Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,86 @@
|
|
| 1 |
---
|
| 2 |
license: unknown
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: unknown
|
| 3 |
+
base_model: openthaigpt/openthaigpt-1.0.0-beta-7b-chat-ckpt-hf
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
model-index:
|
| 7 |
+
- name: out
|
| 8 |
+
results: []
|
| 9 |
+
language:
|
| 10 |
+
- th
|
| 11 |
+
pipeline_tag: text-generation
|
| 12 |
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
[<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)
|
| 18 |
+
# ping98k/th-7b-20gb-base
|
| 19 |
+
|
| 20 |
+
This model is a fine-tuned version of [openthaigpt/openthaigpt-1.0.0-beta-7b-chat-ckpt-hf](https://huggingface.co/openthaigpt/openthaigpt-1.0.0-beta-7b-chat-ckpt-hf) on the 20GB Thai dataset.
|
| 21 |
+
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 1.5721
|
| 23 |
+
|
| 24 |
+
## Model description
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Intended uses & limitations
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training and evaluation data
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Training procedure
|
| 37 |
+
|
| 38 |
+
### Training hyperparameters
|
| 39 |
+
|
| 40 |
+
The following hyperparameters were used during training:
|
| 41 |
+
- learning_rate: 0.00015
|
| 42 |
+
- train_batch_size: 4
|
| 43 |
+
- eval_batch_size: 4
|
| 44 |
+
- seed: 42
|
| 45 |
+
- distributed_type: multi-GPU
|
| 46 |
+
- num_devices: 2
|
| 47 |
+
- gradient_accumulation_steps: 4
|
| 48 |
+
- total_train_batch_size: 32
|
| 49 |
+
- total_eval_batch_size: 8
|
| 50 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 51 |
+
- lr_scheduler_type: cosine
|
| 52 |
+
- lr_scheduler_warmup_steps: 100
|
| 53 |
+
- num_epochs: 1
|
| 54 |
+
|
| 55 |
+
### Training results
|
| 56 |
+
|
| 57 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 58 |
+
|:-------------:|:-----:|:-----:|:---------------:|
|
| 59 |
+
| 4.0347 | 0.0 | 1 | 4.0530 |
|
| 60 |
+
| 2.2753 | 0.05 | 1179 | 2.2083 |
|
| 61 |
+
| 2.1613 | 0.1 | 2358 | 2.0422 |
|
| 62 |
+
| 2.0696 | 0.15 | 3537 | 1.9526 |
|
| 63 |
+
| 1.945 | 0.2 | 4716 | 1.8886 |
|
| 64 |
+
| 1.6807 | 0.25 | 5895 | 1.8340 |
|
| 65 |
+
| 1.5838 | 0.3 | 7074 | 1.7961 |
|
| 66 |
+
| 1.7497 | 0.35 | 8253 | 1.7548 |
|
| 67 |
+
| 1.535 | 0.4 | 9432 | 1.7237 |
|
| 68 |
+
| 1.9632 | 0.45 | 10611 | 1.6878 |
|
| 69 |
+
| 1.9091 | 0.5 | 11790 | 1.6631 |
|
| 70 |
+
| 1.6837 | 0.55 | 12969 | 1.6344 |
|
| 71 |
+
| 1.7054 | 0.6 | 14148 | 1.6131 |
|
| 72 |
+
| 1.463 | 0.65 | 15327 | 1.5980 |
|
| 73 |
+
| 1.5538 | 0.7 | 16506 | 1.5853 |
|
| 74 |
+
| 1.5095 | 0.75 | 17685 | 1.5780 |
|
| 75 |
+
| 1.7322 | 0.8 | 18864 | 1.5742 |
|
| 76 |
+
| 1.5645 | 0.85 | 20043 | 1.5727 |
|
| 77 |
+
| 1.72 | 0.9 | 21222 | 1.5722 |
|
| 78 |
+
| 1.5882 | 0.95 | 22401 | 1.5721 |
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
### Framework versions
|
| 82 |
+
|
| 83 |
+
- Transformers 4.35.1
|
| 84 |
+
- Pytorch 2.0.1+cu118
|
| 85 |
+
- Datasets 2.14.7
|
| 86 |
+
- Tokenizers 0.14.1
|