From d98383197fe73e191fca2dbee552b4953ee492e4 Mon Sep 17 00:00:00 2001
From: s444415 <mickoz5@st.amu.edu.pl>
Date: Fri, 16 Dec 2022 12:32:08 +0000
Subject: [PATCH 01/17] modifcation to train and eval

---
 donut-eval.py  | 34 +++++++++++++++++++---------------
 donut-train.py |  8 ++++----
 2 files changed, 23 insertions(+), 19 deletions(-)

diff --git a/donut-eval.py b/donut-eval.py
index c5548e6..cbcc0b7 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -11,21 +11,21 @@ import json
 import torch
 from tqdm.auto import tqdm
 import numpy as np
-
+import pandas as pd
 from donut import JSONParseEvaluator
 
 
 # In[2]:
 
 
-processor = DonutProcessor.from_pretrained("Zombely/plwiki-proto-fine-tuned")
-model = VisionEncoderDecoderModel.from_pretrained("Zombely/plwiki-proto-fine-tuned")
+processor = DonutProcessor.from_pretrained("Zombely/plwiki-proto-fine-tuned-v2")
+model = VisionEncoderDecoderModel.from_pretrained("Zombely/plwiki-proto-fine-tuned-v2")
 
 
 # In[3]:
 
 
-dataset = load_dataset("Zombely/pl-text-images-5000-whole", split="validation")
+dataset = load_dataset("Zombely/diachronia-ocr", split='train')
 
 
 # In[4]:
@@ -38,11 +38,11 @@ model.to(device)
 
 output_list = []
 accs = []
-
+has_metadata = bool(dataset[0].get('ground_truth'))
 
 for idx, sample in tqdm(enumerate(dataset), total=len(dataset)):
     # prepare encoder inputs
-    pixel_values = processor(sample["image"].convert("RGB"), return_tensors="pt").pixel_values
+    pixel_values = processor(sample['image'].convert("RGB"), return_tensors="pt").pixel_values
     pixel_values = pixel_values.to(device)
     # prepare decoder inputs
     task_prompt = "<s_cord-v2>"
@@ -68,16 +68,20 @@ for idx, sample in tqdm(enumerate(dataset), total=len(dataset)):
     seq = seq.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
     seq = re.sub(r"<.*?>", "", seq, count=1).strip()  # remove first task start token
     seq = processor.token2json(seq)
+    if has_metadata:
+        ground_truth = json.loads(sample["ground_truth"])
+        ground_truth = ground_truth["gt_parse"]
+        evaluator = JSONParseEvaluator()
+        score = evaluator.cal_acc(seq, ground_truth)
 
-    ground_truth = json.loads(sample["ground_truth"])
-    ground_truth = ground_truth["gt_parse"]
-    evaluator = JSONParseEvaluator()
-    score = evaluator.cal_acc(seq, ground_truth)
-
-    accs.append(score)
+        accs.append(score)
+    print(seq)
     output_list.append(seq)
+df = pd.DataFrame(map(lambda x: x.get('text_sequence', ''), output_list))
+df.to_csv('out.tsv', sep='\t', header=False, index=False)
 
-scores = {"accuracies": accs, "mean_accuracy": np.mean(accs)}
-print(scores, f"length : {len(accs)}")
-print("Mean accuracy:", np.mean(accs))
+if has_metadata:
+    scores = {"accuracies": accs, "mean_accuracy": np.mean(accs)}
+    print(scores, f"length : {len(accs)}")
+    print("Mean accuracy:", np.mean(accs))
 
diff --git a/donut-train.py b/donut-train.py
index 8eaa167..7ba98f0 100644
--- a/donut-train.py
+++ b/donut-train.py
@@ -22,7 +22,7 @@ from pytorch_lightning.plugins import CheckpointIO
 
 
 DATASET_PATH = "Zombely/pl-text-images-5000-whole"
-PRETRAINED_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned"
+PRETRAINED_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v2"
 START_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned"
 OUTPUT_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v2"
 LOGGING_PATH = "plwiki-proto-ft-second-iter"
@@ -30,8 +30,8 @@ CHECKPOINT_PATH = "./checkpoint"
 
 
 train_config = {
-    "max_epochs":30,
-    "val_check_interval":0.5, # how many times we want to validate during an epoch
+    "max_epochs":1,
+    "val_check_interval":1.0, # how many times we want to validate during an epoch
     "check_val_every_n_epoch":1,
     "gradient_clip_val":1.0,
     "num_training_samples_per_epoch": 800,
@@ -339,7 +339,7 @@ class PushToHubCallback(Callback):
 
