Hugging Face's logo

scrollprize
/
PHerc.1667-iteration-2

Image Segmentation
Transformers
Safetensors
inkdetection_resnet3d
feature-extraction
vesuvius-challenge
ink-detection
herculaneum
resnet3d
u-net
3d-segmentation
volumetric-imaging
custom_code
Model card Files
xet
 Community
YoussefMoNader commited on
Commit
62fc291
·
verified ·
1 Parent(s): 04b9ad0

Update model card: rename to PHerc.1667 iteration N

Browse files
Files changed (1) hide show
  1. README.md +9 -9
README.md CHANGED
@@ -12,7 +12,7 @@ tags:
12
  pipeline_tag: image-segmentation
13
  ---
14
 
15
- # vesuvius-inkdetection-ink2
16
 
17
  > **Trained on segment l_2 with l_2_inklabels2.png (8,970 tiles).**
18
 
@@ -57,7 +57,7 @@ import torch
57
  from transformers import AutoModel
58
 
59
  model = AutoModel.from_pretrained(
60
- "YoussefMoNader/vesuvius-inkdetection-ink2",
61
  trust_remote_code=True,
62
  ).eval().cuda()
63
 
@@ -86,7 +86,7 @@ import torch.nn.functional as F
86
  from transformers import AutoModel
87
 
88
  model = AutoModel.from_pretrained(
89
- "YoussefMoNader/vesuvius-inkdetection-ink2", trust_remote_code=True,
90
  ).eval().cuda()
91
 
92
  WINDOW, STRIDE = 256, 128 # 128 = 2x oversample; 64 for 8x oversample
@@ -164,12 +164,12 @@ optimizer / scheduler state.
164
 
165
  | model | training segment(s) | label | tiles | effective epochs |
166
  |-------|---------------------|-------|-------|-------------------|
167
- | [`vesuvius-inkdetection-ink0`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink0) | 500p2a + 658 + 20250910185200 + 20250919125754* | (cross-segment baseline) | 20,075 | ~5 |
168
- | [`vesuvius-inkdetection-ink1`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink1) | `l_2` | `l_2_inklabels.png` | 3,396 | ~30 |
169
- | [`vesuvius-inkdetection-ink2`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink2) | `l_2` | `l_2_inklabels2.png` | 8,970 | ~12 |
170
- | [`vesuvius-inkdetection-ink3`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink3) | `l_2` | `l_2_inklabels3.png` | 15,286 | ~7 |
171
- | [`vesuvius-inkdetection-ink4`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink4) | `l_2` | `l_2_inklabels4.png` | 24,773 | ~5 |
172
- | [`vesuvius-inkdetection-ink5`](https://huggingface.co/YoussefMoNader/vesuvius-inkdetection-ink5) | `l_2` | `l_2_inklabels5.png` | 33,061 | 3 |
173
 
174
  All six share the architecture, hyperparameters, and a fixed step
175
  budget of 12,396 optimizer steps; the only thing that varies between
 
12
  pipeline_tag: image-segmentation
13
  ---
14
 
15
+ # PHerc.1667-iteration-2
16
 
17
  > **Trained on segment l_2 with l_2_inklabels2.png (8,970 tiles).**
18
 
 
57
  from transformers import AutoModel
58
 
59
  model = AutoModel.from_pretrained(
60
+ "YoussefMoNader/PHerc.1667-iteration-2",
61
  trust_remote_code=True,
62
  ).eval().cuda()
63
 
 
86
  from transformers import AutoModel
87
 
88
  model = AutoModel.from_pretrained(
89
+ "YoussefMoNader/PHerc.1667-iteration-2", trust_remote_code=True,
90
  ).eval().cuda()
91
 
92
  WINDOW, STRIDE = 256, 128 # 128 = 2x oversample; 64 for 8x oversample
 
164
 
165
  | model | training segment(s) | label | tiles | effective epochs |
166
  |-------|---------------------|-------|-------|-------------------|
167
+ | [`PHerc.1667-iteration-0`](https://huggingface.co/scrollprize/PHerc.1667-iteration-0) | 500p2a + 658 + 20250910185200 + 20250919125754* | (cross-segment baseline) | 20,075 | ~5 |
168
+ | [`PHerc.1667-iteration-1`](https://huggingface.co/scrollprize/PHerc.1667-iteration-1) | `l_2` | `l_2_inklabels.png` | 3,396 | ~30 |
169
+ | [`PHerc.1667-iteration-2`](https://huggingface.co/scrollprize/PHerc.1667-iteration-2) | `l_2` | `l_2_inklabels2.png` | 8,970 | ~12 |
170
+ | [`PHerc.1667-iteration-3`](https://huggingface.co/scrollprize/PHerc.1667-iteration-3) | `l_2` | `l_2_inklabels3.png` | 15,286 | ~7 |
171
+ | [`PHerc.1667-iteration-4`](https://huggingface.co/scrollprize/PHerc.1667-iteration-4) | `l_2` | `l_2_inklabels4.png` | 24,773 | ~5 |
172
+ | [`PHerc.1667-iteration-5`](https://huggingface.co/scrollprize/PHerc.1667-iteration-5) | `l_2` | `l_2_inklabels5.png` | 33,061 | 3 |
173
 
174
  All six share the architecture, hyperparameters, and a fixed step
175
  budget of 12,396 optimizer steps; the only thing that varies between