meetween/Llama-speechlmm-1.0-l-TSUM

Model Information

This is the version of meetween/Llama-speechlmm-1.0-l that was fine-tuned for Text Summarization.

License: see LICENSE

Model Architecture

Identical to base model. This model does not include an audio adapter or a video adapter.

This model was obtained by fine-tuning a LoRA on the decoder. This repository contains the weights with the LoRA merged into the main weights.

How to Use

Identical to base model.

Training Data

This model has been fine-tuned on the same AMI and ICSI text summarization data from the training data of the base model.

Evaluation Results

Model Name	Topic Segmentation	Summary of Summaries	ICSI			AutoMin
R-1	R-2	R-L	R-1	R-2	R-L
Base Model	No	No	26.2	3.2	23.3	28.3	2.9	25.9
Base Model	Yes	No	21.9	4.8	20.2	21.3	4.7	20.4
Base Model	Yes	Yes	22.3	2.0	18.9	22.3	2.7	20.6
+LoRA decoder (this model)	No	No	32.1	4.5	24.2	31.4	4.6	29.4
	Yes	No	23.1	6.3	21.7	22.8	6.1	21.7
	Yes	Yes	27.9	3.6	25.8	27.7	3.8	25.9

Framework versions

• Transformers 4.45.0
• Pytorch 2.3.1+cu124.post2
• Datasets 3.2.0
• Tokenizers 0.20.0