 
 
-login(os.environ.get("HUG_TOKKEN", ""))
+login(os.environ.get("HUG_TOKKEN", None), True)
 
 
 # ### Wandb.ai link: https://wandb.ai/michalkozlowski936/Donut?workspace=user-michalkozlowski936

From 8ccd1aabb64fd5d49533cff810ca9cc36baed740 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Fri, 16 Dec 2022 13:53:03 +0100
Subject: [PATCH 02/17] config and params for donut-eval

---
 config.yaml   |   7 +++
 donut-eval.py | 129 ++++++++++++++++++++++++--------------------------
 2 files changed, 68 insertions(+), 68 deletions(-)
 create mode 100644 config.yaml

diff --git a/config.yaml b/config.yaml
new file mode 100644
index 0000000..cb7e730
--- /dev/null
+++ b/config.yaml
@@ -0,0 +1,7 @@
+pretrained_processor_path: "Zombely/plwiki-proto-fine-tuned-v2"
+pretrained_model_path: "Zombely/plwiki-proto-fine-tuned-v2"
+validation_dataset_path: "Zombely/diachronia-ocr"
+validation_dataset_split: "train"
+has_metadata: False
+print_output: True
+output_file_dir: "../../gonito-outs"
\ No newline at end of file
diff --git a/donut-eval.py b/donut-eval.py
index cbcc0b7..23efe11 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -1,9 +1,6 @@
 #!/usr/bin/env python
 # coding: utf-8
 
-# In[1]:
-
-
 from transformers import DonutProcessor, VisionEncoderDecoderModel
 from datasets import load_dataset
 import re
@@ -13,75 +10,71 @@ from tqdm.auto import tqdm
 import numpy as np
 import pandas as pd
 from donut import JSONParseEvaluator
+import argparse
+from sconf import Config
 
+def main(config):
+    processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
+    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
+    dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    model.eval()
+    model.to(device)
+    output_list = []
+    accs = []
 
-# In[2]:
+    for idx, sample in tqdm(enumerate(dataset), total=len(dataset)):
+        # prepare encoder inputs
+        pixel_values = processor(sample['image'].convert("RGB"), return_tensors="pt").pixel_values
+        pixel_values = pixel_values.to(device)
+        # prepare decoder inputs
+        task_prompt = "<s_cord-v2>"
+        decoder_input_ids = processor.tokenizer(task_prompt, add_special_tokens=False, return_tensors="pt").input_ids
+        decoder_input_ids = decoder_input_ids.to(device)
+        
+        # autoregressively generate sequence
+        outputs = model.generate(
+                pixel_values,
+                decoder_input_ids=decoder_input_ids,
+                max_length=model.decoder.config.max_position_embeddings,
+                early_stopping=True,
+                pad_token_id=processor.tokenizer.pad_token_id,
+                eos_token_id=processor.tokenizer.eos_token_id,
+                use_cache=True,
+                num_beams=1,
+                bad_words_ids=[[processor.tokenizer.unk_token_id]],
+                return_dict_in_generate=True,
+            )
 
+        # turn into JSON
+        seq = processor.batch_decode(outputs.sequences)[0]
+        seq = seq.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
+        seq = re.sub(r"<.*?>", "", seq, count=1).strip()  # remove first task start token
+        seq = processor.token2json(seq)
+        if config.has_metadata:
+            ground_truth = json.loads(sample["ground_truth"])
+            ground_truth = ground_truth["gt_parse"]
+            evaluator = JSONParseEvaluator()
+            score = evaluator.cal_acc(seq, ground_truth)
 
