|
--- |
|
language: |
|
- multilingual |
|
base_model: /kaggle/input/mistral-7b/Mistral-7B-v0.1 |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- STEM |
|
model-index: |
|
- name: mistral-7b-llm-science-exam |
|
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. --> |
|
|
|
# mistral-7b-llm-science-exam |
|
|
|
This model is a fine-tuned version of [/kaggle/input/mistral-7b/Mistral-7B-v0.1](https://huggingface.co//kaggle/input/mistral-7b/Mistral-7B-v0.1) on the llm-science-exam dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.3951 |
|
- Map@3: 0.8976 |
|
|
|
## 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: 1e-05 |
|
- train_batch_size: 2 |
|
- eval_batch_size: 2 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 4 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_steps: 50 |
|
- num_epochs: 1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Map@3 | |
|
|:-------------:|:-----:|:----:|:---------------:|:------:| |
|
| 3.3769 | 0.11 | 50 | 1.8621 | 0.9238 | |
|
| 1.5772 | 0.23 | 100 | 0.5619 | 0.9119 | |
|
| 0.9202 | 0.34 | 150 | 0.3942 | 0.9095 | |
|
| 0.9485 | 0.45 | 200 | 0.4117 | 0.8976 | |
|
| 0.9698 | 0.56 | 250 | 0.4145 | 0.9048 | |
|
| 0.8731 | 0.68 | 300 | 0.4054 | 0.9048 | |
|
| 0.8929 | 0.79 | 350 | 0.3967 | 0.8976 | |
|
| 0.9737 | 0.9 | 400 | 0.3951 | 0.8976 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.34.0.dev0 |
|
- Pytorch 2.0.0 |
|
- Datasets 2.14.4 |
|
- Tokenizers 0.14.0 |
|
|
