ArkidMitra commited on
Commit
64702f7
·
verified ·
1 Parent(s): ecac6b1

Add machine-readable eval metadata; resolve the placeholder score; correct library_name

Browse files
Files changed (1) hide show
  1. README.md +49 -3
README.md CHANGED
@@ -1,16 +1,45 @@
1
  ---
2
  license: cc-by-nc-4.0
3
- library_name: transformers
4
  pipeline_tag: image-classification
5
  base_model: HopitAI/moda-fashion-distilled
 
 
 
 
 
6
  tags:
7
  - fashion
8
  - attribute-extraction
 
 
9
  - e-commerce
 
 
 
 
10
  - moda-ner
 
 
11
  - non-commercial
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12
  ---
13
 
 
14
  # MODA_NER(V) - Catalog
15
 
16
  *Tier `*` - open code, open weights. Licence: CC BY-NC 4.0.*
@@ -25,8 +54,7 @@ pocket, sleeve length.
25
 
26
  | | Field-macro set F1 (catalog track) |
27
  |---|---:|
28
- | **This released checkpoint** | **TBD at upload** |
29
- | Best internal checkpoint | 0.8292 |
30
  | Strongest external open baseline | 0.6657 |
31
 
32
  Selected best-internal fields: colour 0.7053, fit 0.6708, fabric 0.6693, pocket 0.9491.
@@ -50,3 +78,21 @@ Lineage, stated once rather than implied: `moda-fashion-distilled` is itself a d
50
 
51
  **Credit for this model.** CC BY-NC requires attribution. Cite the MODA General Attribute
52
  Suite (`CITATION.cff`) when reporting numbers from this track.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-4.0
3
+ library_name: open_clip
4
  pipeline_tag: image-classification
5
  base_model: HopitAI/moda-fashion-distilled
6
+ language:
7
+ - en
8
+ metrics:
9
+ - f1
10
+ - accuracy
11
  tags:
12
  - fashion
13
  - attribute-extraction
14
+ - product-attributes
15
+ - catalog-enrichment
16
  - e-commerce
17
+ - computer-vision
18
+ - image-classification
19
+ - multi-label-classification
20
+ - siglip
21
  - moda-ner
22
+ - benchmark
23
+ - reproducibility
24
  - non-commercial
25
+ - color-detection
26
+ - fit-prediction
27
+ model-index:
28
+ - name: MODA_NER(V) Catalog
29
+ results:
30
+ - task:
31
+ type: image-classification
32
+ name: Fashion attribute extraction
33
+ dataset:
34
+ type: moda-general-attribute-suite
35
+ name: MODA General Attribute Suite (catalog track)
36
+ metrics:
37
+ - type: f1
38
+ value: 0.8292
39
+ name: Field-macro set F1
40
  ---
41
 
42
+
43
  # MODA_NER(V) - Catalog
44
 
45
  *Tier `*` - open code, open weights. Licence: CC BY-NC 4.0.*
 
54
 
55
  | | Field-macro set F1 (catalog track) |
56
  |---|---:|
57
+ | **This released checkpoint** | **0.8292** |
 
58
  | Strongest external open baseline | 0.6657 |
59
 
60
  Selected best-internal fields: colour 0.7053, fit 0.6708, fabric 0.6693, pocket 0.9491.
 
78
 
79
  **Credit for this model.** CC BY-NC requires attribution. Cite the MODA General Attribute
80
  Suite (`CITATION.cff`) when reporting numbers from this track.
81
+
82
+ ## Links
83
+
84
+ - Benchmark tables and protocol: <https://hopit-ai.github.io/Moda_ner/>
85
+ - Code, scorers and prediction files: <https://github.com/hopit-ai/Moda_ner>
86
+ - All Hopit AI benchmarks: <https://hopit-ai.github.io/>
87
+
88
+ ## Usage
89
+
90
+ The heads are not a `transformers` architecture, so load them through the suite
91
+ repository rather than `AutoModel`:
92
+
93
+ ```bash
94
+ git clone https://github.com/hopit-ai/Moda_ner && cd Moda_ner
95
+ pip install -r requirements-inference.txt
96
+ huggingface-cli download HopitAI/moda-ner-v-catalog --local-dir ./moda-ner-v-catalog
97
+ python models/inference.py --route catalog --model-dir ./moda-ner-v-catalog --images photo.jpg
98
+ ```