-processor = DonutProcessor.from_pretrained("Zombely/plwiki-proto-fine-tuned-v2")
-model = VisionEncoderDecoderModel.from_pretrained("Zombely/plwiki-proto-fine-tuned-v2")
+            accs.append(score)
+        if config.print_output:
+            print(seq)
+        output_list.append(seq)
+    if config.output_file_dir:
+        df = pd.DataFrame(map(lambda x: x.get('text_sequence', ''), output_list))
+        df.to_csv(f'{config.output_file_dir}/{config.pretrained_processor_path}-out.tsv', sep='\t', header=False, index=False)
 
+    if config.has_metadata:
+        scores = {"accuracies": accs, "mean_accuracy": np.mean(accs)}
+        print(scores, f"length : {len(accs)}")
+        print("Mean accuracy:", np.mean(accs))
 
-# In[3]:
-
-
-dataset = load_dataset("Zombely/diachronia-ocr", split='train')
-
-
-# In[4]:
-
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-
-model.eval()
-model.to(device)
-
-output_list = []
-accs = []
-has_metadata = bool(dataset[0].get('ground_truth'))
-
-for idx, sample in tqdm(enumerate(dataset), total=len(dataset)):
-    # prepare encoder inputs
-    pixel_values = processor(sample['image'].convert("RGB"), return_tensors="pt").pixel_values
-    pixel_values = pixel_values.to(device)
-    # prepare decoder inputs
-    task_prompt = "<s_cord-v2>"
-    decoder_input_ids = processor.tokenizer(task_prompt, add_special_tokens=False, return_tensors="pt").input_ids
-    decoder_input_ids = decoder_input_ids.to(device)
-    
-    # autoregressively generate sequence
-    outputs = model.generate(
-            pixel_values,
-            decoder_input_ids=decoder_input_ids,
-            max_length=model.decoder.config.max_position_embeddings,
-            early_stopping=True,
-            pad_token_id=processor.tokenizer.pad_token_id,
-            eos_token_id=processor.tokenizer.eos_token_id,
-            use_cache=True,
-            num_beams=1,
-            bad_words_ids=[[processor.tokenizer.unk_token_id]],
-            return_dict_in_generate=True,
-        )
-
-    # turn into JSON
-    seq = processor.batch_decode(outputs.sequences)[0]
-    seq = seq.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
-    seq = re.sub(r"<.*?>", "", seq, count=1).strip()  # remove first task start token
-    seq = processor.token2json(seq)
-    if has_metadata:
-        ground_truth = json.loads(sample["ground_truth"])
-        ground_truth = ground_truth["gt_parse"]
-        evaluator = JSONParseEvaluator()
-        score = evaluator.cal_acc(seq, ground_truth)
-
-        accs.append(score)
-    print(seq)
-    output_list.append(seq)
-df = pd.DataFrame(map(lambda x: x.get('text_sequence', ''), output_list))
-df.to_csv('out.tsv', sep='\t', header=False, index=False)
-
-if has_metadata:
-    scores = {"accuracies": accs, "mean_accuracy": np.mean(accs)}
-    print(scores, f"length : {len(accs)}")
-    print("Mean accuracy:", np.mean(accs))
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--config", type=str, required=True)
+    args, left_argv = parser.parse_known_args()
+    config = Config(args.config)
+    config.argv_update(left_argv)
 
+    main(config)
\ No newline at end of file

From b7296bb2a94d003133d4ca853b12e7c963357740 Mon Sep 17 00:00:00 2001
From: s444415 <mickoz5@st.amu.edu.pl>
Date: Fri, 16 Dec 2022 14:16:46 +0000
Subject: [PATCH 03/17] config changes

---
 config.yaml   | 7 ++++---
 donut-eval.py | 4 ++--
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/config.yaml b/config.yaml
index cb7e730..4ff337a 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,7 +1,8 @@
-pretrained_processor_path: "Zombely/plwiki-proto-fine-tuned-v2"
-pretrained_model_path: "Zombely/plwiki-proto-fine-tuned-v2"
+pretrained_processor_path: "naver-clova-ix/donut-proto"
+pretrained_model_path: "naver-clova-ix/donut-proto"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False
 print_output: True
-output_file_dir: "../../gonito-outs"
\ No newline at end of file
+output_file_dir: "../../gonito-outs"
+test_name: "proto-test"
diff --git a/donut-eval.py b/donut-eval.py
index 23efe11..5754483 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -63,7 +63,7 @@ def main(config):
         output_list.append(seq)
     if config.output_file_dir:
         df = pd.DataFrame(map(lambda x: x.get('text_sequence', ''), output_list))
-        df.to_csv(f'{config.output_file_dir}/{config.pretrained_processor_path}-out.tsv', sep='\t', header=False, index=False)
+        df.to_csv(f'{config.output_file_dir}/{config.test_name}-out.tsv', sep='\t', header=False, index=False)
 
     if config.has_metadata:
         scores = {"accuracies": accs, "mean_accuracy": np.mean(accs)}
@@ -77,4 +77,4 @@ if __name__ == "__main__":
     config = Config(args.config)
     config.argv_update(left_argv)
 
