format
Browse files
README.md
CHANGED
|
@@ -88,7 +88,7 @@ A fine-tuned version of [google/long-t5-tglobal-base](https://huggingface.co/goo
|
|
| 88 |
|
| 89 |
- At the time of writing, the model is not _fully converged_ despite training for 20+ epochs. This checkpoint is serviceable enough (see examples).
|
| 90 |
- I plan to update this page with newer checkpoints and post some metrics over time.
|
| 91 |
-
- Compare performance to [LED-base](https://huggingface.co/pszemraj/led-base-book-summary) trained on the same dataset.
|
| 92 |
|
| 93 |
## Training and evaluation data
|
| 94 |
|
|
@@ -111,7 +111,7 @@ The following hyperparameters were used during the **final** training round\*:
|
|
| 111 |
- lr_scheduler_warmup_ratio: 0.02
|
| 112 |
- num_epochs: 2
|
| 113 |
|
| 114 |
-
\*_Prior training sessions used roughly similar parameters, multiple sessions were required as this takes eons to
|
| 115 |
|
| 116 |
### Training results
|
| 117 |
|
|
|
|
| 88 |
|
| 89 |
- At the time of writing, the model is not _fully converged_ despite training for 20+ epochs. This checkpoint is serviceable enough (see examples).
|
| 90 |
- I plan to update this page with newer checkpoints and post some metrics over time.
|
| 91 |
+
- Compare performance to [LED-base](https://huggingface.co/pszemraj/led-base-book-summary) trained on the same dataset (API gen parameters are the same).
|
| 92 |
|
| 93 |
## Training and evaluation data
|
| 94 |
|
|
|
|
| 111 |
- lr_scheduler_warmup_ratio: 0.02
|
| 112 |
- num_epochs: 2
|
| 113 |
|
| 114 |
+
\*_Prior training sessions used roughly similar parameters, multiple sessions were required as this takes eons to train_
|
| 115 |
|
| 116 |
### Training results
|
| 117 |
|