Molecule Captioning On Chebi 20

Metrics

BLEU-2

BLEU-4

METEOR

ROUGE-1

ROUGE-2

ROUGE-L

Text2Mol

Results

Performance results of various models on this benchmark

								Paper Title	Repository
Mol-LLM (Mistral-Instruct-v0.2)	73.2	-	-	-	-	-	-	-	-
Mol-LLM (LLaMA2-Chat)	72.70	-	-	-	-	-	-	-	-
MolReFlect	67.6	60.8	68.0	70.3	57.1	64.4	-	MolReFlect: Towards Fine-grained In-Context Alignment between Molecules and Texts	-
BioT5+	66.6	59.1	68.1	71.0	58.4	65.0	-	BioT5+: Towards Generalized Biological Understanding with IUPAC Integration and Multi-task Tuning
BioT5	63.5	55.6	65.6	69.2	55.9	63.3	60.3	BioT5: Enriching Cross-modal Integration in Biology with Chemical Knowledge and Natural Language Associations
Text+Chem T5-augm-Base	62.5	54.2	64.8	68.2	54.3	62.2	-	Unifying Molecular and Textual Representations via Multi-task Language Modelling
MolCA, Galac1.3B	62.0	53.1	65.1	68.1	53.7	61.8	-	MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter
MolCA, Galac125M	61.6	52.9	63.9	67.4	53.3	61.5	-	MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter
Mol2Lang-VLM	61.2	52.7	63.3	67.4	53.2	61.4	59.8	Mol2Lang-VLM: Vision- and Text-Guided Generative Pre-trained Language Models for Advancing Molecule Captioning through Multimodal Fusion	-
MolReGPT (GPT-4-0314)	60.7	52.5	61.0	63.4	47.6	56.2	58.5	Empowering Molecule Discovery for Molecule-Caption Translation with Large Language Models: A ChatGPT Perspective
LaMolT5-Large	60.2	52.1	63.4	65.5	51.2	59.8	59.7	Automatic Annotation Augmentation Boosts Translation between Molecules and Natural Language
MoMu+MolT5-Large	59.9	51.5	59.7	-	-	-	58.2	A Molecular Multimodal Foundation Model Associating Molecule Graphs with Natural Language
PEIT-GEN	59.8	53.4	67.6	70.0	58.2	65.3	-	Property Enhanced Instruction Tuning for Multi-task Molecule Generation with Large Language Models
MolT5-Large	59.4	50.8	61.4	65.4	51.0	59.4	58.2	Translation between Molecules and Natural Language
MolXPT	59.4	50.5	62.6	66	51.1	59.7	59.4	MolXPT: Wrapping Molecules with Text for Generative Pre-training
Mol-LLM (SELFIES)	58.7	51.5	61.7	62.7	48.7	57.1	-	Mol-LLM: Multimodal Generalist Molecular LLM with Improved Graph Utilization	-
MolFM-Base	58.5	49.8	60.7	65.3	50.8	59.4	57.6	MolFM: A Multimodal Molecular Foundation Model
Text+Chem T5-Base	58	49	60.4	64.7	49.8	58.6	-	Unifying Molecular and Textual Representations via Multi-task Language Modelling
LaMolT5-Base	57.4	48.5	59.6	63.4	47.8	56.4	59.9	Automatic Annotation Augmentation Boosts Translation between Molecules and Natural Language
MolReGPT (GPT-3.5-turbo)	56.5	48.2	62.3	45.0	54.3	58.5	56.0	Empowering Molecule Discovery for Molecule-Caption Translation with Large Language Models: A ChatGPT Perspective

0 of 32 row(s) selected.

Command Palette

Molecule Captioning On Chebi 20

Metrics

Results