Model save
Browse files
README.md
CHANGED
|
@@ -1,10 +1,9 @@
|
|
| 1 |
---
|
| 2 |
-
library_name:
|
| 3 |
license: llama3
|
| 4 |
-
base_model: meta-llama/Meta-Llama-3-
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
| 7 |
-
- full
|
| 8 |
- generated_from_trainer
|
| 9 |
model-index:
|
| 10 |
- name: filtered_construction_train_data
|
|
@@ -16,9 +15,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# filtered_construction_train_data
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
@@ -37,15 +36,15 @@ More information needed
|
|
| 37 |
### Training hyperparameters
|
| 38 |
|
| 39 |
The following hyperparameters were used during training:
|
| 40 |
-
- learning_rate:
|
| 41 |
- train_batch_size: 1
|
| 42 |
-
- eval_batch_size:
|
| 43 |
- seed: 42
|
| 44 |
- distributed_type: multi-GPU
|
| 45 |
- num_devices: 8
|
| 46 |
- gradient_accumulation_steps: 16
|
| 47 |
- total_train_batch_size: 128
|
| 48 |
-
- total_eval_batch_size:
|
| 49 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 50 |
- lr_scheduler_type: cosine
|
| 51 |
- lr_scheduler_warmup_ratio: 0.1
|
|
@@ -55,12 +54,14 @@ The following hyperparameters were used during training:
|
|
| 55 |
|
| 56 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 57 |
|:-------------:|:------:|:----:|:---------------:|
|
| 58 |
-
| 0.
|
|
|
|
| 59 |
|
| 60 |
|
| 61 |
### Framework versions
|
| 62 |
|
|
|
|
| 63 |
- Transformers 4.49.0
|
| 64 |
- Pytorch 2.5.1+cu124
|
| 65 |
- Datasets 3.2.0
|
| 66 |
-
- Tokenizers 0.21.0
|
|
|
|
| 1 |
---
|
| 2 |
+
library_name: peft
|
| 3 |
license: llama3
|
| 4 |
+
base_model: meta-llama/Meta-Llama-3-70B-Instruct
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
|
|
|
| 7 |
- generated_from_trainer
|
| 8 |
model-index:
|
| 9 |
- name: filtered_construction_train_data
|
|
|
|
| 15 |
|
| 16 |
# filtered_construction_train_data
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-70B-Instruct) on the None dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.3455
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 36 |
### Training hyperparameters
|
| 37 |
|
| 38 |
The following hyperparameters were used during training:
|
| 39 |
+
- learning_rate: 0.0001
|
| 40 |
- train_batch_size: 1
|
| 41 |
+
- eval_batch_size: 1
|
| 42 |
- seed: 42
|
| 43 |
- distributed_type: multi-GPU
|
| 44 |
- num_devices: 8
|
| 45 |
- gradient_accumulation_steps: 16
|
| 46 |
- total_train_batch_size: 128
|
| 47 |
+
- total_eval_batch_size: 8
|
| 48 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 49 |
- lr_scheduler_type: cosine
|
| 50 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 54 |
|
| 55 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 56 |
|:-------------:|:------:|:----:|:---------------:|
|
| 57 |
+
| 0.2838 | 0.4203 | 30 | 0.3709 |
|
| 58 |
+
| 0.2593 | 0.8406 | 60 | 0.3455 |
|
| 59 |
|
| 60 |
|
| 61 |
### Framework versions
|
| 62 |
|
| 63 |
+
- PEFT 0.12.0
|
| 64 |
- Transformers 4.49.0
|
| 65 |
- Pytorch 2.5.1+cu124
|
| 66 |
- Datasets 3.2.0
|
| 67 |
+
- Tokenizers 0.21.0
|