Upload 8 files

.gitattributes

@@ -2,7 +2,6 @@
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.gz filter=lfs diff=lfs merge=lfs -text
 *.h5 filter=lfs diff=lfs merge=lfs -text
@@ -22,10 +21,8 @@
 *.pt filter=lfs diff=lfs merge=lfs -text
 *.pth filter=lfs diff=lfs merge=lfs -text
 *.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
 saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
 *.tflite filter=lfs diff=lfs merge=lfs -text
 *.tgz filter=lfs diff=lfs merge=lfs -text
 *.wasm filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1 @@
+.venv

README.md

@@ -1,3 +1,168 @@
 ---
-license: afl-3.0
+tags:
+- flair
+- token-classification
+- sequence-tagger-model
+- endpoints-template
+language: en
+datasets:
+- shipping_label_ner
+widget:
+- text: "On September 1st George won 1 dollar while watching Game of Thrones."
 ---
+
+# Fork of [flair/ner-english-shipping-labels](https://huggingface.co/flair/ner-english-ontonotes-large)
+
+> This is fork of [flair/ner-english-ontonotes-large](https://huggingface.co/flair/ner-english-ontonotes-large) implementing a custom `handler.py` as an example for how to use `flair` models with [inference-endpoints](https://hf.co/inference-endpoints)
+
+## English NER in Flair (Ontonotes large model)
+
+This is the large 5-class NER model for English that ships with [Flair](https://github.com/flairNLP/flair/).
+
+F1-Score: **77.78** (Ontonotes)
+
+Predicts 6 tags:
+
+| **tag**                        | **meaning** |
+|---------------------------------|-----------|
+| NAME    | Name of person | 
+| ORG         | organizaiton name | 
+| GCNUMBER         | GC tracking number| 
+| BGNUMBER        | BG tracking number | 
+| COUNTRY         | Country name | 
+| LOCATION         | city and picode | 
+
+Based on document-level XLM-R embeddings and [FLERT](https://arxiv.org/pdf/2011.06993v1.pdf/).
+
+---
+
+### Demo: How to use in Flair
+
+Requires: **[Flair](https://github.com/flairNLP/flair/)** (`pip install flair`)
+
+```python
+from flair.data import Sentence
+from flair.models import SequenceTagger
+
+# load tagger
+tagger = SequenceTagger.load("flair/ner-english-ontonotes-large")
+
+# make example sentence
+sentence = Sentence("On September 1st George won 1 dollar while watching Game of Thrones.")
+
+# predict NER tags
+tagger.predict(sentence)
+
+# print sentence
+print(sentence)
+
+# print predicted NER spans
+print('The following NER tags are found:')
+# iterate over entities and print
+for entity in sentence.get_spans('ner'):
+    print(entity)
+
+```
+
+This yields the following output:
+```
+Span [2,3]: "September 1st"   [− Labels: DATE (1.0)]
+Span [4]: "George"   [− Labels: PERSON (1.0)]
+Span [6,7]: "1 dollar"   [− Labels: MONEY (1.0)]
+Span [10,11,12]: "Game of Thrones"   [− Labels: WORK_OF_ART (1.0)]
+```
+
+So, the entities "*September 1st*" (labeled as a **date**), "*George*" (labeled as a **person**), "*1 dollar*" (labeled as a **money**) and "Game of Thrones" (labeled as a **work of art**) are found in the sentence "*On September 1st George Washington won 1 dollar while watching Game of Thrones*". 
+
+
+---
+
+### Training: Script to train this model
+
+The following Flair script was used to train this model: 
+
+```python
+from flair.data import Corpus
+from flair.datasets import ColumnCorpus
+from flair.embeddings import WordEmbeddings, StackedEmbeddings, FlairEmbeddings
+
+# 1. load the corpus (Ontonotes does not ship with Flair, you need to download and reformat into a column format yourself)
+corpus: Corpus = ColumnCorpus(
+                "resources/tasks/onto-ner",
+                column_format={0: "text", 1: "pos", 2: "upos", 3: "ner"},
+                tag_to_bioes="ner",
+            )
+
+# 2. what tag do we want to predict?
+tag_type = 'ner'
+
+# 3. make the tag dictionary from the corpus
+tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
+
+# 4. initialize fine-tuneable transformer embeddings WITH document context
+from flair.embeddings import TransformerWordEmbeddings
+
+embeddings = TransformerWordEmbeddings(
+    model='xlm-roberta-large',
+    layers="-1",
+    subtoken_pooling="first",
+    fine_tune=True,
+    use_context=True,
+)
+
+# 5. initialize bare-bones sequence tagger (no CRF, no RNN, no reprojection)
+from flair.models import SequenceTagger
+
+tagger = SequenceTagger(
+    hidden_size=256,
+    embeddings=embeddings,
+    tag_dictionary=tag_dictionary,
+    tag_type='ner',
+    use_crf=False,
+    use_rnn=False,
+    reproject_embeddings=False,
+)
+
+# 6. initialize trainer with AdamW optimizer
+from flair.trainers import ModelTrainer
+
+trainer = ModelTrainer(tagger, corpus, optimizer=torch.optim.AdamW)
+
+# 7. run training with XLM parameters (20 epochs, small LR)
+from torch.optim.lr_scheduler import OneCycleLR
+
+trainer.train('resources/taggers/ner-english-ontonotes-large',
+              learning_rate=5.0e-6,
+              mini_batch_size=4,
+              mini_batch_chunk_size=1,
+              max_epochs=20,
+              scheduler=OneCycleLR,
+              embeddings_storage_mode='none',
+              weight_decay=0.,
+              )
+```
+
+
+
+---
+
+### Cite
+
+Please cite the following paper when using this model.
+
+```
+@misc{schweter2020flert,
+    title={FLERT: Document-Level Features for Named Entity Recognition},
+    author={Stefan Schweter and Alan Akbik},
+    year={2020},
+    eprint={2011.06993},
+    archivePrefix={arXiv},
+    primaryClass={cs.CL}
+}
+```
+
+---
+
+### Issues?
+
+The Flair issue tracker is available [here](https://github.com/flairNLP/flair/issues/).

