---
license: other
base_model: liswei/OpenELM-270M-zh-cp
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: zh-sft
  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. -->

# zh-sft

This model is a fine-tuned version of [saves/OpenELM-270M/zh-cp-galore](https://huggingface.co/saves/OpenELM-270M/zh-cp-galore) on the en2tw-alignment-sft and the TaiwanChat datasets.
It achieves the following results on the evaluation set:
- Loss: 1.5575

## 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: 0.0001
- train_batch_size: 2
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 16
- total_train_batch_size: 64
- 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: 1000
- num_epochs: 3.0

### Training results

| Training Loss | Epoch  | Step  | Validation Loss |
|:-------------:|:------:|:-----:|:---------------:|
| 2.285         | 0.0588 | 500   | 2.2416          |
| 2.0921        | 0.1176 | 1000  | 2.1271          |
| 2.1212        | 0.1764 | 1500  | 2.0457          |
| 1.9794        | 0.2351 | 2000  | 1.9954          |
| 1.8983        | 0.2939 | 2500  | 1.9546          |
| 1.8976        | 0.3527 | 3000  | 1.9214          |
| 1.9345        | 0.4115 | 3500  | 1.8950          |
| 1.8782        | 0.4703 | 4000  | 1.8705          |
| 1.806         | 0.5291 | 4500  | 1.8493          |
| 1.8282        | 0.5878 | 5000  | 1.8275          |
| 1.7949        | 0.6466 | 5500  | 1.8115          |
| 1.7408        | 0.7054 | 6000  | 1.7943          |
| 1.6978        | 0.7642 | 6500  | 1.7782          |
| 1.7152        | 0.8230 | 7000  | 1.7644          |
| 1.7186        | 0.8818 | 7500  | 1.7511          |
| 1.6821        | 0.9406 | 8000  | 1.7357          |
| 1.6238        | 0.9993 | 8500  | 1.7211          |
| 1.4753        | 1.0581 | 9000  | 1.7177          |
| 1.4412        | 1.1169 | 9500  | 1.7048          |
| 1.4273        | 1.1757 | 10000 | 1.6991          |
| 1.4464        | 1.2345 | 10500 | 1.6840          |
| 1.4484        | 1.2933 | 11000 | 1.6749          |
| 1.4752        | 1.3520 | 11500 | 1.6666          |
| 1.4023        | 1.4108 | 12000 | 1.6602          |
| 1.3717        | 1.4696 | 12500 | 1.6467          |
| 1.411         | 1.5284 | 13000 | 1.6376          |
| 1.41          | 1.5872 | 13500 | 1.6298          |
| 1.4263        | 1.6460 | 14000 | 1.6193          |
| 1.3655        | 1.7048 | 14500 | 1.6108          |
| 1.3813        | 1.7635 | 15000 | 1.6027          |
| 1.3913        | 1.8223 | 15500 | 1.5948          |
| 1.4214        | 1.8811 | 16000 | 1.5872          |
| 1.3626        | 1.9399 | 16500 | 1.5810          |
| 1.4187        | 1.9987 | 17000 | 1.5737          |
| 1.154         | 2.0575 | 17500 | 1.5879          |
| 1.2142        | 2.1162 | 18000 | 1.5826          |
| 1.1634        | 2.1750 | 18500 | 1.5811          |
| 1.1774        | 2.2338 | 19000 | 1.5750          |
| 1.196         | 2.2926 | 19500 | 1.5732          |
| 1.1546        | 2.3514 | 20000 | 1.5697          |
| 1.1804        | 2.4102 | 20500 | 1.5666          |
| 1.1517        | 2.4690 | 21000 | 1.5646          |
| 1.1941        | 2.5277 | 21500 | 1.5633          |
| 1.1836        | 2.5865 | 22000 | 1.5611          |
| 1.1603        | 2.6453 | 22500 | 1.5599          |
| 1.2281        | 2.7041 | 23000 | 1.5588          |
| 1.1626        | 2.7629 | 23500 | 1.5578          |
| 1.077         | 2.8217 | 24000 | 1.5579          |
| 1.1677        | 2.8804 | 24500 | 1.5575          |
| 1.1624        | 2.9392 | 25000 | 1.5574          |
| 1.217         | 2.9980 | 25500 | 1.5575          |


### Framework versions

- Transformers 4.40.1
- Pytorch 2.3.0+cu118
- Datasets 2.18.0
- Tokenizers 0.19.1