p04798526
/
OpenDeltaMirror
mirror of https://github.com/thunlp/OpenDelta.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Update README.md

This commit is contained in:
DingDing 2022-02-14 23:28:40 +08:00 committed by GitHub
parent 0201e3703b
commit 8c279ccf47
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 4 additions and 120 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

124
examples/examples_image-classification/README.md
View File

@ -28,9 +28,7 @@ pip install transformers==4.16.2
pip install datsets==1.18.0 pip install datsets==1.18.0
``` ```
### 2. upgrade the transformers to 4.10.0 ### 2. run
### 3. run
```bash ```bash
python run_image_classification.py configs/lora_beans.json python run_image_classification.py configs/lora_beans.json
``` ```
@ -46,121 +44,7 @@ Solution 1: open a python console, running the error command again, may not be u
Solution 2: download the dataset by yourself on a internect connected machine, saved to disk and transfer to your server, at last load_from_disk. Solution 2: download the dataset by yourself on a internect connected machine, saved to disk and transfer to your server, at last load_from_disk.
## Link to original training scripts
You may find solution to other question about the scripts and irrelevant to Opendelta in
https://github.com/huggingface/transformers/tree/master/examples/pytorch/image-classification
# Image classification examples
The following examples showcase how to fine-tune a `ViT` for image-classification using PyTorch.
## Using datasets from 🤗 `datasets`
Here we show how to fine-tune a `ViT` on the [beans](https://huggingface.co/datasets/beans) dataset.
👀 See the results here: [nateraw/vit-base-beans](https://huggingface.co/nateraw/vit-base-beans).
```bash
python run_image_classification.py \
--dataset_name beans \
--output_dir ./beans_outputs/ \
--remove_unused_columns False \
--do_train \
--do_eval \
--push_to_hub \
--push_to_hub_model_id vit-base-beans \
--learning_rate 2e-5 \
--num_train_epochs 5 \
--per_device_train_batch_size 8 \
--per_device_eval_batch_size 8 \
--logging_strategy steps \
--logging_steps 10 \
--evaluation_strategy epoch \
--save_strategy epoch \
--load_best_model_at_end True \
--save_total_limit 3 \
--seed 1337
```
Here we show how to fine-tune a `ViT` on the [cats_vs_dogs](https://huggingface.co/datasets/cats_vs_dogs) dataset.
👀 See the results here: [nateraw/vit-base-cats-vs-dogs](https://huggingface.co/nateraw/vit-base-cats-vs-dogs).
```bash
python run_image_classification.py \
--dataset_name cats_vs_dogs \
--output_dir ./cats_vs_dogs_outputs/ \
--remove_unused_columns False \
--do_train \
--do_eval \
--push_to_hub \
--push_to_hub_model_id vit-base-cats-vs-dogs \
--fp16 True \
--learning_rate 2e-4 \
--num_train_epochs 5 \
--per_device_train_batch_size 32 \
--per_device_eval_batch_size 32 \
--logging_strategy steps \
--logging_steps 10 \
--evaluation_strategy epoch \
--save_strategy epoch \
--load_best_model_at_end True \
--save_total_limit 3 \
--seed 1337
```
## Using your own data
To use your own dataset, the training script expects the following directory structure:
```bash
root/dog/xxx.png
root/dog/xxy.png
root/dog/[...]/xxz.png
root/cat/123.png
root/cat/nsdf3.png
root/cat/[...]/asd932_.png
```
Once you've prepared your dataset, you can can run the script like this:
```bash
python run_image_classification.py \
--dataset_name nateraw/image-folder \
--train_dir <path-to-train-root> \
--output_dir ./outputs/ \
--remove_unused_columns False \
--do_train \
--do_eval
```
### 💡 The above will split the train dir into training and evaluation sets
- To control the split amount, use the `--train_val_split` flag.
- To provide your own validation split in its own directory, you can pass the `--validation_dir <path-to-val-root>` flag.
## Sharing your model on 🤗 Hub
0. If you haven't already, [sign up](https://huggingface.co/join) for a 🤗 account
1. Make sure you have `git-lfs` installed and git set up.
```bash
$ apt install git-lfs
$ git config --global user.email "you@example.com"
$ git config --global user.name "Your Name"
```
2. Log in with your HuggingFace account credentials using `huggingface-cli`
```bash
$ huggingface-cli login
# ...follow the prompts
```
3. When running the script, pass the following arguments:
```bash
python run_image_classification.py \
--push_to_hub \
--push_to_hub_model_id <name-your-model> \
...
```