-    main(config)
\ No newline at end of file
+    main(config)

From de8f89ddb15ace41d14af6ac5606ea90ad4ad19b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Fri, 16 Dec 2022 15:20:10 +0100
Subject: [PATCH 04/17] config for model

---
 donut-eval.py | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/donut-eval.py b/donut-eval.py
index 5754483..456659a 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 # coding: utf-8
 
-from transformers import DonutProcessor, VisionEncoderDecoderModel
+from transformers import DonutProcessor, VisionEncoderDecoderModel, VisionEncoderDecoderConfig
 from datasets import load_dataset
 import re
 import json
@@ -14,8 +14,15 @@ import argparse
 from sconf import Config
 
 def main(config):
+
+    max_length = 768
+    image_size = [1920, 2560]
+    config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
+    config_vision.encoder.image_size = image_size # (height, width)
+    config_vision.decoder.max_length = max_length
+
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
-    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
+    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config)
     dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
     device = "cuda" if torch.cuda.is_available() else "cpu"
     model.eval()

From ed4cabfe350184da1d759d865e3416bc24bbeb52 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Fri, 16 Dec 2022 15:24:04 +0100
Subject: [PATCH 05/17] config update

---
 config.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/config.yaml b/config.yaml
index 4ff337a..be75a38 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,8 +1,8 @@
-pretrained_processor_path: "naver-clova-ix/donut-proto"
-pretrained_model_path: "naver-clova-ix/donut-proto"
+pretrained_processor_path: "naver-clova-ix/donut-base"
+pretrained_model_path: "naver-clova-ix/donut-base"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False
 print_output: True
 output_file_dir: "../../gonito-outs"
-test_name: "proto-test"
+test_name: "base-test"

From ff2ba1533ad83b2b3cb48aa86f9ec6bcad19cbfd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Fri, 16 Dec 2022 15:24:37 +0100
Subject: [PATCH 06/17] fix

---
 donut-eval.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/donut-eval.py b/donut-eval.py
index 456659a..98e59f1 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -15,11 +15,11 @@ from sconf import Config
 
 def main(config):
 
-    max_length = 768
-    image_size = [1920, 2560]
-    config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
-    config_vision.encoder.image_size = image_size # (height, width)
-    config_vision.decoder.max_length = max_length
+    # max_length = 768
+    # image_size = [1920, 2560]
+    # config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
+    # config_vision.encoder.image_size = image_size # (height, width)
+    # config_vision.decoder.max_length = max_length
 
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
     model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config)

From 53a1db3dce1f29df2fb56b01044d6e19a9d1e205 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:24:47 +0100
Subject: [PATCH 07/17] config fix

---
 config.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/config.yaml b/config.yaml
index be75a38..0b2a262 100644
--- a/config.yaml
+++ b/config.yaml
@@ -5,4 +5,4 @@ validation_dataset_split: "train"
 has_metadata: False
 print_output: True
 output_file_dir: "../../gonito-outs"
-test_name: "base-test"
+test_name: "base-test"
\ No newline at end of file

From cfcb15e99911a3d74101119ee9e621ac45820b72 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:26:21 +0100
Subject: [PATCH 08/17] f

---
 config.yaml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/config.yaml b/config.yaml
index 0b2a262..1ae36ee 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,5 +1,5 @@
-pretrained_processor_path: "naver-clova-ix/donut-base"
-pretrained_model_path: "naver-clova-ix/donut-base"
+pretrained_processor_path: "Zombely/plwiki-proto-fine-tuned-v2"
+pretrained_model_path: "Zombely/plwiki-proto-fine-tuned-v2"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False