handler.py

@@ -0,0 +1,46 @@
+from typing import Any, Dict, List
+import os
+from flair.data import Sentence
+from flair.models import SequenceTagger
+
+class EndpointHandler():
+    def __init__(
+        self,
+        path: str,
+    ):
+        self.tagger = SequenceTagger.load(os.path.join(path,"model.bin"))
+
+    def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
+        """
+        Args:
+            inputs (:obj:`str`):
+                a string containing some text
+        Return:
+            A :obj:`list`:. The object returned should be like [{"entity_group": "XXX", "word": "some word", "start": 3, "end": 6, "score": 0.82}] containing :
+                - "entity_group": A string representing what the entity is.
+                - "word": A substring of the original string that was detected as an entity.
+                - "start": the offset within `input` leading to `answer`. context[start:stop] == word
+                - "end": the ending offset within `input` leading to `answer`. context[start:stop] === word
+                - "score": A score between 0 and 1 describing how confident the model is for this entity.
+        """
+        inputs = data.pop("inputs", data)
+        sentence: Sentence = Sentence(inputs)
+
+        # Also show scores for recognized NEs
+        self.tagger.predict(sentence, label_name="predicted")
+
+        entities = []
+        for span in sentence.get_spans("predicted"):
+            if len(span.tokens) == 0:
+                continue
+            current_entity = {
+                "entity_group": span.tag,
+                "word": span.text,
+                "start": span.tokens[0].start_position,
+                "end": span.tokens[-1].end_position,
+                "score": span.score,
+            }
+
+            entities.append(current_entity)
+
+        return entities

