ailab-bio/PROTAC-Splitter-EncoderDecoder-lr_cosine_restarts-rand-smiles
011
<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
ailab-bio/PROTAC-Splitter-EncoderDecoder-lrcosinerestarts-rand-smiles
This model is a fine-tuned version of seyonec/ChemBERTa-zinc-base-v1 on the ailab-bio/PROTAC-Splitter-Dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.3584
- Linker Has Attachment Point(s): 0.9983
- E3 Valid: 0.9934
- E3 Graph Edit Distance Norm: inf
- E3 Has Attachment Point(s): 0.9934
- E3 Heavy Atoms Difference: 0.1883
- Poi Valid: 0.9567
- Poi Tanimoto Similarity: 0.0
- Has All Attachment Points: 0.9939
- Num Fragments: 2.9997
- Reassembly: 0.6015
- Poi Equal: 0.7927
- Poi Graph Edit Distance: inf
- Linker Graph Edit Distance Norm: inf
- Linker Valid: 0.9983
- Linker Heavy Atoms Difference Norm: 0.0036
- Poi Has Attachment Point(s): 0.9567
- Valid: 0.9493
- Poi Heavy Atoms Difference Norm: 0.0441
- Tanimoto Similarity: 0.0
- Linker Tanimoto Similarity: 0.0
- Linker Graph Edit Distance: inf
- Linker Equal: 0.8512
- Heavy Atoms Difference Norm: 0.0570
- E3 Heavy Atoms Difference Norm: 0.0026
- Heavy Atoms Difference: 4.2382
- Poi Graph Edit Distance Norm: inf
- Linker Heavy Atoms Difference: 0.2012
- Reassembly Nostereo: 0.6357
- E3 Graph Edit Distance: inf
- E3 Tanimoto Similarity: 0.0
- Has Three Substructures: 0.9997
- E3 Equal: 0.8282
- Poi Heavy Atoms Difference: 1.3882
- All Ligands Equal: 0.5946
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- trainbatchsize: 128
- evalbatchsize: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lrschedulertype: cosinewithrestarts
- lrschedulerwarmup_steps: 800
- training_steps: 100000
- mixedprecisiontraining: Native AMP
Training results
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.0
- Tokenizers 0.19.1