From 31981cfc515fd27ec2b58218373e60bcb0939349 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:27:14 +0100
Subject: [PATCH 09/17] donut eval fix

---
 donut-eval.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/donut-eval.py b/donut-eval.py
index 98e59f1..bc83f27 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -22,7 +22,7 @@ def main(config):
     # config_vision.decoder.max_length = max_length
 
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
-    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config)
+    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
     dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
     device = "cuda" if torch.cuda.is_available() else "cpu"
     model.eval()

From 3cbed8144e2b0ac696877db4edc15e0e56b26afd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:27:56 +0100
Subject: [PATCH 10/17] naver

---
 config.yaml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/config.yaml b/config.yaml
index 1ae36ee..0b2a262 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,5 +1,5 @@
-pretrained_processor_path: "Zombely/plwiki-proto-fine-tuned-v2"
-pretrained_model_path: "Zombely/plwiki-proto-fine-tuned-v2"
+pretrained_processor_path: "naver-clova-ix/donut-base"
+pretrained_model_path: "naver-clova-ix/donut-base"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False

From bdb7f5ef7e3f186101ce6b5028af4419690032e6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:29:25 +0100
Subject: [PATCH 11/17] additional config

---
 donut-eval.py | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/donut-eval.py b/donut-eval.py
index bc83f27..bdf7c93 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -15,14 +15,12 @@ from sconf import Config
 
 def main(config):
 
-    # max_length = 768
-    # image_size = [1920, 2560]
-    # config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
-    # config_vision.encoder.image_size = image_size # (height, width)
-    # config_vision.decoder.max_length = max_length
+    config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
+    config_vision.encoder.image_size = [1920, 2560] # (height, width)
+    config_vision.decoder.max_length = 768
 
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
-    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
+    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config_vision)
     dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
     device = "cuda" if torch.cuda.is_available() else "cpu"
     model.eval()

From 296647793ea437c375bd76f111e73ddbdb5d8cb4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:33:03 +0100
Subject: [PATCH 12/17] added more config to eval

---
 donut-eval.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/donut-eval.py b/donut-eval.py
index bdf7c93..acc102b 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -15,12 +15,17 @@ from sconf import Config
 
 def main(config):
 
+    image_size = [1920, 2560]
     config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
-    config_vision.encoder.image_size = [1920, 2560] # (height, width)
+    config_vision.encoder.image_size = image_size # (height, width)
     config_vision.decoder.max_length = 768
 
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
     model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config_vision)
+
+    processor.image_processor.size = image_size[::-1] # should be (width, height)
+    processor.image_processor.do_align_long_axis = False
+
     dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)
     device = "cuda" if torch.cuda.is_available() else "cpu"
     model.eval()

From cc2e6541cf2638e8196230eb36cfa44396e03141 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:44:55 +0100
Subject: [PATCH 13/17] test

---
 donut-eval.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/donut-eval.py b/donut-eval.py
index acc102b..5a03d45 100644
--- a/donut-eval.py
+++ b/donut-eval.py
@@ -15,15 +15,15 @@ from sconf import Config
 
 def main(config):
 
-    image_size = [1920, 2560]
-    config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
-    config_vision.encoder.image_size = image_size # (height, width)
-    config_vision.decoder.max_length = 768
+    # image_size = [1920, 2560]
+    # config_vision = VisionEncoderDecoderConfig.from_pretrained(config.pretrained_model_path)
+    # config_vision.encoder.image_size = image_size # (height, width)
+    # config_vision.decoder.max_length = 768
 
     processor = DonutProcessor.from_pretrained(config.pretrained_processor_path)
-    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path, config=config_vision)
+    model = VisionEncoderDecoderModel.from_pretrained(config.pretrained_model_path)
 
-    processor.image_processor.size = image_size[::-1] # should be (width, height)
+    # processor.image_processor.size = image_size[::-1] # should be (width, height)
     processor.image_processor.do_align_long_axis = False
 
     dataset = load_dataset(config.validation_dataset_path, split=config.validation_dataset_split)

From 8cf6a48c552f926f1f806ff5c6dd091e8c43878e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:49:49 +0100
Subject: [PATCH 14/17] proto

