Update README.md
Browse files
README.md
CHANGED
@@ -24,6 +24,7 @@ datasets:
|
|
24 |
- hpprc/mqa-ja
|
25 |
- google-research-datasets/paws-x
|
26 |
base_model: pkshatech/GLuCoSE-base-ja
|
|
|
27 |
---
|
28 |
|
29 |
# SentenceTransformer
|
@@ -42,14 +43,6 @@ This is a [sentence-transformers](https://www.SBERT.net) model trained. It maps
|
|
42 |
<!-- - **Language:** Unknown -->
|
43 |
<!-- - **License:** Unknown -->
|
44 |
|
45 |
-
### Full Model Architecture
|
46 |
-
|
47 |
-
```
|
48 |
-
SentenceTransformer(
|
49 |
-
(0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: LukeModel
|
50 |
-
(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
|
51 |
-
)
|
52 |
-
```
|
53 |
|
54 |
## Usage
|
55 |
|
@@ -132,8 +125,8 @@ You can finetune this model on your own dataset.
|
|
132 |
- Tokenizers: 0.19.1
|
133 |
## Benchmarks
|
134 |
|
135 |
-
|
136 |
-
Evaluated with [MIRACL-ja](https://huggingface.co/datasets/miracl/miracl), [JQARA]
|
137 |
|
138 |
| model | size | MIRACL<br>Recall@5 | JQaRA<br>nDCG@10 | MLDR<br>nDCG@10 |
|
139 |
|--------|--------|---------------------|-------------------|-------------------|
|
@@ -141,8 +134,8 @@ Evaluated with [MIRACL-ja](https://huggingface.co/datasets/miracl/miracl), [JQAR
|
|
141 |
| GLuCoSE | 0.1B | 53.3 | 30.8 | 25.2 |
|
142 |
| GLuCoSE v2 | 0.1B | 85.5 | 60.6 | 33.8 |
|
143 |
|
144 |
-
|
145 |
-
Evaluated with [JMTEB]
|
146 |
* Time-consuming [‘amazon_review_classification’, ‘mrtydi’, ‘jaqket’, ‘esci’] were excluded and evaluated.
|
147 |
* The average is a macro-average per task.
|
148 |
|
@@ -153,11 +146,16 @@ Evaluated with [JMTEB][https://github.com/sbintuitions/JMTEB].
|
|
153 |
| GLuCoSE v2 | 0.1B | 80.5 | 82.8 | 83.0 | 49.8 | 62.4 | 71.7 |
|
154 |
|
155 |
|
156 |
-
##
|
|
|
157 |
|
158 |
-
|
|
|
159 |
|
160 |
<!--
|
|
|
|
|
|
|
161 |
## Glossary
|
162 |
|
163 |
*Clearly define terms in order to be accessible across audiences.*
|
|
|
24 |
- hpprc/mqa-ja
|
25 |
- google-research-datasets/paws-x
|
26 |
base_model: pkshatech/GLuCoSE-base-ja
|
27 |
+
license: apache-2.0
|
28 |
---
|
29 |
|
30 |
# SentenceTransformer
|
|
|
43 |
<!-- - **Language:** Unknown -->
|
44 |
<!-- - **License:** Unknown -->
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
## Usage
|
48 |
|
|
|
125 |
- Tokenizers: 0.19.1
|
126 |
## Benchmarks
|
127 |
|
128 |
+
### Zero-shot Search
|
129 |
+
Evaluated with [MIRACL-ja](https://huggingface.co/datasets/miracl/miracl), [JQARA](https://huggingface.co/datasets/hotchpotch/JQaRA) and [MLDR-ja](https://huggingface.co/datasets/Shitao/MLDR).
|
130 |
|
131 |
| model | size | MIRACL<br>Recall@5 | JQaRA<br>nDCG@10 | MLDR<br>nDCG@10 |
|
132 |
|--------|--------|---------------------|-------------------|-------------------|
|
|
|
134 |
| GLuCoSE | 0.1B | 53.3 | 30.8 | 25.2 |
|
135 |
| GLuCoSE v2 | 0.1B | 85.5 | 60.6 | 33.8 |
|
136 |
|
137 |
+
### JMTEB
|
138 |
+
Evaluated with [JMTEB](https://github.com/sbintuitions/JMTEB).
|
139 |
* Time-consuming [‘amazon_review_classification’, ‘mrtydi’, ‘jaqket’, ‘esci’] were excluded and evaluated.
|
140 |
* The average is a macro-average per task.
|
141 |
|
|
|
146 |
| GLuCoSE v2 | 0.1B | 80.5 | 82.8 | 83.0 | 49.8 | 62.4 | 71.7 |
|
147 |
|
148 |
|
149 |
+
## Authors
|
150 |
+
Chihiro Yano, Go Mocho, Hideyuki Tachibana, Hiroto Takegawa, Yotaro Watanabe
|
151 |
|
152 |
+
## License
|
153 |
+
This model is published under the [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0).
|
154 |
|
155 |
<!--
|
156 |
+
## Citation
|
157 |
+
|
158 |
+
### BibTeX
|
159 |
## Glossary
|
160 |
|
161 |
*Clearly define terms in order to be accessible across audiences.*
|