65 lines
1.7 KiB
Markdown
65 lines
1.7 KiB
Markdown
---
|
|
license: apache-2.0
|
|
tags:
|
|
- generated_from_trainer
|
|
metrics:
|
|
- accuracy
|
|
model-index:
|
|
- name: python-bytes-distilgpt2
|
|
results: []
|
|
widget:
|
|
- text: "fastAPI is a great new web framework to easily build web APIs."
|
|
example_title: "fastAPI"
|
|
- text: "The new tool uses sqlite for performance"
|
|
example_title: "sqlite"
|
|
---
|
|
|
|
<!-- 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. -->
|
|
|
|
# python-bytes-distilgpt2
|
|
|
|
This model is not affiliated with the Python Bytes podcast in any way.
|
|
|
|
This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on [Python Bytes show notes](https://github.com/mikeckennedy/python_bytes_show_notes/tree/master/transcripts).
|
|
|
|
It achieves the following results on the evaluation set:
|
|
- Loss: 3.0372
|
|
- Accuracy: 0.3969
|
|
|
|
## Model description
|
|
|
|
This model generates conversation between the two show hosts (Michael Kennedy and Brian Okken), and sometimes guests appear :).
|
|
|
|
## Intended uses & limitations
|
|
|
|
This model was trained specifically for educational purposes and is intended for other users to use it in a similar manner.
|
|
|
|
## Training and evaluation data
|
|
|
|
Data is located [on GitHub](https://github.com/mikeckennedy/python_bytes_show_notes/tree/master/transcripts)
|
|
|
|
## Training procedure
|
|
|
|
### Training hyperparameters
|
|
|
|
The following hyperparameters were used during training:
|
|
- learning_rate: 5e-05
|
|
- train_batch_size: 4
|
|
- eval_batch_size: 4
|
|
- seed: 42
|
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
|
- lr_scheduler_type: linear
|
|
- num_epochs: 3.0
|
|
|
|
### Training results
|
|
|
|
|
|
|
|
### Framework versions
|
|
|
|
- Transformers 4.22.0.dev0
|
|
- Pytorch 1.12.0+cu113
|
|
- Datasets 2.4.0
|
|
- Tokenizers 0.12.1
|