youtube video comment generation
This model is a fine-tuned version of gpt2-large on the US_Video_Comment dataset. It achieves the following results on the evaluation set:
- Loss: 2.0918
- Accuracy: 0.6277
Model description and usage
This model can create a comment on a provided Youtube video title. To test this model you can visit space but most likely it is down due to tariffs of hosting.
You also can check how the model works in the Inference API, just insert the Youtube video title.
Intended uses & limitations
More information needed
Training and evaluation data
Dataset consisted of English Youtube videos titles and comments to them. Dataset link US_Video_Comment
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- training_steps: 2000
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 2.1201 | 1.53 | 500 | 2.1152 | 0.6220 |
| 2.016 | 3.07 | 1000 | 2.0957 | 0.6254 |
| 1.9383 | 4.6 | 1500 | 2.0898 | 0.6271 |
| 1.8823 | 6.14 | 2000 | 2.0918 | 0.6277 |
Framework versions
- Transformers 4.29.0.dev0
- Pytorch 2.0.0-rc1
- Datasets 2.11.0
- Tokenizers 0.13.3
- Downloads last month
- 30
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.