Downloads · 30 days
6
27% of all-time downloads
damgomz/fp_bs8_lr2e4_x4
fp_bs8_lr2e4_x4 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.
Downloads · 30 days
6
27% of all-time downloads
All-time downloads
22
Public
Parameters
32.5M
1.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors130 MB · 98%
From the Hugging Face model README
| Metric | Value |
|---|---|
| Duration (in seconds) | 237362.58397865295 |
| Emissions (Co2eq in kg) | 0.1785690373864647 |
| CPU power (W) | 42.5 |
| GPU power (W) | [No GPU] |
| RAM power (W) | 15.0 |
| CPU energy (kWh) | 2.802192151747148 |
| GPU energy (kWh) | [No GPU] |
| RAM energy (kWh) | 0.9890020349254236 |
| Consumed energy (kWh) | 3.791194186672571 |
| Country name | Switzerland |
| Cloud provider | nan |
| Cloud region | nan |
| CPU count | 6 |
| CPU model | Intel(R) Xeon(R) Platinum 8360Y CPU @ 2.40GHz |
| GPU count | nan |
| GPU model | nan |
| Metric | Value |
|---|---|
| CPU energy (kWh) | 0.456922974158907 |
| Emissions (Co2eq in kg) | 0.09296701205830574 |
5 juillet 2024 !
| Config | Value |
|---|---|
| checkpoint | albert-base-v2 |
| model_name | fp_bs8_lr2e4_x4 |
| sequence_length | 400 |
| num_epoch | 6 |
| learning_rate | 0.0002 |
| 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 | 82664 |
| Epoch | Train Loss | Test Loss |
|---|---|---|
| 0.0 | 14.987936 | 11.931193 |
| 0.5 | 7.140273 | 7.057710 |
| 1.0 | 7.035364 | 7.019231 |
| 1.5 | 7.000329 | 7.007203 |
| 2.0 | 6.982339 | 6.987988 |
| 2.5 | 6.980809 | 6.980021 |
| 3.0 | 6.973144 | 6.982472 |
| 3.5 | 6.966241 | 6.977357 |
| 4.0 | 6.965905 | 6.969510 |
| 4.5 | 6.957660 | 6.966595 |
| 5.0 | 6.968893 | 6.967915 |
| 5.5 | 6.962095 | 6.962197 |
| 6.0 | 6.951328 | 6.962886 |