|
--- |
|
license: llama2 |
|
library_name: peft |
|
tags: |
|
- trl |
|
- sft |
|
- generated_from_trainer |
|
base_model: sarvamai/OpenHathi-7B-Hi-v0.1-Base |
|
datasets: |
|
- generator |
|
model-index: |
|
- name: openhathi_instruct |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# openhathi_instruct |
|
|
|
This model is a fine-tuned version of [sarvamai/OpenHathi-7B-Hi-v0.1-Base](https://huggingface.co/sarvamai/OpenHathi-7B-Hi-v0.1-Base) on the generator dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 1.2988 |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 0.0005 |
|
- train_batch_size: 1 |
|
- eval_batch_size: 1 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_ratio: 0.1 |
|
- num_epochs: 10 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| No log | 1.0 | 4 | 2.5575 | |
|
| 2.8332 | 2.0 | 8 | 1.8445 | |
|
| 2.0322 | 3.0 | 12 | 1.6430 | |
|
| 1.6405 | 4.0 | 16 | 1.4686 | |
|
| 1.3921 | 5.0 | 20 | 1.3654 | |
|
| 1.3921 | 6.0 | 24 | 1.3289 | |
|
| 1.1752 | 7.0 | 28 | 1.3054 | |
|
| 1.0246 | 8.0 | 32 | 1.3079 | |
|
| 0.9262 | 9.0 | 36 | 1.2984 | |
|
| 0.852 | 10.0 | 40 | 1.2988 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.11.1 |
|
- Transformers 4.41.2 |
|
- Pytorch 2.1.1+cu121 |
|
- Datasets 2.19.1 |
|
- Tokenizers 0.19.1 |
