cpalenmichel
commited on
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,129 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cc-by-4.0
|
3 |
+
datasets:
|
4 |
+
- bltlab/queryner
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
metrics:
|
8 |
+
- f1
|
9 |
+
pipeline_tag: token-classification
|
10 |
+
---
|
11 |
+
|
12 |
+
# Model Card for Model ID
|
13 |
+
|
14 |
+
E-commerce query segmentation model in English.
|
15 |
+
This model is trained on QueryNER training dataset with the addition of augmentations so the model should be more robust to spelling mistakes and mentions unseen in the training data.
|
16 |
+
|
17 |
+
|
18 |
+
## Model Details
|
19 |
+
|
20 |
+
### Model Description
|
21 |
+
|
22 |
+
This is a token classification model using BERT base uncased as the base model.
|
23 |
+
The model is fine-tuned on the (QueryNER training dataset)[https://huggingface.co/datasets/bltlab/queryner] and augmented data as described in the QueryNER paper.
|
24 |
+
|
25 |
+
|
26 |
+
- **Developed by:** [BLT Lab](https://github.com/bltlab) in collaboration with eBay.
|
27 |
+
- **Funded by:** eBay
|
28 |
+
- **Shared by:** (@cpalenmichel)[https://github.com/cpalenmichel]
|
29 |
+
- **Model type:** Token Classification / Sequence Labeling / Chunking
|
30 |
+
- **Language(s) (NLP):** English
|
31 |
+
- **License:** CC-BY 4.0
|
32 |
+
- **Finetuned from model:** BERT base uncased
|
33 |
+
|
34 |
+
### Model Sources
|
35 |
+
|
36 |
+
Underlying model is based on [BERT base-uncased](https://huggingface.co/google-bert/bert-base-uncased).
|
37 |
+
|
38 |
+
- **Repository:** [https://github.com/bltlab/query-ner](https://github.com/bltlab/query-ner)
|
39 |
+
- **Paper:** Accepted at LREC-COLING Coming soon
|
40 |
+
|
41 |
+
## Uses
|
42 |
+
|
43 |
+
### Direct Use
|
44 |
+
|
45 |
+
Intended use is research purposes and e-commerce query segmentation.
|
46 |
+
|
47 |
+
### Downstream Use
|
48 |
+
|
49 |
+
Potential downstream use cases include weighting entity spans, linking to knowledge bases, removing spans as a recovery strategy for null and low recall queries.
|
50 |
+
|
51 |
+
### Out-of-Scope Use
|
52 |
+
|
53 |
+
This model is trained only on the training data of the QueryNER dataset. It may not perform well on other domains without additional training data and further fine-tuning.
|
54 |
+
|
55 |
+
## Bias, Risks, and Limitations
|
56 |
+
|
57 |
+
See paper limitations section.
|
58 |
+
|
59 |
+
## How to Get Started with the Model
|
60 |
+
|
61 |
+
See huggingface tutorials for token classification and access the model using AutoModelForTokenClassification.
|
62 |
+
Note that we do some post processing to make use of only the first subtoken's tag unlike the inference API.
|
63 |
+
|
64 |
+
## Training Details
|
65 |
+
|
66 |
+
### Training Data
|
67 |
+
|
68 |
+
See paper for details.
|
69 |
+
|
70 |
+
|
71 |
+
### Training Procedure
|
72 |
+
|
73 |
+
See paper for details.
|
74 |
+
|
75 |
+
#### Training Hyperparameters
|
76 |
+
|
77 |
+
See paper for details.
|
78 |
+
|
79 |
+
|
80 |
+
## Evaluation
|
81 |
+
|
82 |
+
Evaluation details provided in the paper.
|
83 |
+
Scoring was done using [SeqScore](https://github.com/bltlab/seqscore) using the conlleval repair method for invalid label transition sequences.
|
84 |
+
|
85 |
+
### Testing Data, Factors & Metrics
|
86 |
+
|
87 |
+
#### Testing Data
|
88 |
+
|
89 |
+
QueryNER test set: [https://huggingface.co/datasets/bltlab/queryner](https://huggingface.co/datasets/bltlab/queryner)
|
90 |
+
|
91 |
+
|
92 |
+
#### Factors
|
93 |
+
Evaluation is reported with micro-F1 at the entity level on the QueryNER test set.
|
94 |
+
We used conlleval repair method for invalid label transitions.
|
95 |
+
|
96 |
+
#### Metrics
|
97 |
+
We use micro-F1 at the entity level as this is fairly common practice for NER models.
|
98 |
+
|
99 |
+
### Results
|
100 |
+
|
101 |
+
[More Information Needed]
|
102 |
+
|
103 |
+
|
104 |
+
## Environmental Impact
|
105 |
+
Rough estimate
|
106 |
+
|
107 |
+
- **Hardware Type:** 1 RTX 3090 GPU
|
108 |
+
- **Hours used:** < 2 hours
|
109 |
+
- **Cloud Provider:** Private
|
110 |
+
- **Compute Region:** northamerica-northeast1
|
111 |
+
- **Carbon Emitted:** 0.02
|
112 |
+
|
113 |
+
|
114 |
+
## Citation
|
115 |
+
|
116 |
+
Accepted at LREC-COLING coming soon
|
117 |
+
|
118 |
+
**BibTeX:**
|
119 |
+
|
120 |
+
Accepted at LREC-COLING coming soon
|
121 |
+
|
122 |
+
|
123 |
+
## Model Card Authors
|
124 |
+
|
125 |
+
Chester Palen-Michel (@cpalenmichel)[https://github.com/cpalenmichel]
|
126 |
+
|
127 |
+
## Model Card Contact
|
128 |
+
|
129 |
+
Chester Palen-Michel (@cpalenmichel)[https://github.com/cpalenmichel]
|