|
--- |
|
license: apache-2.0 |
|
base_model: Locutusque/TinyMistral-248M-v2.5 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: TinyMistral-FFT |
|
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. --> |
|
|
|
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl) |
|
<details><summary>See axolotl config</summary> |
|
|
|
axolotl version: `0.4.0` |
|
```yaml |
|
base_model: Locutusque/TinyMistral-248M-v2.5 |
|
model_type: MistralForCausalLM |
|
is_mistral_derived_model: true |
|
|
|
load_in_8bit: false |
|
load_in_4bit: false |
|
strict: false |
|
|
|
dataset_processes: 20 |
|
|
|
datasets: |
|
- path: epfl-llm/guidelines |
|
type: completion |
|
field: clean_text |
|
- path: JeanKaddour/minipile |
|
type: completion |
|
field: text |
|
|
|
dataset_prepared_path: TinyMistral-FFT-data |
|
val_set_size: 0.001 |
|
output_dir: ./TinyMistral-FFT |
|
|
|
sequence_len: 2048 |
|
sample_packing: false |
|
pad_to_sequence_len: true |
|
|
|
adapter: |
|
lora_model_dir: |
|
lora_r: |
|
lora_alpha: |
|
lora_dropout: |
|
lora_target_linear: |
|
lora_fan_in_fan_out: |
|
|
|
# wandb configuration |
|
wandb_project: TinyMistral-FFT |
|
wandb_watch: |
|
wandb_run_id: |
|
wandb_log_model: |
|
|
|
gradient_accumulation_steps: 2 |
|
micro_batch_size: 4 |
|
num_epochs: 1 |
|
optimizer: paged_adamw_32bit |
|
lr_scheduler: constant |
|
cosine_min_lr_ratio: |
|
|
|
learning_rate: 0.00005 |
|
|
|
train_on_inputs: true |
|
group_by_length: false |
|
bf16: true |
|
fp16: false |
|
tf32: false |
|
|
|
gradient_checkpointing: false |
|
early_stopping_patience: |
|
resume_from_checkpoint: |
|
auto_resume_from_checkpoints: True |
|
local_rank: |
|
logging_steps: 1 |
|
xformers_attention: |
|
flash_attention: true |
|
flash_attn_cross_entropy: false |
|
flash_attn_rms_norm: true |
|
flash_attn_fuse_qkv: false |
|
flash_attn_fuse_mlp: true |
|
|
|
warmup_steps: 10 |
|
evals_per_epoch: 100 |
|
# eval_steps: 10 |
|
eval_table_size: |
|
saves_per_epoch: 50 |
|
debug: |
|
deepspeed: #deepspeed/zero2.json # multi-gpu only |
|
weight_decay: 0 |
|
|
|
# tokens: |
|
|
|
|
|
special_tokens: |
|
bos_token: "<|bos|>" |
|
eos_token: "<|endoftext|>" |
|
unk_token: "<unk>" |
|
``` |
|
|
|
</details><br> |
|
|
|
# TinyMistral-FFT |
|
|
|
This model is a fine-tuned version of [Locutusque/TinyMistral-248M-v2.5](https://huggingface.co/Locutusque/TinyMistral-248M-v2.5) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.9626 |
|
|
|
## 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: 5e-05 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: constant |
|
- lr_scheduler_warmup_steps: 10 |
|
- num_epochs: 1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:------:|:---------------:| |
|
| 4.5414 | 0.0 | 1 | 4.3416 | |
|
| 4.4364 | 0.01 | 1973 | 3.6048 | |
|
| 3.1588 | 0.02 | 3946 | 3.4869 | |
|
| 3.1823 | 0.03 | 5919 | 3.4237 | |
|
| 2.975 | 0.04 | 7892 | 3.3813 | |
|
| 3.2737 | 0.05 | 9865 | 3.3476 | |
|
| 3.7929 | 0.06 | 11838 | 3.3174 | |
|
| 3.3775 | 0.07 | 13811 | 3.2947 | |
|
| 3.6789 | 0.08 | 15784 | 3.2756 | |
|
| 3.4811 | 0.09 | 17757 | 3.2590 | |
|
| 3.3961 | 0.1 | 19730 | 3.2406 | |
|
| 3.4742 | 0.11 | 21703 | 3.2255 | |
|
| 3.5353 | 0.12 | 23676 | 3.2130 | |
|
| 2.5729 | 0.13 | 25649 | 3.2018 | |
|
| 3.0246 | 0.14 | 27622 | 3.1915 | |
|
| 3.5242 | 0.15 | 29595 | 3.1814 | |
|
| 2.6597 | 0.16 | 31568 | 3.1728 | |
|
| 3.0312 | 0.17 | 33541 | 3.1635 | |
