764/events.out.tfevents.1670224272.vidani1.13838.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64330d43196ac2d2ba10a7635fde4247128ea4909b4a0bd2473ff3d3a356f674
+size 5702

runs/Dec05_07-07-43_vidani1/events.out.tfevents.1670224272.vidani1.13838.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:892dc7694bf97e72d406a70090dc14ad11f3d7711b31aa5b61c2f724051698a6
+size 8480

runs/Dec05_07-16-06_vidani1/1670224578.67999/events.out.tfevents.1670224578.vidani1.2893.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79212da24a403472b0cd848129e205123cff73056db19ce45977ebca6c10ef7b
+size 5702

runs/Dec05_07-16-06_vidani1/events.out.tfevents.1670224578.vidani1.2893.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38ea89999eb28c95738760d75dbc14d6aa41dc14054ffaa29ae2a83e9479af3f
+size 8480

runs/Dec05_07-51-46_vidani1/1670226717.1223223/events.out.tfevents.1670226717.vidani1.13119.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:576950f5e52efa9dfe3e7ccdbd6faf5be0e7bcdc8523d52ad77023941912230a
+size 5702

runs/Dec05_07-51-46_vidani1/events.out.tfevents.1670226717.vidani1.13119.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b525cf77d43aebd6e5ae197f64cc4b8f9a6625163c857360d337b025fec678
+size 8480

runs/Dec05_07-56-25_vidani1/1670226993.9422476/events.out.tfevents.1670226993.vidani1.13119.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:510268d95730db74eab6a4bf6efbe1063b52e6cbe277d128265de85a021c12a9
+size 5702

runs/Dec05_07-56-25_vidani1/events.out.tfevents.1670226993.vidani1.13119.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c5207710d1995b0739abf30db4164a8061a1e852bff57a6f6df43d41568a0be
+size 8480

runs/Dec05_08-00-03_vidani1/1670227214.9180048/events.out.tfevents.1670227214.vidani1.32400.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad88f7ff0b3e8fca24473770e7791e367154fcca078ef35c7c3ab715d679365a
+size 5702

runs/Dec05_08-00-03_vidani1/events.out.tfevents.1670227214.vidani1.32400.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f12ba7cefc9a8991328973e728e51ac86696d2ddd35d53cae62e3b5c00de04ec
+size 8479

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3ef3a018de8ac98506192b15649328bdd91c9df6ddd83057cba7c61ff2f381ea
+oid sha256:61f25a751b6e638a739a6c3292bc56fe0fb9c66b5974c11fe2ca9b72ef90f005
 size 3567