Yinkai Wang commited on
Commit
856204a
·
verified ·
1 Parent(s): 4c9f423

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +59 -2
README.md CHANGED
@@ -9,5 +9,62 @@ tags:
9
  - biology
10
  ---
11
 
12
- paper:
13
- - [2601.17204](https://arxiv.org/abs/2501.01950)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  - biology
10
  ---
11
 
12
+ # SpecBridge Adapter Weights (MSGYM / MSnLib / Spectraverse)
13
+
14
+ This Hugging Face repository contains **SpecBridge adapter checkpoints only** (weights for three benchmark datasets).
15
+ Each checkpoint is trained for **spectra → molecule retrieval** by aligning a DreaMS-conditioned spectrum representation to a frozen molecular embedding space (ChemBERTa), as described in the paper below.
16
+
17
+ **Paper (arXiv):** https://arxiv.org/abs/2601.17204
18
+ **DOI:** https://doi.org/10.48550/arXiv.2601.17204
19
+
20
+ ---
21
+
22
+ ## Files in this repo
23
+
24
+ | Dataset | Checkpoint filename |
25
+ |---|---|
26
+ | MassSpecGym (MSGYM) | `SpecBridge_MSGYM_checkpoint.pt` |
27
+ | MSnLib | `SpecBridge_MSnLib_checkpoint.pt` |
28
+ | Spectraverse | `SpecBridge_Spectraverse_checkpoint.pt` |
29
+
30
+ ---
31
+
32
+ ## Download & load weights
33
+
34
+ ### Python (Hugging Face Hub)
35
+
36
+ ```python
37
+ from huggingface_hub import hf_hub_download
38
+ import torch
39
+
40
+ repo_id = "Spony/SpecBridge"
41
+
42
+ ckpt_path = hf_hub_download(
43
+ repo_id=repo_id,
44
+ filename="SpecBridge_MSGYM_checkpoint.pt",
45
+ )
46
+
47
+ state = torch.load(ckpt_path, map_location="cpu")
48
+ print(type(state))
49
+ ```
50
+
51
+ > These checkpoints are intended to be used with the SpecBridge codebase (training/evaluation scripts) and a DreaMS SSL checkpoint.
52
+
53
+ ---
54
+
55
+ ## Citation
56
+
57
+ If you use these weights, please cite:
58
+
59
+ ```bibtex
60
+ @misc{wang2026specbridge,
61
+ title = {SpecBridge: Bridging Mass Spectrometry and Molecular Representations via Cross-Modal Alignment},
62
+ author = {Yinkai Wang and Yan Zhou Chen and Xiaohui Chen and Li-Ping Liu and Soha Hassoun},
63
+ year = {2026},
64
+ eprint = {2601.17204},
65
+ archivePrefix = {arXiv},
66
+ primaryClass = {cs.LG},
67
+ doi = {10.48550/arXiv.2601.17204},
68
+ url = {https://arxiv.org/abs/2601.17204}
69
+ }
70
+ ```