Commit
•
eff5d43
1
Parent(s):
22c06fc
Model save
Browse files
README.md
CHANGED
|
@@ -14,6 +14,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 14 |
# results_packing
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [hivaze/ParaLex-Llama-3-8B-SFT](https://huggingface.co/hivaze/ParaLex-Llama-3-8B-SFT) on the None dataset.
|
|
|
|
|
|
|
| 17 |
|
| 18 |
## Model description
|
| 19 |
|
|
@@ -40,7 +42,19 @@ The following hyperparameters were used during training:
|
|
| 40 |
- total_train_batch_size: 17
|
| 41 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 42 |
- lr_scheduler_type: cosine
|
| 43 |
-
- num_epochs:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
|
| 45 |
### Framework versions
|
| 46 |
|
|
|
|
| 14 |
# results_packing
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [hivaze/ParaLex-Llama-3-8B-SFT](https://huggingface.co/hivaze/ParaLex-Llama-3-8B-SFT) on the None dataset.
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.3215
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 42 |
- total_train_batch_size: 17
|
| 43 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 44 |
- lr_scheduler_type: cosine
|
| 45 |
+
- num_epochs: 8
|
| 46 |
+
|
| 47 |
+
### Training results
|
| 48 |
+
|
| 49 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 50 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
| 51 |
+
| 0.4022 | 1.25 | 5 | 0.3324 |
|
| 52 |
+
| 0.3492 | 2.5 | 10 | 0.3161 |
|
| 53 |
+
| 0.3181 | 3.75 | 15 | 0.3138 |
|
| 54 |
+
| 0.2808 | 5.0 | 20 | 0.3177 |
|
| 55 |
+
| 0.2571 | 6.25 | 25 | 0.3206 |
|
| 56 |
+
| 0.2424 | 7.5 | 30 | 0.3215 |
|
| 57 |
+
|
| 58 |
|
| 59 |
### Framework versions
|
| 60 |
|