|
| 3.2913 | 0.18 | 35514 | 3.1564 | |
|
| 2.8945 | 0.19 | 37487 | 3.1487 | |
|
| 3.2407 | 0.2 | 39460 | 3.1423 | |
|
| 3.076 | 0.21 | 41433 | 3.1358 | |
|
| 3.4588 | 0.22 | 43406 | 3.1298 | |
|
| 3.1972 | 0.23 | 45379 | 3.1236 | |
|
| 2.8544 | 0.24 | 47352 | 3.1182 | |
|
| 2.949 | 0.25 | 49325 | 3.1116 | |
|
| 3.7614 | 0.26 | 51298 | 3.1078 | |
|
| 2.7729 | 0.27 | 53271 | 3.1022 | |
|
| 3.371 | 0.28 | 55244 | 3.0972 | |
|
| 3.1048 | 0.29 | 57217 | 3.0932 | |
|
| 3.0419 | 0.3 | 59190 | 3.0877 | |
|
| 3.0947 | 0.31 | 61163 | 3.0821 | |
|
| 3.4587 | 0.32 | 63136 | 3.0783 | |
|
| 2.8448 | 0.33 | 65109 | 3.0760 | |
|
| 3.3145 | 0.34 | 67082 | 3.0711 | |
|
| 3.1927 | 0.35 | 69055 | 3.0668 | |
|
| 3.3117 | 0.36 | 71028 | 3.0643 | |
|
| 3.2579 | 0.37 | 73001 | 3.0613 | |
|
| 3.1899 | 0.38 | 74974 | 3.0597 | |
|
| 3.0391 | 0.39 | 76947 | 3.0563 | |
|
| 2.6476 | 0.4 | 78920 | 3.0542 | |
|
| 2.9163 | 0.41 | 80893 | 3.0504 | |
|
| 2.4931 | 0.42 | 82866 | 3.0489 | |
|
| 3.3614 | 0.43 | 84839 | 3.0451 | |
|
| 3.1546 | 0.44 | 86812 | 3.0416 | |
|
| 2.8995 | 0.45 | 88785 | 3.0403 | |
|
| 2.8657 | 0.46 | 90758 | 3.0370 | |
|
| 3.4511 | 0.47 | 92731 | 3.0343 | |
|
| 3.2269 | 0.48 | 94704 | 3.0323 | |
|
| 2.6914 | 0.49 | 96677 | 3.0302 | |
|
| 3.087 | 0.5 | 98650 | 3.0282 | |
|
| 3.3036 | 0.51 | 100623 | 3.0266 | |
|
| 3.2269 | 0.52 | 102596 | 3.0251 | |
|
| 3.1237 | 0.53 | 104569 | 3.0223 | |
|
| 2.9733 | 0.54 | 106542 | 3.0197 | |
|
| 3.0594 | 0.55 | 108515 | 3.0186 | |
|
| 2.9842 | 0.56 | 110488 | 3.0168 | |
|
| 3.0986 | 0.57 | 112461 | 3.0158 | |
|
| 3.0296 | 0.58 | 114434 | 3.0141 | |
|
| 3.0091 | 0.59 | 116407 | 3.0139 | |
|
| 2.7111 | 0.6 | 118380 | 3.0107 | |
|
| 3.115 | 0.61 | 120353 | 3.0080 | |
|
| 3.2585 | 0.62 | 122326 | 3.0063 | |
|
| 3.0651 | 0.63 | 124299 | 3.0038 | |
|
| 2.965 | 0.64 | 126272 | 3.0035 | |
|
| 2.9165 | 0.65 | 128245 | 3.0023 | |
|
| 2.8069 | 0.66 | 130218 | 3.0007 | |
|
| 2.9818 | 0.67 | 132191 | 2.9995 | |
|
| 2.8997 | 0.68 | 134164 | 2.9978 | |
|
| 2.948 | 0.69 | 136137 | 2.9966 | |
|
| 3.034 | 0.7 | 138110 | 2.9953 | |
|
| 3.1774 | 0.71 | 140083 | 2.9936 | |
|
| 3.3357 | 0.72 | 142056 | 2.9919 | |
|
| 3.2333 | 0.73 | 144029 | 2.9897 | |
|
| 3.1183 | 0.74 | 146002 | 2.9889 | |
|
| 3.1148 | 0.75 | 147975 | 2.9887 | |
|
| 2.8678 | 0.76 | 149948 | 2.9867 | |
|
| 2.6597 | 0.77 | 151921 | 2.9850 | |
|
| 3.1122 | 0.78 | 153894 | 2.9842 | |
|
| 3.1959 | 0.79 | 155867 | 2.9825 | |
|
| 2.8623 | 0.8 | 157840 | 2.9808 | |
|
| 2.9416 | 0.81 | 159813 | 2.9809 | |
|
| 3.0551 | 0.82 | 161786 | 2.9792 | |
|
| 2.9538 | 0.83 | 163759 | 2.9777 | |
|
| 2.8278 | 0.84 | 165732 | 2.9767 | |
|
| 3.4942 | 0.85 | 167705 | 2.9762 | |
|
| 2.838 | 0.86 | 169678 | 2.9740 | |
|
| 3.0352 | 0.87 | 171651 | 2.9720 | |
|
| 2.8865 | 0.88 | 173624 | 2.9724 | |
|
| 3.0911 | 0.89 | 175597 | 2.9708 | |
|
| 2.8237 | 0.9 | 177570 | 2.9703 | |
|
| 2.9927 | 0.91 | 179543 | 2.9695 | |
|
| 3.2014 | 0.92 | 181516 | 2.9680 | |
|
| 2.3033 | 0.93 | 183489 | 2.9666 | |
|
| 2.6264 | 0.94 | 185462 | 2.9668 | |
|
| 3.1788 | 0.95 | 187435 | 2.9659 | |
|
| 3.066 | 0.96 | 189408 | 2.9645 | |
|
| 2.5523 | 0.97 | 191381 | 2.9640 | |
|
| 2.4562 | 0.98 | 193354 | 2.9630 | |
|
| 3.3801 | 0.99 | 195327 | 2.9626 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.37.0 |
|
- Pytorch 2.0.1+cu117 |
|
- Datasets 2.15.0 |
|
- Tokenizers 0.15.0 |
|
|