File size: 753 Bytes
f452fcc 0e5a3b5 f452fcc 54c40b3 0e5a3b5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 |
---
license: apache-2.0
base_model: mistralai/Mistral-7B-v0.1
---
# Tensoic - Mistral-7B-v0.1-alpaca-2k-test
Our first fine tune of the Mistral 7B on the Alpaca-2k-test dataset. Feel free to play around!
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
### Framework versions
- Transformers 4.34.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.14.0
|