---
library_name: transformers
tags: [ ]
model_index:
  - name: Llama-speechlmm-1.0-l-SSUM
    results: [ ]
---

## Model Information

This is the version of [meetween/Llama-speechlmm-1.0-l](https://huggingface.co/meetween/Llama-speechlmm-1.0-l) that was
fine-tuned for Speech Summarization.

**License:** see [LICENSE](LICENSE)

## Model Architecture

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

This model was obtained by fine-tuning the adapter and 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 speech summarization data from the training data of the base
model.

## Evaluation Results

<table>
    <thead>
        <tr>
            <th rowspan=2">Model Name</th>
            <th rowspan=2">Topic<br/>Segmentation</th>
            <th rowspan=2">Summary<br/>of<br/>Summaries</th>
            <th colspan=3>ICSI</th>
        </tr>
        <tr>
            <th>R-1</th>
            <th>R-2</th>
            <th>R-L</th>
        </tr>
    </thead>
    <tbody>
        <tr>
            <td colspan=6>Cascade (Whisper + Textual summ.)</td>
        </tr>
        <tr>
            <td>Base Model</td>
            <td>No</td>
            <td>No</td>
            <td>27.6</td>
            <td>3.8</td>
            <td>25.3</td>
        </tr>
        <tr>
            <td>Base Model</td>
            <td>Yes</td>
            <td>No</td>
            <td>25.9</td>
            <td>5.3</td>
            <td>23.8</td>
        </tr>
        <tr>
            <td>Base Model</td>
            <td>Yes</td>
            <td>Yes</td>
            <td>21.1</td>
            <td>2.3</td>
            <td>18.3</td>
        </tr>
        <tr>
            <td><a href="https://huggingface.co/meetween/Llama-speechlmm-1.0-l-TSUM">meetween/Llama-speechlmm-1.0-l-TSUM</a></td>
            <td>No</td>
            <td>No</td>
            <td>31.0</td>
            <td>4.3</td>
            <td>27.6</td>
        </tr>
        <tr>
            <td>This Model</td>
            <td>No</td>
            <td>No</td>
            <td>28.9</td>
            <td>3.8</td>
            <td>26.0</td>
        </tr>
        <tr style="border-bottom:1px solid black"><td colspan="100%"></td>
        <tr>
            <td colspan=6>end-to-end directly from audio</td>
        </tr>
        <tr>
            <td>Base Model</td>
            <td>N/A</td>
            <td>N/A</td>
            <td>26.6</td>
            <td>3.5</td>
            <td>23.9</td>
        </tr>
        <tr>
            <td>+LoRA decoder</td>
            <td>N/A</td>
            <td>N/A</td>
            <td>27.9</td>
            <td>3.3</td>
            <td>25.2</td>
        </tr>
        <tr>
            <td>+adapter finetune +LoRA decoder (this model)</td>
            <td>No</td>
            <td>No</td>
            <td>32.1</td>
            <td>4.1</td>
            <td>29.1</td>
        </tr>
    </tbody>
</table>

## Framework versions

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