flan-t5-base-samsum / README.md
achimoraites's picture
add more examples
f392da2
|
raw
history blame
5.09 kB
---
license: apache-2.0
tags:
- generated_from_trainer
- summarization
datasets:
- samsum
metrics:
- rouge
widget:
- text: >-
Olivia: Hey Carter, are you still developing that restaurant business? Carter: Hi Olivia Carter: Yes, we want to launch next month :) Olivia: Next month? That's soon! Congrats :) Carter: thanks, I'm a bit nervous but I seriously believe we're delivering something innovative and needed Olivia: I think it's a great concept and I am sure you'll do great! Olivia: I am currently involved with a new restaurant in the city centre Carter: Which one? Olivia: Spicy and chilled Carter: I heard about it :) Is it any good? ;) Olivia: I love the restaurant and really like working there Carter: good for you! Olivia: and here's the question - are you still looking for restaurant to include in your discount app? Carter: sure, but I think it would be better to discuss it in person - would you like to meet up? Olivia: That would be great!
example_title: "Dialogue 1"
- text: >-
Chad: Elton John is goat Eva: what do you mean by goat? Frank: greatest of all time Chad: indeed Eva: ahh... it makes sense now :P
example_title: "Dialogue 2"
- text: >-
Astonishing X-Men is the name of four X-Men comic book series from Marvel Comics, the first two of which were limited series. The third volume, an ongoing series, began in 2004, with its first run written by Joss Whedon and art by John Cassaday. It was then written by Warren Ellis with art by Simone Bianchi and Phil Jimenez.[1] Daniel Way and Christos Gage then took over the title writing alternating stories. They were followed by James Asmus who wrote one issue, then Greg Pak, who took over for four issues in November 2011.[2] Marjorie Liu wrote the final 21 issues of the series until its end at issue #68 in 2013. The title's fourth volume and second ongoing series launched in 2017 during the "ResurrXion" storyline.[3] The first run was written by Charles Soule and illustrated by a rotating cast of artists. Matthew Rosenberg and artist Greg Land would then take over the series before its end in 2018. The original Astonishing X-Men was a four-issue limited series that replaced Uncanny X-Men during the 1995 alternate universe storyline "Age of Apocalypse", in which all X-titles were given new names and issue numbers. In the storyline, Professor X was murdered 20 years in the past by his own son, Legion. Magneto, witnessing his friend's death, committed himself to Xavier's dream and created his own team of X-Men. However, he was unable to prevent the rise of the despotic Apocalypse and hence the series primarily dealt with the X-Men's battle against him. Astonishing X-Men, written by Scott Lobdell and illustrated by Joe Madureira, featured a team of X-Men led by Rogue and consisted of Sunfire, Blink, Morph, Sabretooth and Wildchild. source: https://en.wikipedia.org/wiki/Astonishing_X-Men
example_title: "Wikipedia Article"
model-index:
- name: flan-t5-base-samsum
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: samsum
type: samsum
config: samsum
split: test
args: samsum
metrics:
- name: Rouge1
type: rouge
value: 46.8876
language:
- en
---
<!-- 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. -->
# flan-t5-base-samsum
This model is a fine-tuned version of [google/flan-t5-base](https://huggingface.co/google/flan-t5-base) on the samsum dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3709
- Rouge1: 46.8876
- Rouge2: 23.2689
- Rougel: 39.5369
- Rougelsum: 43.1602
- Gen Len: 17.2027
## 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
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
| 1.4403 | 1.0 | 1842 | 1.3829 | 46.5321 | 23.0912 | 39.4008 | 42.8993 | 17.0977 |
| 1.3534 | 2.0 | 3684 | 1.3732 | 47.1111 | 23.4456 | 39.5462 | 43.2534 | 17.4554 |
| 1.2795 | 3.0 | 5526 | 1.3709 | 46.8876 | 23.2689 | 39.5369 | 43.1602 | 17.2027 |
| 1.2313 | 4.0 | 7368 | 1.3736 | 47.4418 | 23.701 | 39.9856 | 43.6294 | 17.2198 |
| 1.1934 | 5.0 | 9210 | 1.3772 | 47.4656 | 23.9199 | 40.0284 | 43.7039 | 17.3162 |
### Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1+cu116
- Datasets 2.9.0
- Tokenizers 0.13.2