---
 config.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/config.yaml b/config.yaml
index 0b2a262..512a31b 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,8 +1,8 @@
-pretrained_processor_path: "naver-clova-ix/donut-base"
-pretrained_model_path: "naver-clova-ix/donut-base"
+pretrained_processor_path: "naver-clova-ix/donut-proto"
+pretrained_model_path: "naver-clova-ix/donut-proto"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False
 print_output: True
 output_file_dir: "../../gonito-outs"
-test_name: "base-test"
\ No newline at end of file
+test_name: "proto-test"
\ No newline at end of file

From 45d3ac7355b7c45cc810aced53f1c1a18fa363e7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Micha=C5=82=20Koz=C5=82owski?=
 <michalkozlowski936@gmail.com>
Date: Sat, 17 Dec 2022 10:52:31 +0100
Subject: [PATCH 15/17] fine

---
 config.yaml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/config.yaml b/config.yaml
index 512a31b..9dc286e 100644
--- a/config.yaml
+++ b/config.yaml
@@ -1,8 +1,8 @@
-pretrained_processor_path: "naver-clova-ix/donut-proto"
-pretrained_model_path: "naver-clova-ix/donut-proto"
+pretrained_processor_path: "Zombely/plwiki-proto-fine-tuned-v2"
+pretrained_model_path: "Zombely/plwiki-proto-fine-tuned-v2"
 validation_dataset_path: "Zombely/diachronia-ocr"
 validation_dataset_split: "train"
 has_metadata: False
 print_output: True
 output_file_dir: "../../gonito-outs"
-test_name: "proto-test"
\ No newline at end of file
+test_name: "fine-tuned-test"
\ No newline at end of file

From 8006c39d83daf47fa9f6ac97479fcbe6e686a302 Mon Sep 17 00:00:00 2001
From: s444415 <mickoz5@st.amu.edu.pl>
Date: Tue, 3 Jan 2023 13:15:05 +0100
Subject: [PATCH 16/17] config edit, dataset create script, donut train config
 edit

---
 config.yaml => config-eval.yaml |   0
 dataset_create.ipynb            | 822 ++++++++++++++++++++++++++++++++
 donut-train.py                  |  12 +-
 3 files changed, 828 insertions(+), 6 deletions(-)
 rename config.yaml => config-eval.yaml (100%)
 create mode 100644 dataset_create.ipynb

