Downloads · 30 days
7
37% of all-time downloads
fp_bs8_lr5e4_x8
damgomz/fp_bs8_lr5e4_x8
fp_bs8_lr5e4_x8 is a fill-mask model from damgomz. Use it when you need the model to fill a missing word. It is set up for transformers.
All-time downloads
19
Public
Parameters
60.8M
973 MB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors243 MB · 99%
At a glance
- Task
- Fill-Mask
- Library
- transformers
- Model type
- albert
- Access
- Public
- Created
- Jul 6, 2024
- Updated
- Jul 6, 2024
- SHA
- ef950913
Try a prompt
Model card
From the Hugging Face model README
Environmental Impact (CODE CARBON DEFAULT)
| Metric | Value |
|---|---|
| Duration (in seconds) | [More Information Needed] |
| Emissions (Co2eq in kg) | [More Information Needed] |
| CPU power (W) | [NO CPU] |
| GPU power (W) | [No GPU] |
| RAM power (W) | [More Information Needed] |
| CPU energy (kWh) | [No CPU] |
| GPU energy (kWh) | [No GPU] |
| RAM energy (kWh) | [More Information Needed] |
| Consumed energy (kWh) | [More Information Needed] |
| Country name | [More Information Needed] |
| Cloud provider | [No Cloud] |
| Cloud region | [No Cloud] |
| CPU count | [No CPU] |
| CPU model | [No CPU] |
| GPU count | [No GPU] |
| GPU model | [No GPU] |
Environmental Impact (for one core)
| Metric | Value |
|---|---|
| CPU energy (kWh) | [No CPU] |
| Emissions (Co2eq in kg) | [More Information Needed] |
Note
5 juillet 2024 !
My Config
| Config | Value |
|---|---|
| checkpoint | albert-base-v2 |
| model_name | fp_bs8_lr5e4_x8 |
| sequence_length | 400 |
| num_epoch | 6 |
| learning_rate | 0.0005 |
| batch_size | 8 |
| weight_decay | 0.0 |
| warm_up_prop | 0.0 |
| drop_out_prob | 0.1 |
| packing_length | 100 |
| train_test_split | 0.2 |
| num_steps | 82101 |
Training and Testing steps
| Epoch | Train Loss | Test Loss |
|---|---|---|
| 0.0 | 14.254172 | 13.220451 |
| 0.5 | 7.311641 | 7.185098 |
| 1.0 | 7.114274 | 7.079700 |
| 1.5 | 7.037040 | 7.022025 |
- Task
- Fill-Mask
- Library
- transformers
- Type
- albert
- Languages
- en
- Created
- Jul 6, 2024
- Updated
- Jul 6, 2024