loss.tsv

@@ -0,0 +1,121 @@
+EPOCH	TIMESTAMP	LEARNING_RATE	TRAIN_LOSS
+1	08:17:20	0.1000	3.6498
+2	08:17:20	0.1000	2.6736
+3	08:17:20	0.1000	2.5362
+4	08:17:20	0.1000	2.2586
+5	08:17:20	0.1000	2.3726
+6	08:17:20	0.1000	2.1890
+7	08:17:21	0.1000	2.1492
+8	08:17:21	0.1000	2.0981
+9	08:17:21	0.1000	1.9099
+10	08:17:21	0.1000	1.7786
+11	08:17:21	0.1000	1.5913
+12	08:17:21	0.1000	1.5533
+13	08:17:22	0.1000	1.5620
+14	08:17:22	0.1000	1.5314
+15	08:17:22	0.1000	1.3500
+16	08:17:22	0.1000	1.3900
+17	08:17:23	0.1000	1.3232
+18	08:17:23	0.1000	1.1288
+19	08:17:23	0.1000	1.1317
+20	08:17:23	0.1000	1.0139
+21	08:17:23	0.1000	0.9761
+22	08:17:23	0.1000	1.2072
+23	08:17:23	0.1000	0.9009
+24	08:17:24	0.1000	0.8894
+25	08:17:24	0.1000	0.9252
+26	08:17:24	0.1000	0.9054
+27	08:17:24	0.1000	0.7883
+28	08:17:24	0.1000	0.8084
+29	08:17:24	0.1000	0.7761
+30	08:17:24	0.1000	0.7852
+31	08:17:25	0.1000	0.7688
+32	08:17:25	0.1000	0.7836
+33	08:17:25	0.1000	0.7338
+34	08:17:25	0.1000	0.6533
+35	08:17:25	0.1000	0.6309
+36	08:17:25	0.1000	0.6480
+37	08:17:25	0.1000	0.5591
+38	08:17:26	0.1000	0.5507
+39	08:17:26	0.1000	0.6041
+40	08:17:26	0.1000	0.6510
+41	08:17:26	0.1000	0.5804
+42	08:17:26	0.1000	0.5140
+43	08:17:26	0.1000	0.4635
+44	08:17:26	0.1000	0.4680
+45	08:17:27	0.1000	0.6061
+46	08:17:27	0.1000	0.5622
+47	08:17:27	0.1000	0.4827
+48	08:17:27	0.0500	0.3977
+49	08:17:27	0.0500	0.4279
+50	08:17:27	0.0500	0.3961
+51	08:17:27	0.0500	0.3775
+52	08:17:28	0.0500	0.3889
+53	08:17:28	0.0500	0.3675
+54	08:17:28	0.0500	0.3871
+55	08:17:28	0.0500	0.3514
+56	08:17:28	0.0500	0.3670
+57	08:17:28	0.0500	0.3303
+58	08:17:29	0.0500	0.3669
+59	08:17:29	0.0500	0.3577
+60	08:17:29	0.0500	0.3258
+61	08:17:29	0.0500	0.3332
+62	08:17:29	0.0500	0.3685
+63	08:17:29	0.0500	0.3106
+64	08:17:30	0.0500	0.2856
+65	08:17:30	0.0500	0.2952
+66	08:17:30	0.0500	0.2865
+67	08:17:30	0.0500	0.2844
+68	08:17:30	0.0500	0.2515
+69	08:17:31	0.0500	0.3155
+70	08:17:31	0.0500	0.2477
+71	08:17:31	0.0500	0.3029
+72	08:17:31	0.0500	0.2660
+73	08:17:31	0.0500	0.3264
+74	08:17:31	0.0500	0.2799
+75	08:17:32	0.0250	0.2766
+76	08:17:32	0.0250	0.2247
+77	08:17:32	0.0250	0.2453
+78	08:17:32	0.0250	0.2271
+79	08:17:32	0.0250	0.2329
+80	08:17:32	0.0250	0.2389
+81	08:17:33	0.0125	0.2508
+82	08:17:33	0.0125	0.2192
+83	08:17:33	0.0125	0.2158
+84	08:17:33	0.0125	0.2222
+85	08:17:33	0.0125	0.2167
+86	08:17:33	0.0125	0.2078
+87	08:17:33	0.0125	0.2183
+88	08:17:34	0.0125	0.2368
+89	08:17:34	0.0125	0.1910
+90	08:17:34	0.0125	0.1775
+91	08:17:34	0.0125	0.2070
+92	08:17:34	0.0125	0.1957
+93	08:17:34	0.0125	0.2066
+94	08:17:34	0.0125	0.1958
+95	08:17:35	0.0063	0.1911
+96	08:17:35	0.0063	0.1824
+97	08:17:35	0.0063	0.1891
+98	08:17:35	0.0063	0.2174
+99	08:17:35	0.0031	0.2056
+100	08:17:35	0.0031	0.1873
+101	08:17:36	0.0031	0.1985
+102	08:17:36	0.0031	0.1865
+103	08:17:36	0.0016	0.1874
+104	08:17:36	0.0016	0.2153
+105	08:17:36	0.0016	0.2044
+106	08:17:36	0.0016	0.1859
+107	08:17:36	0.0008	0.2098
+108	08:17:37	0.0008	0.2072
+109	08:17:37	0.0008	0.1640
+110	08:17:37	0.0008	0.2128
+111	08:17:37	0.0008	0.1954
+112	08:17:37	0.0008	0.1887
+113	08:17:37	0.0008	0.1824
+114	08:17:37	0.0004	0.2271
+115	08:17:38	0.0004	0.1896
+116	08:17:38	0.0004	0.1872
+117	08:17:38	0.0004	0.2119
+118	08:17:38	0.0002	0.2057
+119	08:17:38	0.0002	0.1747
+120	08:17:38	0.0002	0.1977

model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c71dfca8fc533690b01b75ddf3613d10b2f64511f97b57d29ef98124023f79f3
+size 408136407

requirements.txt

@@ -0,0 +1 @@
+flair==0.13.1