diff --git a/config.yaml b/config-eval.yaml
similarity index 100%
rename from config.yaml
rename to config-eval.yaml
diff --git a/dataset_create.ipynb b/dataset_create.ipynb
new file mode 100644
index 0000000..4fbcd73
--- /dev/null
+++ b/dataset_create.ipynb
@@ -0,0 +1,822 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "from huggingface_hub import login\n",
+    "from datasets import load_dataset\n",
+    "import os\n",
+    "import json\n",
+    "import shutil"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f0476002f8d14822a24f1376cfe29a07",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "login(os.environ.get(\"HUG_TOKKEN\"))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "df_train = pd.read_csv('../fiszki-ocr/train/in.tsv', sep='\\t', header=None, index_col=False)\n",
+    "files = [file[0] for file in df_train.iloc()]\n",
+    "df_train_out = pd.read_csv('../fiszki-ocr/train/expected.tsv', sep='\\t', header=None, index_col=False)\n",
+    "files_out = [file_out[0] for file_out in df_train_out.iloc()]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "whole = []\n",
+    "for file, out in zip(files, files_out):\n",
+    "        whole.append({\"file_name\": file, \"ground_truth\": json.dumps({\"gt_parse\": {\"text_sequance\": out}}, ensure_ascii=False)})"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train = whole[:85]\n",
+    "validation = whole[85:]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "train_files = [file.get(\"file_name\") for file in train]\n",
+    "validation_files = [file.get(\"file_name\") for file in validation]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for image in os.listdir(\"../fiszki-ocr/images\"):\n",
+    "    if image in train_files:\n",
+    "        shutil.copy(f\"/home/pc/work/fiszki-ocr/images/{image}\", f\"./images-split-fiszki/train/{image}\")\n",
+    "    if image in validation_files:\n",
+    "        shutil.copy(f\"/home/pc/work/fiszki-ocr/images/{image}\", f\"./images-split-fiszki/validation/{image}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\n",
+    "with open('./images-split-fiszki/train/metadata.jsonl', 'w', encoding='utf-8') as f:\n",
+    "    for entry in train:\n",
+    "        json.dump(entry, f, ensure_ascii=False)\n",
+    "        f.write(\"\\n\")\n",
+    "with open('./images-split-fiszki/validation/metadata.jsonl', 'w', encoding='utf-8') as f:\n",
+    "    for entry in validation:\n",
+    "        json.dump(entry, f, ensure_ascii=False)\n",
+    "        f.write(\"\\n\")\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ca154573c11a44a8a1fa7dede4c54e26",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Resolving data files:   0%|          | 0/86 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Using custom data configuration images-split-fiszki-0b6e02834f7867a1\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Downloading and preparing dataset imagefolder/images-split-fiszki to /home/pc/.cache/huggingface/datasets/imagefolder/images-split-fiszki-0b6e02834f7867a1/0.0.0/37fbb85cc714a338bea574ac6c7d0b5be5aff46c1862c1989b20e0771199e93f...\n",
+      "                "
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "2677f9a18a4d40768ebfee41eb5ee208",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #4:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b742285b54724ef895dc3f1c76510030",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #1:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "27b3c6bbb7fe4220b20a13c6b720b99e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #15:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "b9961292c96c404582fe522ff8d93e1d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #10:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d496ade67a244136b1fe5a00e539dc9f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #6:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "61b0ebdef7814d0ab6f9fa796b67f033",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #14:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9040a6b8a24f4ab793d0cf459f5f35b3",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #2:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e4e6f1800d37456ebc095f7a096082fe",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #7:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "badee192d70a4d109cf38b3539876221",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #8:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "05801dc38fd24f4382f488c8a3fa92bc",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #3:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "704b4bd67b044e9c8d3cb009df4be325",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #0:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ee591babc11e479c8263368893964589",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #9:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "2e6e2f9a00774a6ba35330a0e1104968",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #5:   0%|          | 0/6 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "c97fd7b70b544c068d13eef90ad05127",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #11:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "9d9741e14c7945c4aac512ebe6effbba",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #12:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "f0ec07904f434cf7b8d7e98702979c83",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #13:   0%|          | 0/5 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3881282964584fe8906257ca4edb825b",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files: 0it [00:00, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3d543609550c438c891b36e2406cb1ae",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Extracting data files: 0it [00:00, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "                "
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d48e39b33fb74375894bff21bd91dd56",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #2:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3e8add0aead64b06b4b630a9e3cd7614",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #4:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "7e3311255c414944965ac6d19e3520bb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #3:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "89fe0a3793d0442ab9d91a98e39b05f1",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #0:   0%|          | 0/2 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "e89ba8660b684c028d15b5b62f22c3ba",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #5:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "db6d55a219704ffa8f73a31d928fe47e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #1:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "ca45c0d6589d4b858a9914ef9f8845d4",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #9:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "45ea9e09950d4ec0b0529db382b14d6f",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #10:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "cfeb70d806d344b683aa9e772b468e6e",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #14:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3dd158c57c8c46b19b86ddd7e31915fd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #15:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "367863a5314d494f929aae0ca91e0a33",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #7:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a1a0e3b6a0234188b34c11fae2f6503d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #6:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d8e9df2e9ace4365b3e6faf80c2b7cbb",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #11:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "6df8b58b7a934f8eaf0422ce9f704d38",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #13:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d1f58d8da7f24d6394e2c2ace1372d92",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #8:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "208c67f7ffb64f548726bbe2443f6930",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files #12:   0%|          | 0/1 [00:00<?, ?obj/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "905e84687479471daaadfd9850c52a88",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading data files: 0it [00:00, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "3db88be1336b4ea0b03638761a6d69e7",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Extracting data files: 0it [00:00, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "96090b6f92eb46be8b44dde7d96f225a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Generating train split: 0 examples [00:00, ? examples/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "419401799c864422b9669c66c44159bd",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Generating validation split: 0 examples [00:00, ? examples/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Dataset imagefolder downloaded and prepared to /home/pc/.cache/huggingface/datasets/imagefolder/images-split-fiszki-0b6e02834f7867a1/0.0.0/37fbb85cc714a338bea574ac6c7d0b5be5aff46c1862c1989b20e0771199e93f. Subsequent calls will reuse this data.\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "835ebdd301dc469dbb0ad6f1838403a5",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/2 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "dataset = load_dataset('./images-split-fiszki')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing split train to the Hub.\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "a92c7f7732054b479a26b1f32621cf20",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/1 [00:00<?, ?ba/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "d934bd92d4af41e492ecac230b635903",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Pushing dataset shards to the dataset hub:   0%|          | 0/1 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Pushing split validation to the Hub.\n"
+     ]
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "dd0c40c587e84870ad9c3089d401b80a",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "  0%|          | 0/1 [00:00<?, ?ba/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "eeac023e001349a48133da21f7656378",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Pushing dataset shards to the dataset hub:   0%|          | 0/1 [00:00<?, ?it/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "dataset.push_to_hub(\"Zombely/fiszki-ocr-train\")"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "hug_donut",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.15 (main, Nov  4 2022, 16:13:54) \n[GCC 11.2.0]"
+  },
+  "orig_nbformat": 4,
+  "vscode": {
+   "interpreter": {
+    "hash": "8f1c1b41577d000ca6512e75d22d324bbd1d5e060e99f4f49d98cf0adf636690"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
diff --git a/donut-train.py b/donut-train.py
index 7ba98f0..95dee08 100644
--- a/donut-train.py
+++ b/donut-train.py
@@ -21,17 +21,17 @@ from pytorch_lightning.plugins import CheckpointIO
 
 
 
-DATASET_PATH = "Zombely/pl-text-images-5000-whole"
+DATASET_PATH = "Zombely/fiszki-ocr-train"
 PRETRAINED_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v2"
-START_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned"
-OUTPUT_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v2"
-LOGGING_PATH = "plwiki-proto-ft-second-iter"
+START_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v2"
+OUTPUT_MODEL_PATH = "Zombely/plwiki-proto-fine-tuned-v3"
+LOGGING_PATH = "fiszki-ocr-fine-tune"
 CHECKPOINT_PATH = "./checkpoint"
 
 
 train_config = {
     "max_epochs":1,
-    "val_check_interval":1.0, # how many times we want to validate during an epoch
+    "val_check_interval":0.5, # how many times we want to validate during an epoch
     "check_val_every_n_epoch":1,
     "gradient_clip_val":1.0,
     "num_training_samples_per_epoch": 800,
@@ -362,7 +362,7 @@ custom_ckpt = CustomCheckpointIO()
 
 
 trainer = pl.Trainer(
-        accelerator="gpu", # change to gpu
+        accelerator="gpu" if torch.cuda.is_available() else 'cpu', # change to gpu
         devices=1,
         max_epochs=train_config.get("max_epochs"),
         val_check_interval=train_config.get("val_check_interval"),

From 0de83d1603a5749f9db195425ea3b9daf0b1d262 Mon Sep 17 00:00:00 2001
From: s444415 <mickoz5@st.amu.edu.pl>
Date: Wed, 4 Jan 2023 09:50:56 +0100
Subject: [PATCH 17/17] folder split

---
 dataset_create.ipynb => notepads/dataset_create.ipynb | 0
 donut-eval.ipynb => notepads/donut-eval.ipynb         | 0
 donut-train.ipynb => notepads/donut-train.ipynb       | 0
 3 files changed, 0 insertions(+), 0 deletions(-)
 rename dataset_create.ipynb => notepads/dataset_create.ipynb (100%)
 rename donut-eval.ipynb => notepads/donut-eval.ipynb (100%)
 rename donut-train.ipynb => notepads/donut-train.ipynb (100%)

diff --git a/dataset_create.ipynb b/notepads/dataset_create.ipynb
similarity index 100%
rename from dataset_create.ipynb
rename to notepads/dataset_create.ipynb
diff --git a/donut-eval.ipynb b/notepads/donut-eval.ipynb
similarity index 100%
rename from donut-eval.ipynb
rename to notepads/donut-eval.ipynb
diff --git a/donut-train.ipynb b/notepads/donut-train.ipynb
similarity index 100%
rename from donut-train.ipynb
rename to notepads/donut-train.ipynb